Hot Cracking Phenomena

Thomas Bِllinghaus
Horst Herold
Hot Cracking Phenomena in Welds
Thomas Bِllinghaus
Horst Herold (Eds.)
Hot Cracking Phenomena
in Welds
With 322 figures and 46 tables
Library of Congress Control Number: 2005921916
ISBN 3-540-22332-0 Springer Berlin Heidelberg New York
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in other ways, and storage in data banks. Duplication of this publication or
parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965,
in its current version, and permission for use must always be obtained from Springer-Verlag. Violations are
liable to prosecution under German Copyright Law.
Springer is a part of Springer Science+Business Media
springeronline.com
© Springer-Verlag Berlin Heidelberg 2005
Printed in Germany
The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws
and regulations and therefore free for general use.
Typesetting: Data conversion by the author.
Final processing by PTP-Berlin Protago-TEX-Production GmbH, Germany
Cover-Design: deblik, Berlin
Printed on acid-free paper 62/3141/Yu – 5 4 3 2 1 0
Dr.-Ing. Thomas Bِllinghaus
Vizeprنsident und Prof. der BAM
Bundesanstalt für Materialforschung und -prüfung
Unter den Eichen 87
12205 Berlin
Germany
Thomas.Boellinghaus@bam.de
Prof. Dr.-Ing. habil. Dr. E.h. Horst Herold
Institut für Füge- und Strahltechnik
Fakultنt für Maschinenbau
Otto-von-Guericke-Universitنt
Universitنtsplatz 2
39106 Magdeburg
Germany
Horst.Herold@mb.uni-magdeburg.de
Preface
During modern fabrication welding of welded components the avoidance
of hot cracking still represents a major topic, sometimes also under new
aspects. Austenitic stainless steels, for instance, widely used in industry
and known to be crack-free joinable by arc welding might turn their
primary solidification mode from ferrite to austenite and thus, might
become increasingly susceptible to hot cracking during increasingly
applied modern laser and hybrid processes.
Additionally, the phenomena of hot cracking in welds have not
completely been understood up to the present. Hydrogen added to the
shielding gas in arc welding processes, for instance, might enhance
solidification cracking by an increasing the heat input, but has been also
tentatively considered to contribute to ductility dip cracking by
embrittlement. Quite numerous technological hot cracking test procedures
have been developed all over the world to rank the hot cracking resistance
of base and filler materials ahead of fabrication welding. Standardization
of such tests appears as very challenging, because the different results are
difficult to compare and to transfer to real component welds.
In order to provide a forum to define the present state of knowledge, to
exchange recent research results, to discuss different viewpoints and to
contribute to the ongoing standardization work on hot cracking phenomena
in welds an international workshop has been organized in March 2004.
The present book contains the 20 individual contributions form experts
all over the world covering four major subjects.
By seven contributions the first chapter provides a complete overview of
the different hot cracking phenomena. Different mechanisms of
solidification cracking proposed in the past decades are summarized and
new insight is particularly given into the mechanism of ductility dip
cracking.
In the second chapter, metallurgy and materials, the effect of different
alloying elements on the hot cracking resistance of various materials are
shown. The initiation of stress corrosion cracking at hot cracks has
additionally been included in one contribution as a special metallurgical
effect.
Since numerical analyses and other simulation techniques represent very
helpful tools to explain cracking phenomena, three individual contributions
show in the third chapter how modelling of hot cracking can be performed
and how such results might support the explanation of mechanisms.
VI Preface
In the final chapter, the various hot cracking tests are presented in seven
individual contributions with a special emphasis on the ongoing process of
standardization. As a final contribution the necessary linking between
testing and practise is outlined on the basis of actual extraordinary cases.
In total, the extensive contributions from eight different countries do not
only provide the latest insight and define the international state of
knowledge on hot cracking phenomena in welds. As a particular item, the
authors included numerous future research perspectives, fairly enough to
excite also the next generation of scientists. By touching all three types of
hot cracking, namely solidification cracking, liquation cracking and
ductility dip cracking and also by explanations of their differences various
articles represent also a very helpful tool for metallurgical and mechanical
engineering students of the higher semesters. Furthermore, the text
contains helpful individual advices, particularly for international welding
engineers confronted with hot cracking in practise.
The editors convey their sincere gratitude to the authors and to all the
participants of the workshop for their individual contributions and their
eager discussions and, in particular, for pushing the scientific knowledge
about hot cracking phenomena in welds a huge step forward.
We especially thank Karen Stelling for her tremendous work in
formatting the individual articles and to prepare this book for printing,
Margit Bauer for her very helpful translation assistance and, in particular,
Thomas Kannengiesser as well as BAM Division V.5 Joining Technology
for the organization and their support during the workshop.
Berlin and Magdeburg, January 2005
Thomas Böllinghaus
Horst Herold
Contents
I Phenomena and Mechanisms
On the Origin of Weld Solidification Cracking............................................3
C.E. Cross
New Insight into the Mechanism of Ductility-Dip Cracking
in Ni-base Weld Metals....................................................................................19
A.J. Ramirez, J.C. Lippold
Influence of Welding Speed on the Hot Cracking Resistance
of the Nickel-Base Alloy NiCr25FeAlY during TIG-Welding...................42
A. Slyvinsky, H. Herold, M. Streitenberger
The Role of Segregation of Oxygen in Welding Alloys
of the INVAR Type ............................................................................................59
K.A. Yushchenko, V.S. Savchenko, T.M. Starushchenko
Character of Hot Crack Formation during Welding
of Cast Heat-Resistant Nickel Alloys...........................................................71
K.A. Yushchenko, V.S. Savchenko, N.O. Chervyakov, A.V. Zvyagintseva
Contribution to HAZ Liquation Cracking of
Austenitic Stainless Steels ............................................................................84
P. Bernasovský
Morphology of Hot Cracks in Single-Phase Weld Metal......................104
B. Yakhushin
II Metallurgy and Materials
The Effect of Silicon and Iron on the Weldability
of Ni-Co-Cr-Si HR-160® Alloy.......................................................................119
I.S. Maroef, M.D. Rowe, G.R. Edwards
The Influence of Different Nb-Contents on the
Hot Cracking Susceptibility of Ni-Base Weld Metals Type 70/20 .......141
R. Vallant
VIII Contents
Hot Cracks as Stress Corrosion Cracking Initiation Sites in
Laser Welded Corrosion Resistant Alloys............................................... 165
K. Stelling, Th. Böllinghaus, M. Wolf, A. Schöler, A. and A. Burkert,
B. Isecke
III Modeling and Simulation
Simulating and Predicting Weld Solidification Cracks......................... 185
Y. Wei, Z. Dong, R. Liu, Z. Dong, Y. Pan
Integrated Mechanical-Metallurgical Approach to Modeling
of Solidification Cracking in Welds ........................................................... 223
V. Ploshikhin, A. Prikhodovsky, M. Makhutin, A. Ilin, H.-W. Zoch
Influence of the Weld Pool Geometry on
Solidification Crack Formation ................................................................... 245
M. Wolf, H. Schobbert, Th. Böllinghaus
IV Testing and Standardization
Recent Developments in Weldability Testing ......................................... 271
J.C. Lippold
Hot Cracking Tests – The Route to International Standardization.... 291
J.C.M. Farrar
Value of Different Hot Cracking Tests for the Manufacturer
of Filler Metals................................................................................................. 305
H. Heuser
Influence of the Deformation Rate of Different Tests
on Hot Cracking Formation ......................................................................... 328
H. Herold, A. Pchennikov, M. Streitenberger
Testing for Susceptibility to Hot Cracking
on Gleeble Physical Simulator.................................................................... 347
S.T. Mandziej
Contents IX
Scientific Bases of the International Standardization Project
"Hot Cracking Tests for Welds"..................................................................377
B. Yakhushin, D. Semin
Discussion and Evaluation of Some Extraordinary Cases
of Hot Cracking ...............................................................................................383
K. Wilken
I Phenomena and Mechanisms
On the Origin of Weld Solidification Cracking
C.E. Cross
Department of Metallurgical & Materials Engineering,
Montana Tech of the University of Montana, USA
Abstract
A review is made of theories regarding weld solidification cracking, identifying
the numerous factors controlling crack initiation. A new approach to
understanding this phenomenon is discussed, considering the conditions
necessary to achieve rupture of liquid films.
Introduction
Weld solidification cracking consists of the rupture of liquid films present
at grain boundaries in the mushy zone trailing the weld pool. The initiation
of such cracks involves a complex interaction between metallurgical and
mechanical factors, driven by the temperature gradients generated during
welding, as suggested in Fig. 1. Thermal-metallurgical interactions control
the solidification microstructure. Thermal-mechanical interactions control
the local and global stresses and strains.
Numerous theories have been proposed to describe the conditions required
for crack initiation, normally oriented toward either metallurgical or
mechanical aspects of the problem. However, seldom is the actual failure
mechanism confronted (i.e. the rupture of a liquid film). In this paper, a review
of solidification theories is made, categorizing each theory in terms
of the factors controlling crack initiation. A new approach is then introduced
which addresses a liquid rupture mechanism, which enables many of
the controlling factors to be combined into a unified theory.
4 Phenomena and Mechanisms
Fig. 1. Diagram indicating complex interaction between process parameters affecting
weld solidification cracking
Identifying Controlling Factors
The path to understanding solidification cracking requires an appreciation
of how various factors interact to affect cracking susceptibility. There is
usually no simple interaction between factors, but instead many complex
interrelationships, making direct correlations difficult between observation
and theory. These factors are identified below, grouped for comparison as
either metallurgical or mechanical in nature. Metallurgical factors are particular
to phase relationships, whereas mechanical factors involve stress
and strain behavior.
Metallurgical
Solidification Range. It has long been appreciated that solidification temperature
range, often referred to as a brittle temperature range (BTR), plays
an important role in solidification cracking susceptibility. Its value can be
defined as the difference between liquidus and solidus temperatures. For
non-equilibrium solidification, the solidus temperature is usually defined
by the lowest melting eutectic in the system. A classical example of this is
the deleterious effect of sulfur and phosphorous impurities on the cracking
susceptibility of ferrous and nickel based alloys. Sulfur can extend the
On the Origin of Weld Solidification Cracking 5
solidus temperature in steel from around 1400°C to 988°C (Fe-S eutectic
temperature).
In aluminum alloys, combinations of copper and magnesium are known
to result in unweldable alloys [1], attributed to the formation of a low melting
Al-Cu-Mg eutectic. However, Al-Mg alloys with an equally large solidification
range, exhibit exceptionally good weldability. Thus, there is
clearly more to this problem than just solidification range.
The solidification range, divided by the temperature gradient (G), gives
the extent of the two-phase mushy zone (i.e. dendrite length). In several
models [2, 3] it is argued that an extended mushy zone will experience
more shrinkage strain and will thus be more susceptible to cracking. It follows
that, when welding a susceptible alloy with high heat input (i.e. low
G), there may be a higher likelihood for cracking to occur. However, other
effects of high heat input (e.g. thermal-mechanical response) may negate
this effect.
In certain weldability tests where an augmented strain is applied (e.g.
varestraint test), the maximum crack length generated at large values of
imposed strain corresponds to a characteristic temperature range equal to
or less than the solidification range [4]. Referred to by Lippold et al. [5] as
a solidification cracking temperature range (SCTR), this parameter has
been effectively used to rank the relative weldability of various different
alloys (also see Recent Developments in Weldability Testing in this text).
Back-filling. Back-filling refers to the drawing of liquid back through the
dendritic network to feed solidification shrinkage. It can also be driven by
capillary flow to fill or “heal” cracks, sometimes observed in metallography
as pools of high solute material (e.g. eutectic) deposited in crackshaped
defects. This interdendritic flow of liquid has been modeled extensively
by numerous researchers [6], with the rate of flow controlled by
dendrite tortuosity, liquid fraction, fluidity, and surface tension. The concept
of back-filling comes from foundry practice, where risers are strategically
placed to feed hot spots to avoid hot tearing. In the case of welding,
the molten weld pool serves as the reservoir of liquid, taking the place of a
riser in a casting.
Dendrite Coherency. Coherency refers to the degree of solid-solid bonding
between secondary dendrite arms occurring at the latter stage of solidification
within the mushy zone (Fig. 2). Theory of Pumphrey and Jennings
[2] proposed that cracking is associated with the thermal contraction experienced
in the coherent region (also defined as a brittle temperature
range), where an alloy with a large coherent region is expected to have a
6 Phenomena and Mechanisms
higher susceptibility to cracking. The problem with this reasoning, as will
be pointed out later, is that alloys with extensive coherent bonding (e.g.
Al-Mg alloys) may actually be more resistant to strain and resultant cracking.
Fig. 2. Schematic drawing showing progressive stages of dendritic solidification
demonstrating regions of liquid back-filling a–b, thin liquid film c, and dendrite
coherency d–f [7]
The maximum coherent region under equilibrium conditions is seen to
occur at the point of maximum solid solution (point B in Fig. 3). This
phase equilibria has been used [2] to explain the peak cracking susceptibility
often observed for aluminum alloys, noting that point B will be
shifted to lower solute values for non-equilibrium solidification.
Fig. 3. Schematic diagram suggesting connection between phase equilibrium and
peak susceptibility in solidification cracking per theory of Pumphrey et al. [2].
Shaded region represents coherent zone
On the Origin of Weld Solidification Cracking 7
Eutectic Fraction. The weight fraction of interdendritic eutectic generated
(fE) increases with solute content (Co), as can be approximated using the
Scheil Equation for non-equilibrium solidification [8]:
[ / ]1/(1 k )
fE = Co CE − , (1)
where CE is the eutectic composition and k is the partition ratio. It has
been suggested that alloys with large solute content, and hence large quantities
of eutectic, will be less susceptible to cracking [1]. This is because:
1) there is a less extensive coherent dendrite structure and 2) shrinkage can
be more readily fed by means of back-filling due to a more open dendrite
array.
The above reasoning can in some cases be used to explain the peak susceptibility
behavior typically observed in aluminum alloys, where high
alloy content in base metal or filler metal often results in improved weldability
[9]. Filler metal alloys are typically high in alloy content for this reason.
Al-Si filler alloys in particular, which generate large quantities of
eutectic, are known for their exceptional weldability. At the opposite extreme,
however, are low eutectic Al-Mg alloys that also experience good
weldability. Al-Mg alloys are expected to have a high degree of coherency
(i.e. large solidification range plus small amount of eutectic) and, thus, can
resist thermal strain.
Surface Tension. Borland defined the effect of surface tension on cracking
in his “Generalized Theory” of cracking [10], combining some aspects of
the shrinkage-brittleness theory of Pumphrey and Jennings [2] with the
strain theory of Pellini [3] (see Strain below). Central to Borland’s theory
is the continuity of liquid at the base of the dendrites, during the last stage
of solidification.
If the last liquid to solidify wets the dendrites (i.e. low γL/S), there will
be a higher likelihood that a continuous network of liquid can provide
back-filling. At the opposite extreme where no wetting occurs (i.e. high
γL/S), bridging between dendrite arms is promoted, resisting strain, and thus
avoiding cracking. It is at intermediate values of wettability, between these
two extremes, where cracking is encountered. Experimental evidence exists
to support this theory, comparing Al-Sn, Al-Cd, and Al-In alloys [10].
Two alloys with poor wetting characteristics (Al-Cd and Al-In: 90° dihedral
angle) were found to be more weldable than the alloy experiencing
moderate wetting (Al-Sn: 65° dihedral angle).
Surface tension must also play a role in the rupture of a liquid film (i.e.
creation of new vapor/liquid interface). The force (F) required to separate a
8 Phenomena and Mechanisms
liquid film wetting two parallel plates has been predicted by Seveiko [11]
to be:
t
F c γA
= 1 ,
(2)
where A represents surface area, γ is vapor/liquid surface tension, and t
is film thickness. In essence, thin films with high surface tension should
prove more resistant to cracking. However, cavitation should provide a
more energetically favorable mechanism for liquid rupture (see Liquid
Rupture below), and so Eq. 2 represents an upper bound on liquid strength.
Another possible influence of surface tension involves temperature gradient
driven fluid flow, commonly known as the Marongoni effect. Considered
from a theoretical standpoint, it has been proposed that variations
in sulfur content may influence interdendritic fluid flow and, hence, backfilling
[12]. For example, it was shown that at high sulfur concentrations
(i.e. dγ/dT >0) the temperature gradient will drive flow out of the dendritic
network, inhibiting feeding of shrinkage.
Grain Boundaries. Solidification cracking normally occurs along weld
metal grain boundaries, although this is not necessarily always the case.
The reason for this tendency is likely tied to the preferential segregation of
solute or impurity elements (e.g. sulfur in steel) to grain boundaries, even
though originally partitioned between dendrites. A grain boundary provides
a well-defined, high energy planar interface upon which a liquid film
can wet. Therefore, it follows that grain shape, structure and size should
have a profound effect on cracking susceptibility.
i. Grain Shape. General wisdom suggests that conditions resulting in columnar
grains growing normal to the welding direction are most deleterious
[13]. Such is the case when welding at rapid travel speeds with the
resultant tear-drop shaped weld pool. It may be that this tear-drop shape
causes solute or impurities to concentrate along the weld centerline, where
one single grain boundary film must accommodate all of the imposed
strain.
ii. Grain Structure. Brooks has shown that different grain boundary
structures in stainless steel weld metal may account for the difference in
weldability between austenite versus ferrite primary solidification [14].
When solidifying as primary austenite, the weld metal grain boundaries are
observed to be better defined (i.e. more straight and continuous) allowing
On the Origin of Weld Solidification Cracking 9
for easy crack propagation, as depicted in Fig. 4. The structure of the grain
boundary is affected by the morphology of the dendrites, which will vary
between alloys.
Fig. 4. Schematic comparing solidification grain boundary structure between primary
austenite (a) and primary ferrite (b) stainless steel, from Brooks et al. [14]
iii. Grain Size. Numerous studies have observed dramatic improvements
in weldability through means of grain refinement, principally in aluminum
alloys. This can be readily explained by considering the distribution of
strain (e.g. from solidification shrinkage and thermal contraction) between
grain boundaries spanning the mushy zone. Smaller grains mean more
grain boundaries are present, which means a smaller amount of strain is
partitioned to each individual grain boundary. A high level of strain, per
grain boundary, is believed to result in cracking (see Strain below). An example
of how grain refinement can effectively reduce solidification cracking
is shown in Fig. 5, where circular patch test results are given for aluminum
alloy 7108 treated with varying amounts of scandium grain refiner
[15].
Porosity. Porosity represents a form of liquid rupture in that it involves the
formation of a liquid/vapor interface, albeit round rather than planar.
Therefore, it is not unreasonable to expect that porosity might play some
role in solidification crack formation. Dixon [16] has observed interdendritic
porosity in close proximity with weld solidification cracks in steel,
using real-time radiography. The flattening and elongation of such a pore
may serve as a crack nucleus, as may the coalescence of micro-pores. A
cracking model involving gas pore coalescence has been proposed [17].
On the other hand, porosity is also known to counter solidification
10 Phenomena and Mechanisms
shrinkage (e.g. killed steel versus rimmed steel), reducing or eliminating
the need for back-filling. Dissolved gasses that result in porosity may also
influence cavitation (see Liquid Rupture below).
Fig. 5. Weld metal grain size versus weldability data from Circular Patch Test for
aluminium alloy 7108 [15]. Grain refinement was achieved using Scandium additions
Mechanical
Strain. Originally viewed as a natural extension of concepts governing
fracture in solids, strain has likewise been assumed to play an important
role in controlling solidification cracking. In his “Strain Theory”, Pellini
[3] proposed that cracking occurs when an intergranular liquid film is
strained beyond some critical value. Furthermore, he points out that the
amount of strain that a liquid film will experience is determined by the
film life, as determined by the solidification range and the weld cooling
rate. It follows that when sulfur is added to steel, the solidification range is
extended, liquid films will be exposed to more strain, and there will be a
higher likelihood that a critical strain will be reached.
Building upon the strain limitation concepts of Pellini, both Prokhorov
[18] and later Senda, et al. [19] established ductility curves, defining the
maximum strain tolerated before cracking occurs. An example of this approach
is shown schematically in Fig. 6, where the solidification range defines
the upper and lower temperature bounds, and the critical ductility
curve is determined experimentally (e.g. by applying controlled strain in
weldabilitiy tests). Cracking will occur if the deformation curve, representing
strain across the mushy zone, intersects the ductility curve.
On the Origin of Weld Solidification Cracking 11
Fig. 6. Schematic showing ductility curve and Brittle Temperature Range (BTR)
from theory of Senda et al. [19]
Stress. Because stress and strain are linked through continuum mechanics,
stress must play a role in any of the cracking mechanisms discussed above
involving strain. Chihoski [20] provided an early analysis of the local
compression and tension stress cells that follow a moving weld pool,
demonstrating how the relative size and location of these cells will vary
with welding parameters.
More recently, finite element models have been used to evaluate local
stresses in weldability studies [21, 22, 23]. Zacharia [21] has shown that by
applying a high cross-weld stress in a Sigmajig test, the trailing end of the
mushy zone will experience a tensile stress resulting in cracking, as indicated
in Fig. 7.
Eq. 2 suggests that liquid films possess a critical strength dependent
upon surface tension and thickness. Also, dendrite coherency should provide
additional resistance to stress. Thus, it seems plausible that a weld
mushy zone must possess some inherent strength. Various experimental
tests have been developed in an attempt to measure the strength of liquidsolid
mushy zones, including the Gleeble™ test [24].
A quartz jacket, placed around a cylindrical test specimen, can be used
to support a molten zone while applying a uni-axial tensile stress. Results
from these tests have typically been inconclusive, reflecting the complex
nature of deformation in the mushy zone.
12 Phenomena and Mechanisms
Fig. 7. Schematic showing two different welds (a) and (b), where the mushy zone
of weld (b) avoids solidification crack by remaining in compression, from Zacharia
[21]
Strain Rate. The importance of strain rate on cracking has been appreciated
from early analyses. From the work of Prokhorov [18] and Senda
[19] discussed earlier, it is clear that the rate of deformation, normal to the
weld, serves to determine whether the critical strain for cracking is
achieved. The slope of the deformation curve in Fig. 6 can be related to
strain rate (dε/dt) and cooling rate (dT/dt) as follows:
dε/dT = (dε/dt) / (dT/dt). (3)
Several weldability tests have been specifically developed to measure
the critical strain rate required for cracking: e.g. VDR test [25] and PVR
test [26]. Another explanation for the role of strain rate will be provided in
the section below in Models for Crack Initiation.
Restraint. It has generally been assumed that high levels of weld restraint
result in higher susceptibility to solidification cracking. This belief may
On the Origin of Weld Solidification Cracking 13
have evolved as an extension of the well-established relationship between
restraint, residual stress, and cold cracking. Numerous different selfrestrained
weldability tests have evolved based upon this belief; e.g.
Houldcroft [27] and Lehigh [28] “fishbone” tests. In one particular
weldability test specifically developed for aluminum (a “window” restraint
test), a slotted weld coupon is welded around its perimeter to a massive 5
cm thick plate (1.2m x 1.4m) to provide excessive restraint [29].
However, recent work by Kannengiesser, et al. [30] has demonstrated
that high restraint does not always lead to higher cracking susceptibility.
What appears to be more important here is the interaction between restraining
forces and local weld strains, and specifically how this affects
strain and strain rate in the vicinity of the mushy zone.
Mechanism for Crack Initiation
Based upon an accumulation of observations made over a 50-year period,
as briefly outlined in the discussions above, a unified model for solidification
cracking is evolving which will eventually address all of the complex
aspects of this problem. Absent in all of the mechanisms discussed above,
however, is a theoretical consideration of how liquid is ruptured. This
would appear to be a rather serious omission, when considering that solidification
cracking involves liquid rupture. Such a consideration has recently
been applied to solidification cracking in castings [7] and has, in a similar
manner, been applied to welding [31]. It is assumed in this model that solidification
cracking is associated with a pressure drop in the interdendritic
liquid, caused by solidification shrinkage, compounded by thermal contraction
strains in the coherent region. When the liquid pressure reaches a
critical negative value, cavitation will occur and result in liquid rupture.
Liquid Rupture
When liquid is placed in a state of hydrostatic tension (i.e. negative pressure),
it becomes metastable. When a sufficiently negative pressure is met,
many fine pores will nucleate spontaneously (i.e. cavitate) in a manner described
by Fischer [32]. This cavitation mode of failure for liquids differs
significantly from fracture in solids, in that liquid rupture is defined by the
critical pressure required to nucleate one single pore. Compare this, for example,
with ductile fracture in solid, which requires the nucleation, growth
and coalescence of many cavities over an extended period of time. The
critical fracture pressure required for homogeneous nucleation of one pore
was shown by Fisher to be:
14 Phenomena and Mechanisms
3 1/ 2
3 ln( / )
16
􀂻 􀂼
􀂺
􀂫 􀂬
􀂪
= −
kT NkT h
pc
π γ ,
(4)
where γ is surface tension, k is Boltzmann’s constant, T is absolute temperature,
N is Avogadro’s number, and h is Planck’s constant.
Campbell [33] took a closer look at Fisher’s liquid fracture pressure, examining
both homogeneous and heterogeneous nucleation. Values of pc
required for homogeneous nucleation are compared in Table I for four different
liquid metals and water.
Table 1. Fracture Pressures [33]
Liquid Surface
Tension
(erg/cm)
Temperature
(°K)
pC
(atm)
Experimental
(atm)
Water 72 300 -1,380 -270
Mercury 490 300 -23,100 -425
Aluminium 850 933 -30,500 –
Copper 1,300 1,356 -48,000 –
Iron 1,850 1,800 -70,800 –
Experimental measurements for fracture pressures are typically orders
of magnitude less than predicted values, indicating that additional factors
must be at play here. One such factor involves dissolved gasses in the melt,
making it easier to initiate fracture by lowering the required pressure by an
amount equal to the internal gas pressure pg (i.e. pc-pg). Another factor
lowering the critical fracture pressure involves heterogeneous nucleation
on substrates (e.g. inclusion particles or oxide films). Gas bubbles entrapped
in grooves on inclusions or oxide films would provide an even
more effective means of initiating failure.
Models for Crack Initiation
Solidification cracking in castings has been related to the pressure drop in
inderdendritic liquid originating from 1) solidification shrinkage, 2) an inability
to properly feed this shrinkage through back-filling, and 3) thermal
strain resulting from cooling. In an early model by Feurer [34], this problem
was partially addressed by assuming that cracking can only occur if
the rate of shrinkage becomes greater than the rate of back-filling. Rappaz,
et al. [7] built upon this model, adding in the strain associated with thermal
contraction, and incorporating Fisher’s criterion for liquid rupture. Thermal
strain is limited to the coherent dendrite region, and crack initiation
On the Origin of Weld Solidification Cracking 15
was assumed to occur in the thin film region immediately ahead of the coherent
dendrite network (Fig. 8).
Fig. 8. Schematic showing location of solidification crack initiation relative to interdendritic
pressure drop and dendrite coherency, from Rappaz et al. [7]
What remains to be done in applying the casting model of Rappaz, et al.
to welding, is to account for the local transverse strain unique to welding.
One might consider this to be ‘solid feeding’ of shrinkage. If there is sufficient
inward movement of base metal behind the weld pool, the interdendritic
pressure cannot drop low enough to initiate rupture. On the other
hand, if there is outward movement of base metal, the cracking problem
will be exacerbated. This would account for the importance of having
compressive rather than tensile local stress fields to avoid cracking, as has
been reported [21]. In a recent study relating local weld strain to cracking
[30], transverse displacement was measured for different conditions of restraint
and joint gap. It was determined that cracking was favored by con16
Phenomena and Mechanisms
ditions producing minimal inward displacement, such as a zero gap joint
(Fig. 9).
-4.00
-3.00
-2.00
-1.00
0.00
1.00
2.00
0 5 10 15 20 25 30 35 40 45 50
Time (seconds)
Displacement (mm)
2.1 mm/s* (No Gap)
2.1 mm/s* (0.1778 mm
Gap)
2.1 mm/s (0.3048 mm
Gap)
Fig. 9. Local weld displacement transverse to weld (feeding shrinkage) for three
different gap conditions, from Kannengiesser et a. [30]
Summary
It has been demonstrated that solidification cracking is a many-faceted
problem, with many influencing factors identified in this paper. Most
attempts to model this phenomenon have concentrated on one or two of
these factors (e.g. solidification range, critical strain, strain rate, etc.), but
rarely have details regarding the actual failure mechanism even been considered.
Models addressing liquid rupture in castings hold promise for
shedding new light on weld solidification cracking, when combined with
existing knowledge and an understanding of localized strain around a moving
weld pool.
On the Origin of Weld Solidification Cracking 17
References
1. Cross CE, Kramer LS, Tack WT, Loechel LW (1990) Aluminum weldability
and hot tearing theory. Welding of Materials. ASM Int: 275–282
2. Pumphrey WI, Jennings PH (1948) A consideration of the nature of brittleness
and temperature above the solidus in castings and welds in aluminum alloys. J
Inst Metals 75: 235–256
3. Pellini WS (1952) Strain theory of hot tearing. Foundry 80: 125–199
4. Cross CE, Olson DL (1986) Hot tearing model to asses aluminum weldability.
Aluminum alloys-their physical and mechanical properties, Vol 3: 1869–1875
5. Lippold JC, Lin W (1996) Weldability of Al-Cu-Li alloys. Aluminum alloys –
their physical and mechanical proterties. Transtec Pub: 1685–1690
6. Poirier DR, Yeum K, Maples AL (1987) A thermodynamic prediction for microporosity
formation in aluminum-rich Al-Cu alloys. Met Trans 18A: 1979–
1987
7. Rappaz M, Drezet JM, Gremaud M (1999) A new hot-tearing criterion. Met
Mat Trans 30A: 449–455
8. Kurz W, Fisher DJ (1986) Fundamentals of solidification. Trans Tech Pub:
232
9. Mousavi MG, Cross CE, Grong Ø, Hval M (1997) Controlling weld metal dilution
for optimized weld performance in aluminum. Sci Tech Weld Join 2:
275–278
10. Borland JC (1960) Generalized theory of super-solidus cracking in welds and
castings – an initial development. Brit Weld J 7: 508–512
11. Saveiko VN (1961) Theory of hot tearing. Russ Cast Prod: 453–456
12. Holt M, Olson DL, Cross CE (1992) Interfacial tension driven fluid flow
model for hot cracking. Scripta Met Mat 26: 1119–1124
13. Savage WF, Aronson AH (1966) Preferred orientation in the weld fusion
zone. Welding J 45: 85s–89s
14. Brooks JA, Thompson AW (1991) Microstructural development and solidification
cracking susceptibility of austenitic stainless steel welds. Int Met Rev:
16–43
15. Mousavi MG, Cross CE, Grong Ø (1999) Effect of scandium and titaniumboron
on grain refinement and hot cracking of aluminum alloy 7108. Sci Tech
Weld Join 4: 381–388
16. Dixon, BF (1988) A fractographic study of weld metal solidification cracking
in steels. PhD Thesis, Univ Melbourne
17. Cross CE, Olson DL, Edwards GR (1994) The role of porosity in initiating
weld metal hot cracks. In: Modeling and control of joining processes. AWS,
Miami, pp 549–557
18. Prokhorov NN (1956) The problem of the strength of metals while solidifying
during welding. Svar Proiz 6: 5–11
19. Senda T, Matsuda F. Takano G (1973) Studies on solidification crack susceptibility
for weld metals with trans-varestraint test. J Japan Weld Soc 42: 48–56
18 Phenomena and Mechanisms
20. Chihoski RA (1972) The character of stress fields around a weld arc moving
on an aluminum sheet. Weld J 51: 9s–18s
21. Zacharia T (1994) Dynamic stresses in weld metal hot cracking. Weld J 73:
164s–172s
22. Feng Z, Zacharia T, David, SA (1997) On the thermomechanical conditions
for weld metal solidification cracking. In: Mathematical modelling of weld
phenomena 3. Inst Materials, London, pp 114–148
23. Dike JJ, Brooks JA, Li M (1998) Comparison of failure criteria in weld solidification
cracking simulations. In: Mathematical modelling of weld phenomena
4. Inst Materials, London, pp 199–222
24. Batra R, Wilber GA, Breit HF Childs WJ ( 1975) Programmed In Situ Melting,
Freezing , and Tensile Testing for Laboratory Study of High Temperature
Properties of As-Cast Metals. Journal of Testing and Evaluation, Vol 3 (1):
68–74
25. Matsuda F, Nakagawa H, Nakata K, Okada H (1979) The VDR cracking test
for solidification crack susceptibility on weld metals and its application to
aluminum alloys. Trans JWRI 8: 85–95
26. Herold H, Streitenberger M, Pchennikov A (2001) Modelling of the PVR-test
to examine the origin of different hot cracking types. In: Mathematical modelling
of weld phenomena 5. Inst Metals, London, pp 783–792
27. Houldcroft PT (1955) A simple cracking test for use with argon-arc welding.
Brit Weld J 2: 471–475
28. Stout RD, Tor SS, McGeady LJ, Doan GE (1946) Quantitative measurement
of cracking tendency in welds. Weld J 25: 522s–532s
29. Robinson IB, Baysinger FR (1966) Welding aluminum alloy 7039. Weld J 45:
433s–444s
30. Kannengiesser Th, McInearney T, Forian W, Böllinghaus Th, Cross CE
(2002) The influence of local weld deformation on hot cracking susceptibility.
Mathematical modelling of weld phenomena 6. Manley, London, pp 803–818
31. Rindler W, Kozeschnik E, Enzinger N, Buchmayr B (2002) A modified hot
tearing criterion for steels. In: Mathematical modelling of weld phenomena 6,
Manley, London, pp 819–835
32. Fisher JC (1948) The fraction of liquids. J Appl Phys 19: 1062–1067
33. Campbell J (1968) Pore nucleation in solidifying metals. In: The solidification
of metals. Iron Steel Inst, London, pp 18–26
34. Feurer, U (1977) Influence of alloy composition and solidification conditions
on dendrite arm spacing, feeding, and hot tear properties of aluminum alloys.
In: Proceedings of the international symposium on engineering alloys, Delft:
131–145
New Insight into the Mechanism of Ductility-Dip
Cracking in Ni-base Weld Metals
A.J. Ramirez1, J.C. Lippold2
1 Synchrotron Light National Laboratory, Campinas-SP, Brazil
2 The Ohio State University, USA
Abstract
The ductility dip cracking resistance of Ni-base Filler Metal 52 (AWS
A5.14, ERNiCrFe-7; ISO SNi6052, 59 % Ni, 29 % Cr, 9 % Fe, 1 % Mn,
1 % Al) and Filler Metal 82 (AWS A5.14, ERNiCr-3; ISO SNi6082,
72 % Ni, 20 % Cr, 1 % Fe, 3 % Mn, 3 % Nb) has been extensively
evaluated using the recently developed strain-to-fracture test in conjunction
with microstructural characterization using electron microscopy.
This paper provides new insight into the creep-like, grain boundary sliding
mechanism that leads to elevated temperature intergranular cracking in
these weld deposits. The effect of precipitation on grain boundary tortuosity
and sliding, and its influence on ductility-dip cracking resistance is discussed
in the context of current theories of high temperature creep. Finally,
the effect of impurity and interstitial elements on ductility-dip cracking is
discussed.
Introduction
Many normally ductile austenitic alloys (FCC structure) such as Ni-base
alloys, Ni-Cu alloys, Cu alloys, stainless steels and steels undergo a severe
ductility reduction at temperatures between 0.5 and 0.8 of their melting
temperature [1, 2, 3]. The exhaustion of ductility during the high temperature
processing or welding of these alloys results in intergranular
cracking often referred to as ductility-dip cracking (DDC).
Previous studies with FM-52 and FM-82 addressed the factors that influence
elevated temperature ductility and the DDC mechanism [4]. These
factors are:
20 Phenomena and Mechanisms
• Weld metal chemical composition including impurity and interstitial
elements;
• Segregation to the grain boundaries;
• Precipitation behavior;
• Grain boundary migration and pinning;
• Grain boundary orientation (macroscopic) relative to the applied strain;
• Dynamic recrystallization.
Despite the extensive research performed on ductility dip cracking, a
complete understanding of this phenomenon has not been achieved. In order
to provide a better understanding of the influence of the above mentioned
factors on the DDC phenomenon it has been necessary to develop
new testing techniques and to perform detailed microstructural characterization.
Several test techniques have been used in the past to quantify DDC susceptibility.
Among these techniques are: elevated temperature tensile test,
hot ductility test, and Varestraint-based tests. However, most of these
techniques have poor repeatability and/or difficulty differentiating between
DDC and other high temperature cracking phenomena. For this reason, the
strain-to-fracture (STF) test was used to reproduce the DDC conditions,
avoiding most of the problems with the previous techniques [5, 6]. Thus,
using samples prepared using the STF test it is possible to more effectively
study the DDC phenomenon in the context of creep-like grain boundary
sliding and analyze the effect of precipitates and impurities on this cracking
process.
Materials and Sample Preparation
Two Ni-base alloys, designated Filler Metal 52 (AWS A5.14, ERNiCrFe-
7; ISO SNi6052) and Filler Metal 82 (AWS A5.14, ERNiCr-3; ISO
SNi6082) were used in this study. Filler Metal 82 and 52 are commonly
used to weld alloys 600/625 and 690, respectively, and to perform dissimilar
welds involving low alloy steels and Ni-base alloys or stainless steels.
The chemical composition of the filler metals is presented in Table 1. Because
of the large variability of DDC resistance from heat to heat, two different
FM-82 heats were included in this study.
Multipass weld deposits were made using the automated gas tungsten
arc welding (GTAW) process. The base material used to prepare this joint
was A36, a low carbon structural steel. However, extreme care was taken
to reduce base metal dilution in the Ni-base weld metal.
New Insight into the Mechanism of Ductility-Dip Cracking 21
Table 1. Chemical composition (wt%) of the filler metals (balance Ni)
Alloy C Mn Fe Cr Nb Ti Al Si Other
FM-52
NX9277
0.026 0.25 8.88 29.1 0.02 0.50 0.71 0.17
0.01 Cu
0.05 Mo
0.004 S
0.004 P
FM-82
YN6830
0.040 2.86 1.18 20.1 2.3 0.37 – 0.12
0.09 Cu
0.05 Co
0.001 S
0.007 P
FM-82
YN7355
0.040 2.75 0.70 20.1 2.6 0.47 – 0.07
0.07 Cu
0.04 Co
0.002 S
0.01 P
Dog bone-like samples, as presented in Fig. 1, with the weld deposit in
the middle were cut out transversely from this joint. Subsequently, a
GTAW autogenous spot weld was made at the sample midspan, within the
pre-deposited weld metal. The circular geometry of this spot weld produces
a radial distribution of grain boundaries such that axial straining during
the STF test allows cracking to occur along the most favorably oriented
grain boundaries. More details about the welding conditions and
sample preparation are presented elsewhere [7, 8].
Strain-to-Fracture Testing
This Gleeble™ based test replicates the thermo-mechanical conditions experimented
by the weld metal resulting in DDC. The sample is heated to
the desired temperature, held for a short period of time (customary 10 s)
and then subjected to a predetermined strain under controlled conditions
[6]. The strain-to-fracture test makes it possible to determine the ductilitydip
temperature range and the strain threshold to cause cracking. The STF
data for the FM-52 and FM-82 presented in Fig. 2 is the reinterpretation of
some recent data obtained by Collins et al.[7, 9]. The new approach takes
into consideration both the cracking threshold and the number of cracks as
a measure of cracking susceptibility.
The high susceptibility of FM-52 to DDC, which is observed during actual
fabrication of heavy sections, when compared with FM-82 was verified
by the STF test results.
22 Phenomena and Mechanisms
Fig. 1. Strain-to-fracture sample
New Insight into the Mechanism of Ductility-Dip Cracking 23
Fig. 2. Strain-to-fracture test results of (a) FM-52 heat NX9277, (b) FM-82 heat
YN7355, (c) FM-82 heat YN6830. Adapted from “An Investigation of Ductility
Dip Cracking in Nickel-Base Filler materials” [7]
24 Phenomena and Mechanisms
FM-52 exhibited a consistently low threshold strain for cracking (between
1 and 2 %) along the entire ductility-dip temperature range (from
700 to 1200 °C), with the minimum at about 1050 °C. In addition, the
number of cracks increased dramatically as the applied strain increased.
On the other hand, FM-82 heats YN-6830 and YN7355 had threshold
strains of approximately 3 % and 4 %, respectively, with the number of
cracks increasing more gradually with the increase in applied strain when
compared with FM-52. The FM-82 heat YN6830 presented a less consistent
ductility-dip trough between 700 and 1200 °C, with two minimums of
ductility at about 775 °C (2.5 % strain) and about 1100 °C (3 % strain).
Between these two minimums the threshold ductility was about 5 %. The
FM-82 heat YN7355 had a more consistent ductility-dip trough between
700 and 1200 °C, with a constant threshold of about 4 % strain throughout
this temperature range.
DDC susceptibility does not depend only on the minimum threshold
strain to start cracking along the whole ductility-dip trough range. In fact,
DDC is a complex phenomenon, which is controlled by the available ductility
along the critical temperature range where the material is subjected to
severe strain and/or strain rate conditions. Finite element modeling has
shown that just behind the weld pool the recently solidified material is under
compressive stresses and tensile stresses start to build just up when this
region reaches lower temperatures [10]. The above explains the higher
susceptibility of FM-82 heat YN7355 to DDC in actual welding conditions
when compared to FM-82 heat YN6830 despite the lower overall threshold
strain to start cracking presented by the second alloy.
Microstructural Characterization
The STF samples were sectioned and prepared for light and scanning electron
microscopy (SEM) by grinding and then polishing, using diamond
pastes of 6, 3 and 1 μm. The final polishing was done with 0.05 μm colloidal
silica. Electrolytic etching was performed with a 10 % aqueous chromic
acid solution at 1.5 to 2.0 V for 30 seconds.
The SEM analyses were conducted using a Phillips XL-30 ESEM FEG
and FEI SIRION FEG microscopes coupled with X-ray energy-dispersive
spectroscopy (XEDS) and electron backscatter diffraction (EBSD). The
EBSD data acquisition and analysis were performed using CHANNEL 5
and PHASE-ID software. Transmission electron microscopy (TEM) foils
were mechanically thinned until approximately 100 μm, followed by jet
New Insight into the Mechanism of Ductility-Dip Cracking 25
Fig. 3. Migrated grain boundaries in the STF samples. In (a) straight migrated
grain boundary in FM-52 heat NX9277 (STF sample deformed 2.9% at 1160 °C)
and (b) grain boundary pinning caused by the frequent medium size eutectic
(NbTi)C precipitates in FM-82 heat YN6830 (STF sample deformed 7.5 % at
972 °C. Secondary electron SEM image
polishing using a solution of 70 %vol ethyl alcohol, 20 %vol Glycerin and
10 %vol of perchloric acid (HClO4) at -20 °C applying 30 V. TEM analyses
were carried out at 200 kV in a Philips CM-200 coupled with XEDS.
26 Phenomena and Mechanisms
The summary of the optical, SEM and TEM observations on the STF
samples is presented on Table 2. However, a detail characterization is presented
elsewhere [11]. The most relevant aspects of FM-52 and FM-82
STF weld metal characterization are the following:
• Alloy FM-52 weld metal has straight migrated GBs and FM-82 has
more tortuous GBs, as shows Fig. 3;
• FM-52 weld metal contains a low concentration of large, medium and
small TiN-like nitrides and numerous small intergranular M23C6. Additionally,
this alloy has some sulfur rich films along the GBs;
• FM-82 weld metal exhibits numerous medium and small, and sporadic
very small intergranular NbC-like carbides.
Table 2. Precipitates observed in the STF samples of FM-52 and FM-82
Precipitates Weld Metal
FM-52 FM-82
Large
Sporadic (Inter., Intra.)
(TiCr)(CN) – 5 μm – Transferred
from the wire. Fig. 4.
None
Medium Sporadic (Inter., Intra.)
(TiCr)(CN) – 1 μm. Fig. 3
Frequent (Inter., Intra.)
(NbTi)C – 1 μm (Nb, Ti,
Cr, Ni, Fe rich). Fig. 3,7.
Small
Sporadic (Inter.) round-like
M23C6 – 100 nm (Cr, Fe rich)
– Associated with small TiNlike
particles.
Sporadic (Inter.) (TiCr)(CN)
– 100 nm.
Numerous (Inter.) square-like
M23C6 – 50 nm (Cr, Fe rich).
Fig. 5.
Frequent (Inter.) (NbTi)C –
50 nm - Isolated, aligned or
grouped. Fig. 8.
Sporadic (Intra.) (NbTi)C –
50 nm – Interdendritic regions.
Fig. 8.
Very small Some (Intra.) presumably
M23C6 – or TiN-like – 10 nm.
Sporadic (Inter.) very likely
(NbTi)C – 10 nm – Associated
with grouped small
(NbTi)C particles. Fig. 8.
Intergranular
Films
Some – Sulfur rich – Associated
with medium size intergranular
(TiCr)(CN) – 1 μm.
Fig. 6.
None
Inter. Intergranular precipitates,
Intra. Intragranular precipitates,
Amount scale: Numerous – Frequent – Some – Sporadic.
New Insight into the Mechanism of Ductility-Dip Cracking 27
Fig. 4. Large titanium nitride (TiN) precipitates in the STF sample of FM-52 heat
NX9277 (strained 1.7 % at 1090 °C). Light microscopy
Fig. 5. M23C6 carbides precipitated along the GB of a STF sample of FM-52 heat
NX9277 strained 1.6 % at 956 °C. In (a) detail of the carbides and (b) SAD pattern
of the austenite grains and the carbides showing the cube-on-cube orientation
relationship between the precipitates and one of the austenite grains. The SAD
pattern presents both phases on the [011] zone axis
28 Phenomena and Mechanisms
Fig. 6. Sulfur enriched film observed along grain boundaries in the FM-52 heat
NX9277 weld metal. In (a) Secondary electron SEM detail of the film and (b)
XEDS spectrum of this film acquired at 5 keV
Extensive GB pinning caused by the medium size eutectic NbC-like
precipitates was observed in FM-82. Some evidence of GB pinning was
seen in FM-52 as well. Nevertheless, the uneven distribution of TiN-like
nitrides in FM-52 did not cause the efficient GB pinning effect that the
NbC-like carbides did in FM-82.
New Insight into the Mechanism of Ductility-Dip Cracking 29
Fig. 7. DDC cracks along migrated grain boundaries and medium size (1 μm) interdendritic
(NbTi)C precipitates in FM-82 heat YN6830 (STF sample strained
7.5 % at 972 °C). In (a) General view and detail of the intergranular cracks and
precipitates, and (b) calculated EBSD pattern for the indicated (NbTi)C precipitate
superimposed on the actual precipitate pattern
30 Phenomena and Mechanisms
Fig. 8. Grouped small (NbTi)C and very small intergranular precipitates on FM-
82 heat YN6830 (STF sample strained 3.5 % at 955 °C). In (a) General view of a
random grain boundary with grouped intergranular small (50 nm) (NbTi)C carbides
and intragranular carbides as indicated by the arrow, and (b) the detail of the
grain boundary region showing the same small carbides and other very small
(10 nm) precipitates, as indicated by the arrows. Bright field TEM images
New Insight into the Mechanism of Ductility-Dip Cracking 31
Fig. 9. FM-82 heat YN6830 strain-to-fracture sample strained 11.3 % at 1150 °C.
In (a) Secondary electron image of the fractured sample. In (b) Strain distribution
map from the same region obtained from EBSD measurements. The thin lines represent
the high angle grain boundaries; the black regions are the open cracks; the
gray scale contouring shows the strain distribution with the dark representing the
lowest and the light the highest strains
The small and/or very small intergranular precipitates observed in both
alloys would effectively pin the GBs. However, these precipitates
nucleated heterogeneously at the GBs after most of the GB migration had
occurred. Thus, the effect of the small and very small precipitates on GB
32 Phenomena and Mechanisms
migration and therefore GB tortuosity was negligible for all the studied
alloys.
In addition to grain boundaries and precipitate characterization, EBSD
analysis of STF samples revealed the microstrain concentration around the
grain boundaries, mainly around triple points and the curved segments of
the GBs.
Fig. 9 shows the microstrain map of the FM-82 YN6830 STF sample
strained 11.3 % at 1150 °C. Strain concentration at the crack tips was revealed
by this analysis as well. However, the strain field ahead the crack
tip is a consequence of the crack propagation phenomenon and not the
cracking cause. The formation of these highly strained regions resulted in
dynamic recrystallization, as has been evidenced by EBSD analysis [12].
Weld Metal Grain Boundaries
Three different types of grain boundaries (GBs) can be recognized within
the weld metal, solidification grain boundaries (SGBs), solidification subgrain
boundaries (SSGBs), and migrated grain boundaries (MGBs). The
combination of solute partitioning and the thermal gradient at the solidification
front cause constitutional supercooling, which results in the instability
of the solidification front itself, leading to the formation of columnar,
cellular, cellular dendritic or dendritic solidification structures within the
grains.
The boundaries that separate these columns or dendrites within the
grains are the SSGBs, which have an important compositional component
as a result of the micro-segregation during the solidification process. The
SSGB are by definition low angle GBs, and possess a very small crystallographic
component. On the other hand, SGBs are high angle GBs were
packets of subgrains (cells and/or dendrites) intersect during the solidification
process. SGBs have both a crystallographic and compositional component.
Once the weld metal has solidified, the crystallographic component
tends to migrate away from its original position to reduce its energy,
leaving the compositional component (segregation) behind. The new GBs,
which have basically a crystallographic component and formed by migration
away from the segregation field, are called MGBs. These different
types of boundaries are shown in Fig. 10.
The MGB carries with it the original misorientation of the SGB and is,
in effect, no different than a base metal grain boundary. In the absence of
grain boundary pinning agents, such as second phases or precipitates, the
MGB will be relatively straight. The distance that it migrates is a function
of the weld cooling rate, with slower cooling rates allowing more time for
New Insight into the Mechanism of Ductility-Dip Cracking 33
migration. Additional migration may also occur during weld metal reheating,
such as during multipass welding.
Fig. 10. Weld metal boundaries in austenitic weld metal. In (a) schematic illustration
[13] and (b) migrated gain boundaries in FM-52 weld metal, which pulled
away from the segregation patterns revealed by the etching
34 Phenomena and Mechanisms
If precipitates or second phases are present, resulting either from solidification
or solid-state precipitation, then the MGBs can be pinned and
more tortuous boundaries will result. The microstructure characterization
presented in the previous section has shown the effect of certain precipitates
on GB tortuosity, with the MGBs in FM-52 weld metals being characteristically
straighter, while the FM-82 boundaries are more tortuous.
Examples of migrated grain boundaries in FM-52 and FM-82 are shown in
Fig. 3 and Fig. 10b.
Ductility Dip Cracking Mechanism in Ni-base Alloys
Before discussing the ductility dip cracking mechanism in Ni-base weld
metals, it is important to describe the principal characteristics of these
cracks.
• They occur along MGBs [2, 4, 7–9];
• The orientation of these MGBs relative to the load applied to the sample
has an important influence on the nature of cracking. In the STF samples,
intergranular cracking normally occurred along the GBs oriented
from 45 to 90° to the applied load [7];
• Evidence of GB sliding has been observed and previously associated
with this type of cracking [1, 4, 14];
• The ductility recovery at the high temperature extreme of the ductilitydip
trough has been associated with the onset of recrystallization [4, 9].
These characteristics suggest that DDC is a GB sliding, creep-like phenomenon,
as was initially proposed by Rhines and Wray [1]. The treatment
of DDC as a creep-like phenomenon, explains the following DDC characteristics.
• The ductility drop between approximately 0.5 and 0.8 of the alloy melting
temperature is limited at the high temperature side by the ductility
recovery due to the onset of recrystallization, which eliminates the deformation
that is accumulating along the GBs and therefore precludes
void formation at either triple points or GB sites. Recrystallization has
been observed by the authors in actual multipass weld metal and in STF
samples tested at temperatures above 1000 °C. In these cases, recrystallization
occurs locally along the GBs and at triple points, where most of
the deformation is concentrated. Recrystallization was also associated
with the regions around the ductility-dip cracks, especially ahead the
cracks. This observation suggests the possible participation of recrystallization
on crack arrest at the higher temperature range. On the other
New Insight into the Mechanism of Ductility-Dip Cracking 35
hand, the low temperature ductility recovery is due to the inoperability
of GB sliding at these lower temperatures [15, 16].
• The GB orientation dependency of DDC is also an important characteristic
of creep, where the intergranular cavities and cracks are normally
formed at about 90° to the applied load. The STF test sample design,
having a spot weld with a radial orientation of migrated grain boundaries
provides a spectrum of GB orientations to the applied load ranging
from 0 to 90 degrees.
• GB sliding in a three dimensional array of grains causes strain concentration
at triple points and at other GB irregularities. This strain concentration
occurs at locations where the GB has sudden directional changes,
such as at “pinning points” associated with intergranular precipitates or
at GB steps. In general, any feature that inhibits GB sliding will be a
preferential site for strain concentration during the deformation process.
Strain concentration at these grain boundary sites can lead to intergranular
cavity formation and subsequently to intergranular DDC cracking.
• The chemical composition of the alloy has an important influence on the
GB sliding phenomenon. Two of the most important factors that chemical
composition controls are 1) the type, size, morphology and distribution
of precipitates within the microstructure and 2) the intergranular
embrittlement caused by impurity and interstitial segregation. Both factors
will be addressed later in this paper.
Influence of Precipitates on the Creep-Like Induced Cracking
Phenomenon
The complex effect of intergranular and intragranular precipitation on the
creep-like DDC phenomenon should be analyzed from two different perspectives.
The first is the effect of precipitation on GB migration and the
subsequent effect that GB migration and pinning has on DDC. The second
is the direct effect that intragranular and mainly intergranular precipitates
have on the GB sliding phenomenon. The effect that precipitates will have
on GB migration and sliding depends on the nature, size, distribution and
evolution of the precipitates in the microstructure during the weld thermal
cycle.
The effect that different sizes and types of precipitates have on the on
GB migration of FM-52 and FM-82 was discussed previously and the precipitate
types summarized in Table 2. In general terms some of these precipitates
pinned the GBs, reducing or preventing GB migration. As GB
migration is inhibited, the average grain size remains smaller, sweeping of
impurities during migration will be reduced, and, perhaps most impor36
Phenomena and Mechanisms
tantly, GB tortuosity increases due to the pining effect (Fig. 3). The
smaller grain size results in lower strain concentration at individual GBs
and reduces the tendency for intergranular void formation and subsequent
cracking [17].
When GB tortuosity is maintained or increased, sliding is restricted due
a mechanical interlocking effect. This causes a reduction in strain concentration
around triple points and therefore, reduces the probability of void
formation at these sites, as revealed by micro-strain measurements using
EBSD [12]. The restriction of GB migration imposed by the precipitates
may maintain and/or promote GB tortuosity. However, the precipitates
have to be present in the microstructure at the right time in order to prevent
GB migration and have an effect on DDC resistance. In addition, precipitates
size and distribution also play an important role in restricting GB migration.
In the case of the FM-52, the large (5 μm) and medium (1 μm) TiN-like
precipitates, which were either transferred from the wire and/or formed in
the liquid during solidification, were present just sporadically in the microstructure.
Thus, the average GB pinning effect of this low fraction of relatively
large particles was negligible, as evidenced by the long and straight
GBs observed in FM-52 weld metal. On the other hand, the numerous medium
size (1 μm) interdendritic (NbTi)C carbides observed in FM-82 effectively
pinned the GBs during the cooling following the solidification
and during the subsequent reheating, resulting in tortuous GBs. The Nbrich
precipitates form as the result of a eutectic reaction at the end of solidification
[18, 19]. Tortuous MGBs make GB sliding more difficult and,
as a result, the DDC resistance of FM-82 is improved relative to FM-52, as
revealed by STF test results and actual welding experience with these filler
metals [7–9, 20].
As previously mentioned, the small precipitates observed in both filler
metals precipitated heterogeneously at the GBs. Therefore, they formed after
most of the GB migration has occurred. Thus, despite the important
pinning effect of these numerous small precipitates, their effect on GB tortuosity
was negligible.
In addition to GB tortuosity, the degree of GB coherency plays an important
role on the GB response to high temperature deformation. Work by
Lim et al. [21] and Lehockey et al. [22] showed that low Σ coincidence site
lattice (CSL) GBs are less prone to cavity formation during high temperature
deformation of nickel. Lim et al. explained this behavior by the faster
accumulation of misfit strain on the higher Σ GBs [21]. However, previous
results have shown that GB coherency is not an important factor on actual
New Insight into the Mechanism of Ductility-Dip Cracking 37
DDC formation during multipass welding of FM-52 and FM-82 due to the
low fraction of low Σ CSL GBs in the weld metal [4, 11].
The effect of intergranular carbides on creep processes is more complex
than just the role they play on the control of GB migration. The intergranular
precipitates cause a locking effect on the GB, making GB sliding more
difficult. The effect of intergranular precipitates depends on the creep
stage, particle size, and particle distribution [17]. Initially the carbides extend
the creep life by the reduction of creep deformation, but at the end of
the creep life, void (cavity) nucleation is associated with the intergranular
carbides. But even during this final stage of creep life, the carbides may
improve the material response reducing the agglomeration of voids, as observed
in some cast stainless steels [23]. Continuous carbides along the
GB, as the numerous small M23C6 precipitates observed on FM-52, are
easy crack propagation paths [24]. However, isolated blocky carbides
along the GB may be beneficial for the high temperature cracking resistance,
due to the reduction in GB sliding. Because DDC is a GB sliding
process, a similar effect of the intergranular precipitates is expected under
DDC conditions, which explains the beneficial effect of the numerous
eutectic, medium-size carbides on the migrated grain boundaries in the
FM-82 weld metal.
Intragranular carbides play a role in the DDC mechanism as well. They
increase high temperature strength and reduce the creep deformation due to
the dislocation pinning effect [25]. Stronger grain interiors make strain relaxation
along the grain boundaries more difficult, favoring triple point
void formation [26]. The amount of intragranular precipitation in the two
filler metals studied here is low and its effect on the DDC phenomenon is
thought to be minimal.
Effect of Impurity and Interstitial Elements on DDC
Impurities and interstitials such as S, P, and H have an important effect on
intermediate temperature ductility and therefore, on DDC susceptibility.
Previous results by the authors have shown that hydrogen increases the
susceptibility to DDC [7–9]. The mechanism of hydrogen enhanced DDC
is thought to be a combination of hydrogen-enhanced local plasticity
around the grain boundaries and hydrogen-induced decohesion of GBs
and/or the interfaces between the intergranular precipitates and the matrix.
Although, hydrogen is generally considered to play a more important role
at lower temperatures, it may also influence ductility within the DDC temperature
range.
38 Phenomena and Mechanisms
Sulfur and phosphorus are elements that segregate to the GBs during solidification,
cooling and reheating and cause GB embrittlement [27, 28].
The increase in DDC susceptibility of Invar (Fe-36Ni) associated with a
sulfur increase from 0.004 to 0.011 wt% has been reported to be related
with the S segregation to the GBs during reheating [29]. A previous study
with FM-82 showed that sulfur additions to the weld metal caused a reduction
in the threshold strain to cause cracking from 6 % to 2 % during the
strain-to-fracture test [9]. Thus, the presence of high sulfur levels at FM-52
migrated grain boundaries indicates that S segregation to the GBs may be
playing an important role on the high susceptibility of this alloy to DDC.
However, the differences in DDC susceptibility among FM-52 and two
FM-82 heats cannot be simply explained based on the bulk S and P content
differences, which are negligible as shown in Table 1.
Summary
The new three-dimensional presentation of the STF results data has permitted
a better interpretation and understanding of the effect of strain on
DDC susceptibility. Further analysis of the data has shown that the variation
in DDC between FM-52 and FM-82 goes beyond the difference in
strain threshold to initiate cracking. Once the cracking threshold is exceeded
in FM-52, the severity of cracking (number of cracks) increases
rapidly with the applied strain, revealing the higher susceptibility of this
alloy to grain boundary embrittlement.
Regarding the differences in DDC susceptibility between the two heats
of FM-82, the STF test results revealed differences in high temperature
ductility. However, the overall threshold strain to initiate cracking is important,
but not enough to characterize the DDC susceptibility of the alloy
under actual production welding conditions. It is proposed that the DDC
resistance is controlled by the available ductility of the weld metal within a
narrow range of the ductility-dip trough where higher strains and/or strain
rates are developed during weld cooling. Based on FEM analysis it is presumed
that this occurs at the lower temperature extreme of the ductility-dip
trough. It is anticipated that a better understanding of the DDC mechanism
will allow STF test results to be a more effective predictor of DDC susceptibility
and may even lead to a modification of the test procedure itself.
Consequently, the better understanding of the DDC mechanism and STF
and other test results will potentially make it possible to anticipate DDC
problems before they occur in actual production and to develop strategies
to avoid DDC.
New Insight into the Mechanism of Ductility-Dip Cracking 39
The combination of STF results with exhaustive microstructural and
mechanical (microstrain) characterization has permitted DDC to be better
described as a grain boundary sliding, creep-like phenomenon. The ductility-
dip trough is limited at low temperatures by the inoperability of GB
sliding and at high temperatures by the onset of dynamic recrystallization.
The ductility-dip crack orientation dependence to the applied load has been
verified and correlated with the GB sliding phenomenon. GB triple points
and any other irregularities along the GB, such as intergranular precipitates,
GB directional changes or GB steps will oppose GB sliding, causing
local strain concentration. These strain concentration sites along the GBs
will be void initiation sites. If the temperature and strain conditions are appropriate,
the growth and linking of these voids causes microcrack formation,
resulting in DDC.
The effective GB pinning caused by the medium size (1 μm) (NbTi)C
carbides in the two heats of FM-82, caused the formation of tortuous GBs
in the weld metals of these heats. This GB tortuosity has a mechanical
locking effect on the GBs, limiting GB sliding. The reduced GB sliding
caused a reduction of strain accumulation at the triple points and a reduction
in the void formation at these sites. The limited GB pinning effect of
the sporadic TiN-like precipitates observed on FM-52 weld metal did not
promote GB tortuosity and facilitated GB sliding and the consequent strain
concentration and void formation at triple points. These voids progressed
to form microcracks, which resulted in DDC.
The resultant beneficial or detrimental effect of the intergranular precipitates
on DDC resistance is determined by the balance between their effect
on preventing GB sliding and nucleating voids along the boundary.
The effect of the precipitates is defined by their type, size, distribution and
time of precipitation. The importance and specific effect of each one of
these factors on DDC resistance is not totally clear and requires further research.
However, the results presented here clearly indicate that the improvement
of DDC resistance by GB tortuosity and presence of medium
size intergranular (NbTi)C observed on FM-82 were more effective than
the numerous small intergranular M23C6 carbides observed on FM-52. This
DDC resistance improvement observed in the FM-82 is due to the balanced
effect of GB sliding control caused by both the intergranular precipitates
and the GB tortuosity. The poor DDC resistance exhibited by FM-52
is thought to be due to the rapid formation of voids and microcracks due to
the straight GBs and the large number of M23C6 carbides along the GBs.
Another aspect that has an important effect on DDC is the segregation
of impurities to the GBs. Sulfur had been shown to impair DDC resistance
and the observation of S rich films along migrated grain boundaries in FM-
52 is thought to be one of the factors responsible for the low DDC resis40
Phenomena and Mechanisms
tance of this alloy. Finally, hydrogen migration to the GBs and the intergranular
precipitate/matrix interfaces appears to have an important influence
on DDC susceptibility.
Acknowledgments
The authors would like to acknowledge BWX Technologies, Inc. for providing
partial funding for this research and the members of the Welding
and Joining Metallurgy Group at The Ohio State University, especially
Matt Collins, Nathan Nissley, Gustavo Guaytima, and Jeff Sowards for
their assistance and insightful discussions.
References
1. Rhines FN, Wray PJ (1961) Trans ASM 54: 117–128
2. Nissley NE, Collins MG, Guaytima G, Lippold JC (2002) (IIW Doc. IX-2050-
02) accepted for publication in Welding the World
3. Mintz B, Yue S, Jonas JJ (1991) Inter Mater Rev 36, 5: 187–217
4. Collins MG, Ramirez AJ, Lippold JC (2003) Weld J 83 (2), Part III: 39s–49s
5. Nissley NE (2002) Development of the strain-to-fracture test to study ductility
dip cracking in austenitic alloys. Masters Thesis, The Ohio State University,
Columbus, OH, USA, p 104
6. Nissley NE, Lippold JC (2003) Weld J 82(12): 355s–364s
7. Collins MG (2002) An investigation of ductility dip cracking in nickel-base
filler materials. Master Thesis, The Ohio State University, Columbus, OH,
USA, p 240
8. Collins MG, Ramirez AJ, Lippold JC (2003) Weld J 82(12), Part II: 348s–
354s
9. Collins MG, Lippold JC (2003) Weld J 82(10), Part I: 288s–295s
10. Feng Z (1993) A methodology for quantifying the thermal and mechanical
conditions for weld metal solidification cracking. The Ohio state University,
PhD Dissertation, Columbus, OH, USA, p 234
11. Ramirez AJ, Lippold JC (2004) Mater Sci Eng A Part I: in press
12. Ramirez AJ, Lippold JC (2002) Internal Research Report, The Ohio State
University, Columbus-OH, pp 1–35
13. Lippold JC, Clark WAT, Tumuluru M (1992) In: Cieslak MJ, Perepezko JH,
Kang S, Glicksman MW (Eds.) The Metal Science of Joining. The Metals,
Minerals and Materials Society, Warrendale, PA, pp 141–146
14. Mintz B, Yue S, Jonas JJ (1991) Inter Mater Rev 36, 5: 187–217
15. Cottrell AH (1961) In: Structural processes in creep. Report of a symposium
organized by the Iron and Steel Institute and the Institute of Metals, London, 3
and 4 May 1961, The Iron Steel Institute, London, pp 1–55
New Insight into the Mechanism of Ductility-Dip Cracking 41
16. Evans RW, Wilshire B (1993) In: Introduction to Creep. The institute of Materials,
London, UK, p 115
17. Kihara S, Newkirk JB, Ohtomo A, Saiga Y (1980) Metall Trans A, 11: 1019–
1031
18. Ramirez AJ, Lippold JC (2004) Mater Sci Eng A, Part II: in press
19. DuPont JN, Robino CV, Marder AR (1998) Weld J 77: 417s–431s
20. Johnson M, Fiore S, Lippold JC (2002) Evaluation of EN82H Filler Metal and
shielding Gas. EWI Project 45371CSP, EWI-OSU Final Report, Columbus-
OH, USA, June, 2002, pp 1–2
21. Lim LC, Raj R (1984) Acta Metall 32: 1183–1190
22. Lehockey EM, Palumbo G (1997) Mater Sci Eng 237A: 168–172
23. Shinoda T, Zaghloul MB, Kondo Y, Tanaka R (1978) Trans ISIJ 18: 139–148
24. Kotval PS, Venables JD, Calder RW (1972) Metall Trans 3: 453–458
25. Etienne CF, Rossum OV, Roode F (1980) In: Proceedings of the International
Conference Engineering Aspects of Creep, Sheffield, England, 15–19 Sept.,
1980, vol. II, The Institution of Mechanical Engineers, London, pp 113–121
26. Mathew MD, Sasikala G, Mannan SL, Rodriguez P (1993) Trans of the
ASME. J of Eng Mat and Technol 115: 163–170
27. Eberhart ME, Latanision RM, Johnson KH (1985) Acta Metall 33: 1769–1783
28. Tang S, Freeman AJ (1993) Physical Review B 47, 5: 2441–2445
29. Nishimoto K, Mori H, Hirata H (2001) In: Proceedings of the Today and Tomorrow
in Science and Technology of Welding and Joining. 7th JWS International
Symposium, Kobe, Japan, 20-22 Nov, 2001, Japan Welding Society,
vol 2, pp 827–832
Influence of Welding Speed on the Hot Cracking
Resistance of the Nickel-Base Alloy NiCr25FeAlY
During TIG-Welding
A. Slyvinsky, H. Herold, M. Streitenberger
Institute of Joining and Beam Technology, Otto-von-Guericke-University,
Magdeburg, Germany
Introduction
Efficiency, availability and operation limits of heat treating equipment and
industrial furnaces are often determined by the thermal load capacity of the
materials used. Within the range of nickel-base alloys, nickel-base hightemperature
materials form a group especially developed to meet the ever
increasing demands that arise in the construction of industrial furnaces and
heat-treating equipment.
In plant construction, the joining technology most often used for such
high-temperature materials is fusion welding. Here, of course, the emphasis
is placed on the production of high quality welded joints which guarantee
a useful life-time as well as retaining heat- and corrosion resistance.
One of the most often encountered difficulties in welding nickel-base alloys
is their hot cracking sensitivity due to thermically determined and material-
specific structural changes. Very often, these disadvantageous material
disconnections in the weld metal structure cannot be removed without
difficulties in the production process. On the other hand, as the hot
cracking resistance of welded seams is clearly dependent on the crystallization
character of the weld pool, there is, independently of the phase precipitation
and the micro-segregation process, a chance to achieve a reduction
of the proportion of directed growth of dendritic structures by
optimizing the welding parameters, mainly the welding speed.
Use and Structural Characteristics of the High-
Temperature Resistant Nickel-Base Alloy NiCr25FeAlY
The demand for operation temperatures up to 1200 °C with adequate lifetime
and corrosion resistance of the equipment is not exceptional today.
Influence of Welding Speed on the Hot Cracking Resistance 43
Thus, the alloy NiCr25FeAlY (Nicrofer 6025 HT, alloy 602 CA, W.-Nr.
2.4633), available in all semi-finished products since 1992, has shown an
outstanding useful life-time and heat- and corrosion resistance up to
1200 °C in practical use [1, 2].
The nominal chemical composition of the material is shown in Table 1.
Table 1. Nominal chemical composition of NiCr25FeAlY, % [3]
Ni Cr Fe C Mn Si Cu Al Ti Y Zr
min 24.0 8.0 0.15 1.8 0.1 0.05 0.01
max
bal.
26.0 11.0 0.25 0.1 0.5 0.1 2.4 0.2 0.12 0.10
Fig. 1. Microstructure in the HAZ of a TIG -weld of the alloy NiCr25FeAlY. In
the HAZ, the Cr-carbides are in solution; from them, eutectic clusters have
formed. The grain boundaries towards the fusion line (top left) are greatly covered
by a carbide layer; REM photo
With NiCr25FeAlY, carbide hardening with the evenly distributed primarily
precipitated carbides M23C6 with M = (Cr, Ni, Fe) was, for the first
time, put into practice on a large scale for workable nickel-base alloys [4].
The material owns a complex range of phases, covering, in the initial state,
􀈖-solid solutions and Cr-carbides of different morphology [5, 6]. This
structure is locally changed by the welding heat input. The metallographic
investigations of the welded joints of NiCr25FeAlY produced by different
welding procedures (TIG-, gas metal arc-, submerged arc- and manual
44 Phenomena and Mechanisms
electric welding) clearly point to a development of local eutectic fields in
the overheated zone of the HAZ (Fig. 1) as well as to numerous filmy
layers of carbides on grain interfacial areas in the HAZ and on dendrite interfacial
areas in the weld metal (Fig. 2).
Fig. 2. Microstructure of the weld metal deposit in a TIG-welded joint of the alloy
NiCr25FeAlY. Noticeable are the strong poly-crystalline layers of Cr-carbides on
dendrite interfacial boundaries. Etching method: Bloch und Wedl II. Scale 625:1
The structural changes shown here cannot be avoided by optimizing the
welding technology. They are typical of all welding procedures used for
nickel-base alloy NiCr25FeAlY, and they are connected with hot cracking.
Determination of the Hot Cracking Sensitivity
The Institute of Joining- and Beam Technology of the Otto von Guericke
University, Magdeburg, (IFST), has the equipment for a PVR-Test (Programmed
Deformation Rate Cracking Test) which is used to determine in
quantity the hot cracking sensitivity of the material NiCR25FeAlY and its
weld using selected welding technology. The principle of the test (Fig. 3)
is to produce a bead-on-plate weld on a flat tensile specimen simultaneously
with a tensile load of linearly increasing speed in welding direction.
In the test, a “critical tension speed” is defined as a criterion of hot
cracking resistance. This is to be understood as the speed at which the first
hot crack in the raw welding bead can be detected under at least tenfold
magnification.
Influence of Welding Speed on the Hot Cracking Resistance 45
Fig. 3. Diagram of the PVR-Test
In the PVR-Test of the nickel-base alloy NiCr25FeAlY, TIG-welding
without filler metal was employed. Compared to gas metal arc-, submerged
arc- and manual electric welding, the TIG-welding procedure can
guarantee a high weld quality with, at the same time, an as low as possible
heat input, especially if, according to the regulations [3], a “hot-cracking
resistant” gas mixture Ar + 1 % N2 is used as shielding gas for the production
of the test welds.
Fig. 4. First solidification cracks in the weld deposit of a PVR-specimen of the
material NiCr25FeAlY. Scale 10:1
46 Phenomena and Mechanisms
The first crack to be registered on this material is always an interdendritic
solidification crack (SC) in the bead-on-plate weld (Figs. 4, 5).
At higher tensile speeds, liquation cracks (LC) in the HAZ at the end of the
specimen are to be observed (Fig. 6). Their character is remarkable as they
solely form in direct combination with the solidification cracks as the continuation
of which they are to be considered in the overheated zone of the
HAZ (Fig. 7). The emergence of cracks as a result of a decrease in weld
ductility (DDC) is not observed.
Fig. 5. Inter-dendritic solidification cracks in the weld deposit of a PVR-specimen
of the material NiCr25FeAlY; Etching method: Bloech und Wedl II
Fig. 6. Liquation crack in the HAZ of a PVR specimen of the material
NICr25FeAlY
Influence of Welding Speed on the Hot Cracking Resistance 47
Fig. 7. Inter-dendritic solidification cracks in the weld deposit and liquation cracks
in the HAZ of a PVR specimen of the material NiCr25FeAlY
From the assessment of the hot crack types produced in the PVR-Test it
follows that the inter-dendritic solidification cracks in the weld deposit appear
to be the most immanent danger in arc welding sheets of
NiCr25FeAlY of medium thickness. This form of solidification cracks is,
in general, not typical of the materials group of “nickel-base super-alloys”,
as these rather tend to form re-flow cracks in the HAZ, according to the
mechanism of the so-called “constitutional re-flow” [7]. To assess the
welding-technological influence on this phenomenon, the energy input per
unit length of weld was varied by different welding speeds during a PVRTest
sequence (see Table 2).
The evaluated test results are shown in Fig. 8.
Table 2. PVR-Test. Assignment of parameters to the test variants
variant
No.
v
[cm/min]
I
[A]
U
[V]
E
[kJ/cm]
LC
[mm]
vcr.
[mm/min]
1 22.2 180 12.0 5.8 200.5 51.8
2 19.1 180 12.0 6.8 159.5 40.0
3 15.5 180 12.0 8.4 160.5 32.4
4 10.6 180 12.0 12.2 105.5 20.8
v welding speed, I welding current, U welding voltage, E energy input per unit
length of weld, LC distance up to the first hot crack, vcr critical tension speed.
48 Phenomena and Mechanisms
Fig. 8. Dependence of hot crack resistance of the material NiCr25FeAlY on the
welding speed (a) and the energy input per unit length of weld (b) in TIG-welding
(determined by the PVR-Test)
The critical tension speed is directly proportional to the welding speed,
and indirectly proportional to the energy input per unit length of weld, i..e.
the hot cracking susceptibility decreases with the increase in welding
speed under constant welding current and -voltage. In paper [8] it was
stated on the basis of numerous experiments with highly alloyed Cr-Nimaterials
that, with alloys susceptible to solidification cracking, the critical
tension speed lies at around 30 mm/min. A high resistance to solidification
cracking is shown by a weld metal deposit with a critical tension speed of
around 70 mm/min and above. According to this classification, and in view
of the test results, the material NiCr25FeAlY offers a resistance to hot
cracking, dependent on the energy input per unit length and the welding
speed, in TIG-welding with the shielding gas mixture Ar + 1 % N2. This
alloy turns out to be rather susceptible to hot cracking when welded at a
welding speed of below ca. 15.4 cm/min which corresponds to an energy
input per unit length of weld of about 8.4 kJ/cm.
Such an influence of welding speed on the hot cracking resistance is not
equally typical of all austenitic high-alloy materials. Thus, e.g., a contrary
effect of welding speed was pointed out for TIG- and submerged-arc
welding of fully austenitic chromium-nickel- and chromium-nickelmolybdenum
steels [9–11]. Here, low welding speeds (up to 10 cm/min),
with other welding parameters being constant, have proved to be favorable
for decreasing the hot cracking susceptibility , due to the type of primary
crystallization. In papers [12, 13], a drastic decrease in the plasticity capacity
of deposited weld metal deposits was observed in parallel with an increase
of welding speed and simultaneous maintenance of a constant weld
Influence of Welding Speed on the Hot Cracking Resistance 49
seam geometry. The influence of welding speed on the hot cracking susceptibility
in laser welding is shown in [14–16] by the example of the alloy
Inconel 718. Here, the hot cracking susceptibility increases with increasing
welding speed. This effect is due to the cross-sectional weld geometry
[14–16]. Investigations on electron-beam welded Inconel 903 showed
similar results, and a decrease in the hot cracking susceptibility by decreasing
the welding speed could be observed at the same time [17].
By contrast, the experiments with low-alloy steel described in [18]
showed an effective influence of increased welding speed with constant
welding current and welding voltage on the solidification crack resistance,
which correlates with the above mentioned results of the PVR-Test.
It must be pointed out that the above mentioned papers which describe
the direct as well as the indirect proportionality of welding speed and hot
crack resistance always relate to the observed influence of selected welding
parameters (welding speed) on the crystallization conditions of actual
weld metal deposits. Therefore, the phenomenon of the influence of welding
speed on the crystallization conditions in the weld metal deposit and
thereby on the hot cracking resistance of the alloy NiCr25FeAlY shall be
further clarified. For this reason, the solidification of the weld pool and the
influence of the welding speed on the crystallization rate and the microstructure
formation will be discussed in the following.
Calculation of the Mathematical Correlation Between the
Crystallization Rate in the Weld Pool and the Welding
Speed
Within the range of selected parameters for TIG-welding (Table 2), low
values of welding current and voltage guarantee a partly constricted arc
and an elliptical shape of the weld pool. At the same time, the crystallization
conditions in the weld pool are influenced by changing the welding
speed.
The welding speed indicates the level of thermal under-cooling of the
molten metal in the tail of the weld pool and turns out to be the most important
parameter that clearly influences the form and the size of the developing
primary dendrites.
From theory [18] we know that, at selected points of the weld pool isotherms,
the crystallization rate R can, for simple reasons of geometry, be
defined by the following empirical equation:
R = v cosβ , (1)
50 Phenomena and Mechanisms
where 􀈕 is the angle between the respective molten surface normal and
the direction of the welding speed v.
Fig. 9. Influence of the welding speed v on the crystallization rate R at selected
points of the weld pool isotherms, after Savage [12]
Eq. 1 was arrived at on the basis of numerous experimental investigations
and is beyond any doubt. The calculation of the crystallization rate
according to Eq. 1, however, causes difficulties as it necessitates experimental
finding of the angle 􀈕.
As a proportionality between the crystallization rate of the molten metal
in the weld pool R and the cooling rate W assumed, another analytic term
of the crystallization rate can be formulated (crystallization rate is here understood
as the travel speed of the phase interface in macro-volumes):
R = k W, (2)
where k is to be understood as a proportionality factor. For defining the
cooling rate W at any point of the phase interface in the weld pool, a
functional-analytic model simplification according to Rykalin can be employed
for the quickly moving high-power line-source in a plate [19, 20].
The theorist takes interest in the quickly moving high-power sources for
reasons of the possible simplification of the equation. The assumed quantities
q = I U (heat power) and v (speed of the heat source) are increased at
possibly constant q/v (energy input per unit length):
q→∞, v →∞, q/v = const.
Apart from this, the selected scheme for the model simplification is especially
suitable for calculating the temperature fields in areas near the
weld [20]. With the linear heat source, the heat is evenly distributed
throughout the thickness of the plate. The heat mainly spreads perpendicuInfluence
of Welding Speed on the Hot Cracking Resistance 51
lar to the travel direction of the heat source. The spread in travel direction
is negligibly small. According to the model selected, the increase in temperature
T – T0 can be calculated dependent from the transverse direction y
and the time t (according to Eq. 83 from [19]):
e (y at bt)
v c t
T −T = q − / 4 +
0
2
δ 4π λ ρ
.
(3)
In this is:
q – heat power of the source, [J/s];
v – feed rate of the heat source, [mm/s];
δ – thickness of the plate, [mm];
λ – coefficient of thermal conductivity, [J/mm s K];
c – specific heat capacity, [J/g K];
ρ – density, [g/mm³];
a – thermal conductivity, [mm²/s];
b – heat transfer coefficient at the plate, [1/s].
The heat transfer coefficient b is used for assessing the amount of heat
emitted by the plate surface to the surrounding medium. In the effective
range of the heat source, this coefficient is zero (b = 0). From this follows
for the equation of the melting temperature at any point of the interface in
the weld pool:
y at
s e
v c t
T q 2/ 4
4
= −
δ π λ ρ
.
(4)
In logarithmical form:
at
t y
v c
Ts q 4
ln
2
1
4
ln ln
2
= − −
δ π λ ρ
, or
(5)
t
v T c
q
at
y
s
ln
2
1
4
ln
4
2
= −
δ π λ ρ
.
(5.1)
From Eq. 4 it follows that the time of existence of the weld pool (y = 0)
is described by the following equation:
v δ T π λ cρ
t q
s
s 2 2 2 4
2
= .
(6)
52 Phenomena and Mechanisms
By substitution of ts after Eq. 6 in Eq. 5.1, the equation for the weld pool
surface appears as follows:
t
t
y2 = 2at ln s ,
(7)
with 0 ≤ t ≤ ts.
By differentiation of Eq. 5, the speed of temperature change at any point
of the weld pool surface is arrived at:
􀂸 􀂸 􀂹
􀂷
􀂨 􀂨 􀂩
􀂧
= −
∂
∂ 1
2 2
2
at
y
t
T
t
T s .
(8)
In Eq. 8, a dependence of the speed of temperature change at the points
at the weld pool surface on the interconnected parameters t and y was defined.
By substitution of Eq. 7 in Eq. 8 we get:
te
t
t
T
t
T s ln s
2
=
∂
∂
.
(9)
Fig. 10. Speed of temperature change (W=∂T/∂t) for points at the weld pool surface
The functional dependence of Eq. 9 is shown in Fig. 10. Obviously, with
0 ≤ t ≤ ts/e, the head part of the weld pool is considered (∂T/∂t > 0). For the
tail part of the weld pool (∂T/∂t < 0) applies ts/e ≤ t ≤ ts, so that Eq. 9 with
ts/e ≤ t ≤ ts defines the cooling rate of the molten metal in the weld pool:
Influence of Welding Speed on the Hot Cracking Resistance 53
te
t
t
W Ts ln s
2
= .
(10)
Point ts in Fig. 10 corresponds with point C in Fig. 9. After equalization
of Eqs. 2 and 10 for these boundary conditions (t = ts; R = v), we get the
new analytic term of the crystallization rate of the weld pool:
te
t
t
R = − v ts ln s , with s
s t t
e
t
≤ ≤ .
(11)
Eq. 11 is valid for any point in the tail end of the weld pool area (line BC-
A in Fig. 9).
This equation can be transformed into the Cartesian coordinate system.
Under consideration of t = x / v and correspondingly ts = Ls / v we arrive
at:
xe
L
x
R = − v Ls ln s .
(12)
In this are:
Ls – length of the weld pool;
x – linear coordinate of the point in the weld pool surface in longitudinal
direction from the line-source (at the tail end of the weld pool
surface Ls/e ≤ x ≤ Ls).
Discussion
In contrast to the empirical equation after Savage (1), the deduced formulae
(11) and (12) consider a number of characteristic particularities of primary
crystallization of the weld metal deposit which were described in
technical literature many years ago. Still in the sixties of the twentieth century,
practical investigations on weld seams [21, 22] showed that solidification
often does not correspond with the “conditions of orthogonal
growth”. This means that the axes of the dendrites which develop when the
weld pool solidifies are not necessarily in orthogonal direction to the isotherm
interfaces. If the direction of growth deviates from the orthogonal,
the crystallization rate of these dendrites increases. This will lead to several
dendrites rushing ahead and with this, a local disturbance of the otherwise
constant isotherm interfaces will arise [22]. For this case, the following
alteration of Eq. 1 was made by Wittke [21]:
54 Phenomena and Mechanisms
β
β
Δ
=
cos
R v cos .
(13)
Here, Δβ is an arithmetical difference between the “real” and the “orthogonal”
angle β. Unfortunately, the practical use of this equation has
turned out to be problematic due to the necessity to determine both the angle
β and the deviation angle Δβ by experiment. Besides, with very high
heat input and at very high welding speed (and consequently at very high
crystallization rate), a disoriented, fine-grain structure with equiaxial dendrites
may emerge along the weld seam center line by heterogeneous nucleus
formation [12]. In this case, the determination of the crystallization
rate in the central area of the weld seam cannot be accomplished by the
empirical Eq. 1. Accordingly, Eq. 1 after Savage is valid only in the special
case of orthogonal and epitaxial conditions of growth.
Fig. 11. Comparison of the traces of curves R = f (ß) after Eq. 1 and R = f (x) after
Eq. 12 (calculated for the welding speed v = 22.2 cm/min)
As in Eq. 1, a dependence of crystallization rate R from the feed rate of
the heat source (welding speed) v was made evident also in the Eqs. 11 and
12. Nevertheless, here the factor of proportionality is a transcendental
function. Its value is variable either from 0 with x = Ls/e to –1 with x = Ls
(Eq. 12) or from 0 with t = ts/e to –1 with t = ts (Eq. 11). Compared to Eq.
1, the analytical Eqs. 11 and 12 consider not only the influence of welding
speed v on the crystallization rate of the weld pool R, but also the influence
of welding processes and welding parameters. This is achieved by considering
the parameters ts (Ls) dependent on welding procedures and welding
data.
Influence of Welding Speed on the Hot Cracking Resistance 55
From the analysis of the Eqs. 11 and 12 follows that the flanks of the
weld pool as well as the areas near the point of deepest penetration in the
weld pool crystallize at a low rate. Thereby, smallest alteration of the coordinate
at the considered point on the weld pool surface causes a considerable
change of this rate (Fig. 11, right).
In contrast to the cosine curve after Eq. 1, Figs. 11 and 12 show that, not
only in point C but also in the far end of the weld pool tail (curve C – C',
Fig. 12), the amounts of crystallization rate and of the welding speed
strongly approach to each other (R ≅ v). That is the reason for special solidification
conditions in the tail of the weld pool (zone C - C'), so that the
dendrites can grow at any angle to the welding heat dissipation surface.
Fig. 12. Alteration in the time of the crystallization rate R in the tail end of the
weld pool
Fig. 13 shows the macro-structure of the welded seams that correspond
with the variants of welding parameters 1 (Fig. 13a) and 4 (Fig. 13b) from
Table 1. The macro-structure shows a mosaic-like character which pleads
for continuous nucleus formation during the dendritic growth. As follows
from Fig. 13b, a three-dimensional crystallization character (dendrites
growing with increasing contortion in welding direction) is typical of TIGwelding
NiCr25FeAlY. With an increase in welding speed v, the area with
fine-grain equiaxial dendrites in the center of the weld also enlarges (Fig.
13a). The building up of such a dendrite form in the center of the weld can
also increase the plasticity reserve of the weld metal deposit and contribute
to an accelerated formation of hot cracks.
56 Phenomena and Mechanisms
Fig. 13. Influence of welding speed on the weld microstructure in TIG-welding of
nickel-base alloy NiCr25FeAlY: (a) v = 22.2 cm/min, (b) v = 10.6 cm/min
Conclusions
In the paper presented, the influence of the welding speed on the hot cracking
resistance of TIG-welds of nickel-base alloy NiCr25FeAlY was investigated.
Increased hot cracking resistance at a higher welding speed and
constant welding current and voltage was observed.
The material under investigation shows a tendency to form interdendritic
solidification cracks dependent on the energy per unit length and
on the welding speed.
The experimental results of the PVR-Test allow it to select such welding
speeds for TIG-welding that minimize the hot cracking sensitivity. Therefore,
an energy amount of 8.4 kJ/cm at a welding speed of 15.5 cm/min
should not be exceeded.
An analytic expression for the calculation of the crystallization rate for
any point in the tail part of a weld pool was deduced which explains the
local influence of crystallization rate and of the welding speed on the hot
Influence of Welding Speed on the Hot Cracking Resistance 57
cracking sensitivity. This correlation is ascertained by the results of the
PVR-Test.
The way in which the crystallization rate throughout the weld pool surface
was changed gives reason for assuming the existence of a zone with a
higher crystallization rate and, resulting from this, a quicker crystallization
without diffusion in the tail zone of the weld pool. This would explain an
increase in the proportion of equiaxial dendrites in the weld center area,
the existence of which improves the hot cracking resistance of alloy
NiCr25FeAlY due to the increase in welding speed.
References
1. Brill U (1995) Practical experience with the new material Nicrofer 6025 HT
in furnace- and heat treatment plant construction (in German). Stahl 6: 37–40
2. Sölch R, Ammann T, Hoffmann T (2002) Use and welding processing of the
high-temperature nickel-base material Nicrofer 6025 HT (W.-Nr. 2.4633) (in
German). DVS-Berichte 219: 59–65
3. N. N. Nicrofer 6025 HT – alloy 602 CA. Material Data Sheet No. 4037, (edn)
October 2002
4. N. N. High-temperature alloys from Krupp VDM for industrial engineering.
Krupp VDM Report No. 25, (edn) September 1999
5. Slyvinsky A (2002) Structural characteristics and welding suitability of the
high-temperature nickel-base alloy NiChr25FeAIY (in German). In: Proceedings
of XXVII. Assistentenseminar „Fügetechnik/Schweißtechnik“, TU Chemnitz,
Oberwiesenthal, pp 145–152
6. Slyvinsky A, Veit P (2003) Structure and properties of the welded joints of a
high-temperature nickel-base alloy (in Russian). Aut Wdg 5: 7–13
7. Schulze G, Krafka H, Neumann P (1996) Welding: Materials – Design – Test
(in German). VDI, Düsseldorf
8. Klug P (1980) A contribution to the testing of hot-crack resistance of highalloy
welding filler materials with the PVR-test of VEW-Kapfenberg (in German).
PhD thesis, TU Graz
9. Kahovsky NI (1975) Welding of Stainless Steels (in Russian). Technika, Kiev
10. Fartushny VG, Kahovsky YuN, Savchenko VS, Demyanenko GP, Nastenko
GF (1974) Increase of the hot-crack resistance of austenitic weld metals (in
Russian). Aut Wdg 5: 39–43
11. Slyvinsky AM, Kahovsky YuN, Nastenko GF (1976) Influence of welding
speed on the crystallisation conditions in the weld pool (in Russian). Aut Wdg
8: 6–8
12. Yakushin BF (1969) Assessment of “Technological Strength” dependent from
welding parameters (in Russian). Wdg Prod 1: 7–9
58 Phenomena and Mechanisms
13. Rybakov AA, Mandelberg SL, Shitova LG, Kireeva TS (1980) Influence of
welding speed on the primary solidification of the weld deposit of low-alloy
steels (in Russian). Aut Wdg 10: 15–18
14. Fontana G, Gobbi S, Rivela C, Zhang L (1999) Laser welding in the manufacture
of superalloy components. Wdg International 8: 631–635
15. Shinozaki K, Kuroki H, Luo X, Ariyoshi H, Shirai M (1999) Effects of welding
parameters on laser weldability of Inconel 718. Study of laser weldability
of Ni-base, heat-resistant superalloys (1st Report). Wdg International 12: 945–
951
16. Shinozaki K, Luo X (1999) Phenomenon and Mechanism of Weld Cracking of
Ni-base Superalloys During Laser Welding. J of the Jap Wdg Soc 7: 22–26
17. Richards NL, Nakkalil R, Chaturvedi MC (1994) The Influence of Electron-
Beam Welding Parameters on Heat-Affected-Zone Microfissuring in
INCOLOY 903. Metall and Mater Trans 25A 8: 1733–1745
18. Rybakov AA, Mandelberg SL (1980) The influence of arc welding parameters
on the formation of solidification cracks in weld seams of low-alloy pipe steels
(in Russian). Aut Wdg 3: 12–15
19. Savage WF (1980) Solidification, segregation and weld imperfections. Wdg in
the world 5–6: 89–113
20. Rykalin NN (1957) Calculation of thermal processes in welding (in German).
VEB Technik, Berlin
21. Radaj D (1988) Heat effects of welding: temperature field, residual stresses,
distortion (in German). Springer, Berlin Heidelberg New York London Paris
Tokyo
22. Wittke K (1966) Characteristics of the primary crystallisation of the weld
metal (in German). Schweißtechnik (Berlin) 6: 289–292
23. Wittke K (1968) Modelling of the primary crystallisation by fusion welding (in
German). Schweißtechnik (Berlin) 7: 295–299
The Role of Segregation of Oxygen in Welding
Alloys of the INVAR Type
K.A. Yushchenko, V.S. Savchenko, T.M. Starushchenko
E.O. Paton Electric Welding Institute, Kyiv, Ukraine
Special thermal-physical properties of INVAR alloys (Fe-36 %Ni) combined
with high ductility and toughness, especially at low temperatures,
enabled their extensive application in a number of engineering fields.
Chemical composition of one of the alloys of the INVAR series is given in
Table 1.
Table 1. Chemical composition of alloy of the 36N type
Content of elements, wt. %
C Si Mn Ni Cr S P [O] [H]
0.05 0.1 0.35 35.6 0.45 0.004 0.006 0.002 0.00015
As shown by investigations, welds in these alloys are very sensitive to
hot cracking. Reportedly [1], the cracks are formed in two temperature
ranges, including temperatures below 1100 °C, i.e. at temperatures below
Ts (solidus temperature). They are located along the grain boundaries and
formed during cooling of the weld metal (Fig. 1).
Fig. 1. Under-bead hot crack formed in multi-layer welding of alloy 36N
60 Phenomena and Mechanisms
Low values of the concentration of sulphur and phosphorus do not prevent
hot cracking of the INVAR type alloys during welding. Investigations
show that formation of the said type of hot cracks can be initiated by oxygen
[1, 2]. Studies of high-temperature ductility of alloys and welds of the
INVAR composition indicate (Fig. 2) that metal loses ductility in a temperature
range of 600–1100 °C at a selected rate of plastic deformation
equal to 0.043 s-1. The higher the oxygen content of a material, the lower
its ductility.
Fig. 2. Temperature dependence of ductility of base and weld metals
Presence of mostly the brittle intergranular fracture with traces of microplastic
deformation can be seen in fractograph of the fracture surface of
the alloy at a temperature of 725 °C (Fig. 3).
Fig. 3. Fractograph of the fracture surface of specimens in alloy 36NKh at 725 °C
The Role of Segregation of Oxygen in Welding Alloys of the INVAR Type 61
It was suggested that the probable cause of hot cracking within a temperature
range of 600–1100 °C (DDC) could be a non-equilibrium diffusion
of atoms of the impurity elements (e.g. oxygen) contained in solid solution
at the grain boundaries. A decisive factor in this case, which
determines the rate of diffusion, is a high-temperature plastic deformation
that always takes place in the weld and HAZ metals during welding. Sensitivity
to formation of cracks in welding was quantitatively estimated by
dynamically deforming the weld metal during the welding process using
the testing machine of the Varestraint-Test type [3]. Plates measuring
4 x 150 x 150 mm were used as the test specimens. Welding of the specimens
was performed by the TIG method under the following conditions:
Iw = 120 A, Uarc = 12 V and vw = 10 m/h, the value of deformation of the
surface layers of a plate being kept constant and equal to 1.25 %.
The flow diagram of the cracking tests provides for performing twolayer
welding. The first layer is deposited on the base metal to evaluate the
properties of a weld. The second layer is deposited to simulate the properties
of the multi-layer welds. It includes two one-layer beads, the second of
them being deposited on the preliminarily made weld to overlap about
60 % of its surface area. Results of testing a series of the specimens differing
in the oxygen content are shown in Fig. 4.
Fig. 4. Effect of oxygen on brittle temperature of alloy Fe- 36Ni
The appearance of the welds after the tests is shown in Fig. 5. Analysis
of the results obtained shows that the welds in alloys of the INVAR type
are characterized by two ductility-dip ranges, i.e. BTR and DTR, with
oxygen affecting the ductility characteristics primarily in the DTR range
(Fig. 5 a). It should be noted that the sensitivity to under-bead micro62
Phenomena and Mechanisms
cracking is higher than in the weld made during testing (Fig. 5b). It is
likely that in this case the total content of oxygen affects the sensitivity to
embrittlement of the first weld both during producing it outside the Varestraint-
test machine and as a result of reheating, as well as during forced
dynamic deformation in testing using the above machine.
The fractographs of the surfaces of the ductility-dip cracks (DDC) in
Fig. 6, and the character of propagation of the cracks (Fig. 5b) show that
this is the case of brittle intergranular fracture with traces of plastic microdeformation.
Fig. 5. Appearance of the welds in alloy 36N after testing: (a) one-pass weld; 1 –
cracks in BTR; 2 – cracks in DTR; (b) two-pass weld, cracks (indicated by arrows)
in the first weld caused by deformation in making the second weld
Fig. 6. Fractograph of the surface of a hot crack formed in BTR
The Role of Segregation of Oxygen in Welding Alloys of the INVAR Type 63
The characteristics of the crack surfaces can be estimated as type "R"
(Matsuda) [4]. Element compositions of the surfaces of the ductility-dip
cracks and the surface of the intact cast metal used as a reference were
studied by electron Auger spectroscopy using the 3D analysis unit of the
LAS-2000 model. The distribution of elements through the thickness (from
the fracture surface) was studied by layer-by-layer etching with argon ions.
The working vacuum during the measurement process was 1.6⋅10-8 Pa.
This allowed the cleanness of a specimen surface to be maintained for 16 h
after removal of adsorbates and contaminants from it (by etching for up to
1 min). Profiles of distribution and variation in contents of main and impurity
elements during etching are shown in Fig. 7. It follows from the presented
data that the chemical composition of metal near the fracture surface
remains unchanged for 4–5 min of ion etching (Fig. 7a). The chemical
composition of matrix of the weld metal is revealed after 5 min of etching
(Fig. 7b), whereas near the fracture surface it cannot be revealed even after
230 min of continuous ion etching.
Fig. 7. Variation in the intensity of graphic display between iron, nickel and oxygen
on the fracture surface (a) and concentration of the above elements in the surface
layer (b) depending on the time of etching (τ) and the distance from boundary
(l): 1 – intact weld metal; 2 – crack surface
64 Phenomena and Mechanisms
It should be noted that carbon, as the interstitial impurity is for the most
part removed from the surface within 30 s of etching. Its content in deeper
layers of metal becomes lower than the Auger spectroscopy sensitivity
limit.
A similar picture is observed also in etching the surface of a hot crack.
Sulphur is removed from the surface within 15 s of etching. No nitrogen is
detected in the spectrum. Hydrogen cannot be detected by Auger spectroscopy
either. However, at increased temperatures (>400 °C), the mobility of
hydrogen is known to be so high that it exerts no effect on the formation
and growth of hot cracks in the weld metal, although the hydrogen content
of the latter is a bit higher than that of the base metal (Table 1).
The oxygen peak in both cases is seen for a much longer time: within
90 min of etching for the base metal, and for more than 230 min for the
fracture surface. It is likely that an oxide film about 1.3 μm thick is present
on the surface of the reference specimen, and that the oxide film on the
surface of a hot crack has thickness of more than 3.45 μm. Intensities of
peaks of individual elements on the crack surfaces in specimens in the aswelded
condition and after holding for 1.8 and 20 days in super high vacuum
(1⋅10-8 Pa) were compared to determine the possibility of existence of
oxygen mass transfer from the depth of the weld metal to the fracture surface.
This was done proceeding from the fact that settling of adsorbates
occurred uniformly and at the same intensity both on the fracture surface
and on the surface of the intact weld.
As proved by the experiments, sorption of oxygen occurs most intensively
during the first hours and days after surface etching. Then the rates
of adsorption and desorption of oxygen level off. The chemisorbed layer is
left on the surface, and the intensity of mass transfer of this impurity from
the depth to the surface remains almost unchanged for 20 days. Therefore,
despite the fact that the reactivity of oxygen is high and the coefficient of
its diffusion in steel at room temperature is very low (∼1⋅10-16 cm2/s), the
surface layer of a crack has an increased concentration of oxygen caused
by its transfer from the depth.
The experimental results suggest that oxygen redistribution takes place
in the metal studied, resulting in increased segregation of this impurity as
well as in an enrichment of boundaries and in a depletion of the bulk of
grains in oxygen. It is believed that it is the peculiarities of the mechanism
of intragranular deformation of the Fe-36Ni metal that can play a decisive
role in changing the resistance to under-bead cracking.
According to the generally accepted concept, the process of realisation
of intragranular deformation occurs either by involving a dislocation
(translation) or turning (rotation) mechanism [5], depending upon the
The Role of Segregation of Oxygen in Welding Alloys of the INVAR Type 65
external factors and on the type of deformed metal. Moreover, characteristic
regularities of the latter most often show up under complex thermal deformation
conditions. It is thought that differences between the above
mechanisms have a substantial effect on the resistance to plastic deformation,
and thus on the formation of under-bead cracks. In this connection, it
is advisable to check peculiarities of deformation (including the intragranular
one) of the weld metal having a different sensitivity to under-bead
cracking.
Fine weld metal structures determining the peculiarities of variations in
their dislocation composition during the plastic deformation process as
well as their resistance to high-temperature deformation were evaluated by
the examples of two systems of alloying the welds (Table 2), i.e. with high
and low sensitivity to cracking.
Comparative evaluation of variations in the fine structure was performed
on specimens with a gauge part diameter of 4 mm cut from the
metal of the upper bead of the multi-pass welds in the as-welded condition
as well as after additional deformation (by 10 %) at a temperature of
700 °C and a rate of 5.66⋅10-2 s-1 corresponding to the DTR temperatures.
In alloying system I (Table 2), increase in the deformation rate by 10 %
leads to strong banded structure formation in the internal volumes (deformation
bands), i.e. mechanical twins, and bands associated with collective
forms of movement of the crystalline lattice defects propagating in the entire
bulk of grains (Fig. 8a). At displacement of the deformation bands, the
low-angle grain boundaries do not exert any delay effects, whereas the
high-angle grain boundaries block further movement of the slip bands into
the neighboring grains (Fig. 8b). Initiation of banded structures occurs in
certain regions of a chaotic dislocation ensemble, which is characterised by
a high density of the lattice defects (ρ ≈ 1⋅1011–1⋅1012 cm-2).
Table 2. Chemical composition of materials under consideration
Content Alloy t of elements, wt. %
C Mn Si Cr Ni N Mo
Others Sensitivity
to underbead
cracking
Ia 0.05 0.35 0.1 0.45 35.6 - - 0.004 S
0.006 P
0.002 [O]
0.0015 [H]
High
IIb 0.030 11.2 0.50 19.6 13.5 0.25 2.6 0.004S Low
a Material 36N.
b Material 03Kh18N19G10AM3 (Fe-Cr-Ni-Mn-N).
66 Phenomena and Mechanisms
Fig. 8. Fine metal structure of the alloy of alloying system I after forced deformation
at 700 °C: (a) intragranular (b) boundary zone (arrow A indicates the region
of initiation of banded structures)
The location of individual dislocations along the crystallographic slip
systems in the bulk of grains in a metal structure of alloying system I, in
the case of absence of forced deformation and emergence of extended
banded formations in grains as the external load is increased, indicates that
intragranular plastic deformation in the metal occurs by the classical dislocation
(translation) mechanisms [5]. In addition, the "strength" (extension)
of translations and a substantial degree of mobility of the lattice defects in
the stress field are caused in many respects by high values of the energy of
stacking faults of the Fe-Ni system metal, which are known to serve as a
weak obstacle for the deformation to propagate. It should be emphasised
that strong collective displacements of the lattice defects in the bulk of
grains, oriented to the boundaries, are accompanied by a "transportation"
of impurity elements and their clustering in a region of intergranular
boundaries, which was experimentally observed in the dark-field structural
and microdiffraction images. It is likely that the specific character of displacement
of the lattice defects during the process of plastic deformation
results in the presence of segregation clusters and new phases along the intergranular
boundaries, as well as their absence in the bulk of grains.
Therefore, as shown by the detailed analysis of structural elements, distribution
of phases and peculiarities of the plastic deformation mechanism,
a very heterogeneous structure with a considerable gradient of dislocation
The Role of Segregation of Oxygen in Welding Alloys of the INVAR Type 67
density and presence of phase precipitates and segregations between the
bulk of grains and high-angle boundaries is formed in the weld metal of
the Fe-Ni alloying system.
The weld metal structure of alloying system II (Fe-Cr-Ni-Mn-N) with a
high cracking resistance is characterised first of all by the presence of
phase precipitates with a high dispersion degree (dph.p ≈ 0.04–0.21 μm) and
by their heterogeneous distribution in the entire volume of the weld metal
(particle spacing lp ≈ 0.25–0.3 μm).
The uniformly increasing degree of deformation in the bulk of grains
entails a higher dislocation density and a greater delay of displacements in
the crystallographic slip planes. The split dislocations are often seen. This
points to a total decrease in the energy of stacking faults, γs.f., of metal at
the transition from alloying system I to system II.
The delay of displacements (translations) of individual dislocations
within the slip systems most probably results from a total decrease in γs.f.,
based on alloying. Naturally, this makes the plastic deformation more difficult
to occur by the classical dislocation mechanism, of which a characteristic
element is a transverse slip of the dislocations.
In this case, for metal with a low value of the stacking fault energy, dislocation
transition to the other slip plane may take place, provided that particular
dislocations combine ("draw together") into the total one. The wider
the stacking fault, (i.e. the lower the stacking fault energy), which is characteristic
of metals of alloying system II, the higher the energy intensity of
the process.
It is presumed that complex dislocation rearrangements of the above
type and interactions of dislocations of different slip systems lead to the
formation of a reticular structure [5] (Fig. 9).
One of the consequences of a change in the deformation mechanism is
the formation, on the basis of the reticular structure, of dispersed fragments
approximately 0.25-0.30 μm in size, framed by the zones with a high dislocation
density (Fig. 9d). Because of this, the neighbouring microvolumes
turn to finite angles. In such cases, relaxation of internal stresses occurs by
plastic turns (rotations).
Therefore, the conducted experiments show that under similar thermal
deformation conditions the weld metals of different alloying systems are
characterised by the formation of structures greatly differing in presence
and distribution of impurities, morphology of phase precipitates, sizes of
grains and substructural elements, and character of displacement of the
crystalline lattice defects within the effective stress field. The latter points
to the fact that the plastic deformation mechanism occurring in certain
microvolumes of the weld metals with different alloying systems is also
68 Phenomena and Mechanisms
different. In the case of alloying system I, the deformation mechanism is
mostly of a dislocation character, whereas in the case of alloying system II,
the other mechanism, i.e. of a rotation character, is involved (Figs. 8 and
9). It can be suggested that a change in the mechanism exerts a decisive effect
on the processes of mass transfer of impurity elements causing embrittlement
along the grain boundaries. In the first case (occurrence of the
translation mechanism, Fig. 8a), an intensive intragranular slip to distances
comparable with the grain size (where dislocations move in the bulk of
grain to its boundaries) favours the dislocation transfer of these elements to
the intragranular boundaries. In the second case (with involvement of the
rotation mechanism, Fig. 9), despite an identical deformation of metal,
there is no explicit trend to movement of dislocations directed to the grain
boundaries. Here, the directed macrodeformation of a specimen at a microlevel
occurs primarily through the rotation movement of elements of an
intragranular structure. As applied to the INVAR type alloys, the noted
specific character of the plastic intragranular deformation mechanism leads
finally to an enrichment of the intergranular boundaries with oxygen,
which causes the formation of under-bead cracks in multi-layer welds (Fig.
10).
Fig. 9. Fine structure of the weld metal of alloying system II: (a) dispersed precipitates
of redundant phases in the bulk of grains (b) dislocation composition of
the boundary zone (c) split dislocations in weld structure (d) fragmentation of
metal structure after forced deformation at 700 °C
The Role of Segregation of Oxygen in Welding Alloys of the INVAR Type 69
Fig. 10. Structures formed in the deformed metal of (a) alloying system I and
(b) alloying system II
Conclusions
1. Under-bead hot cracks in multi-layer welds of the INVAR type are
formed along the grain boundaries.
2. An increased content of impurity elements, and first of all oxygen, is
observed on the surfaces of under-bead cracks in multi-layer welds of
the INVAR type.
3. Enrichment of grain boundaries with oxygen during welding, leading to
hot cracking of the welds, is controlled by the plastic deformation
mechanism of a dislocation (translation) character.
References
1. 􀉘􀉳􀉟􀉧􀉤􀉨 􀉄􀈺, 􀉋􀉬􀉚􀉪􀉭􀉳􀉟􀉧􀉤􀉨 􀉌􀉆, 􀉋􀉚􀉜􀉱􀉟􀉧􀉤􀉨 􀈼􀉋 (1983) 􀉆􀉟􀉯􀉚􀉧􀉢􀉡􀉦
􀉨􀉛􀉪􀉚􀉡􀉨􀉜􀉚􀉧􀉢􀉹 􀉝􀉨􀉪􀉹􀉱􀉢􀉯 􀉬􀉪􀉟􀉳􀉢􀉧, 􀉜􀉵􀉡􀉵􀉜􀉚􀉟􀉦􀉵􀉯 􀉤􀉢􀉫􀉥􀉨􀉪􀉨􀉞􀉨􀉦, 􀉩􀉪􀉢 􀉫􀉜􀉚􀉪􀉤􀉟
􀉂􀉧􀉜􀉚􀉪􀉚. 􀈺􀉜􀉬􀉨􀉦􀉚􀉬􀉢􀉱􀉟􀉫􀉤􀉚􀉹 􀉫􀉜􀉚􀉪􀉤􀉚. 􀊋8: 􀉫.5–7
2. 􀉘􀉳􀉟􀉧􀉤􀉨 􀉄􀈺, 􀉋􀉬􀉚􀉪􀉭􀉳􀉟􀉧􀉤􀉨 􀉌􀉆 (1981) 􀉊􀉨􀉥􀉶 􀉤􀉢􀉫􀉥􀉨􀉪􀉨􀉞􀉚 􀉜 􀉨􀉛􀉪􀉚􀉡􀉨􀉜􀉚􀉧􀉢􀉢
􀉬􀉪􀉟􀉳􀉢􀉧 􀉩􀉪􀉢 􀉫􀉜􀉚􀉪􀉤􀉟 􀉢􀉧􀉜􀉚􀉪􀉚. 􀈺􀉜􀉬􀉨􀉦􀉚􀉬􀉢􀉱􀉟􀉫􀉤􀉚􀉹 􀉫􀉜􀉚􀉪􀉤􀉚. 􀊋8: 􀉫.21–24
70 Phenomena and Mechanisms
3. Savage WF, Lundin GD (1965) The Varenstraint Test. Welding Journal,
44(10): 433–442
4. Matsuda F, Nakagawa H, Ogata S, Katayawa S (1978) Fractographic Investigation
on Solidification Crack in the Varestraint Test of Fully Austentic Steel
– Studies on Fractography of Welded Zone (III). Transactions of IWR I, 7(1):
59–70
5. 􀉏􀉢􀉪􀉬 􀈾􀉠, 􀉅􀉨􀉬􀉟 􀉂 (1972) 􀉌􀉟􀉨􀉪􀉢􀉹 􀉞􀉢􀉫􀉥􀉨􀉤􀉚􀉰􀉢􀉣. 􀉆. 􀈺􀉬􀉨􀉦􀉢􀉡􀉞􀉚􀉬: 599 c
Character of Hot Crack Formation During
Welding of Cast Heat-Resistant Nickel Alloys
K.A. Yushchenko, V.S. Savchenko, N.O. Chervyakov, A.V. Zvyagintseva
E.O. Paton Electric Welding Institute, Kyiv, Ukraine
Heat-resistant nickel alloys are the main structural materials used to manufacture
gas turbine engines widely applied in aircraft engineering and other
industries. Engine sections for high-temperature operation are made using
precipitation-hardening nickel alloys with intermetallic strengthening. To
ensure the structure stability and maintain high long-term strength properties,
nickel alloys are provided with complex alloying systems containing
γ'-forming elements (Al, Ti, Nb), the total content of which in an alloy
amounts to 6–15 % or more. Complex alloying systems and high strength
properties of the alloys lead to crack formation in the weld and heataffected
zone in welding metal of even small thickness. This stipulated the
interest in investigating the principles of hot crack formation first of all in
the heat-affected zone during fusion welding in terms of structural transformations.
Investigations were conducted on nickel alloy IN 738 used as a
structural material for the manufacture of gas turbine blades. The chemical
composition of the alloy is given in Table 1.
Table 1. Chemical composition of alloy IN 738
Content of elements, wt. %
C Cr Co Mo W Nb Al Ti Ca La
0.09 16.0 10.5 1.7 4.6 0.2 3.0 4.4 ≤0.01 ≤0.01
The sensitivity of the HAZ to cracking in plasma-powder welding using
a powder additive with a composition identical to that of the base metal
was evaluated. Welding was performed on an alloy 10 mm thick in the asreceived
condition. Approximate welding parameters are given in Table 2.
Table 2. Parameters used to make joints in nickel alloy by plasma powder welding
Iw
[A]
Uw
[V]
Vw
[m/h]
Powder
particle size
[μm]
Argon flow
rate
[l/min]
100-120 25-26 4 50-150 18-20
72 Phenomena and Mechanisms
Sections for metallographic examinations were subjected to vacuum ion
etching using a high-voltage plasma discharge at a voltage of 2.5 kV and a
current of 0.005 A. This treatment reveals distinct contours of cracks of
different sizes. Dark- and light-field optical microscope images were used
for verification revealing the zones of structural transformations in the
HAZ. Sizes of structural metal components, including the γ'-phase, were
estimated by scanning electron microscopy (SEM) after special etching.
The character of plastic deformation on the incidence of hot crack propagation
was estimated from changes in the surface profile of a welded joint
by interference optical microscopy.
Metallographic analyses of welded joints in alloy IN 738 made with the
weld edges at room temperature revealed microcracks propagating into the
base metal (Fig. 1). A structural zone adjoining the weld, wherein the hot
cracks are mainly located, was detected by dark-field image analysis (Fig.
2a, b).
Fig. 1. Crack in the HAZ of a welded joint in nickel alloy
with the γ’-phase strengthening
Statistical data processing shows that the cracks are located at some distance
from the fusion line. This suggests that the formation of the hot
cracks is not always caused by grain boundary melting during welding.
Therefore, these cracks cannot be classed as solidification cracks.
Zones with properties other than the base metal properties in the asreceived
conditions are formed in the base metal under the effect of the
welding thermal cycle. Investigation of the formation mechanism of such
zones will help to clarify the nature of hot cracking.
Character of Formation of Hot Cracks in Welding 73
Fig. 2. Microcracks in the HAZ of heat-resistant alloy IN 738,
a – crack in the light field
b – the same crack in the dark field
Generally, structural transformations in the HAZ of welded joints depend
upon the welding conditions and, first of all, upon the thermal cycles.
Investigations were conducted to study the effect of the initial temperature
on the formation of the structural zone under consideration, the welding
parameters being kept unchanged.
Welding was performed on metal:
− preliminarily cooled to –196 °C;
− at room temperature;
− on plates heated to 1000 °C.
The HAZ-structure of welded joints in alloy IN 738 is shown in Fig. 3.
It can be seen from this figure that the zone, L, undergoes substantial
changes across its width. A generalization of the results obtained, shown in
Fig. 4, indicates an exponential dependence of variations in the zone width
upon the temperature. As confirmed by statistics, the width of the zone
correlates with a mean length of the hot cracks resulting from different initial
temperatures of the metal prior to welding.
74 Phenomena and Mechanisms
Fig. 3. Changes in width L of the zone of complete γ'→γ→γ' transformations in
welding of heat-resistant nickel alloy with γ'-phase strengthening
Welding was performed at the following temperatures of the alloy IN738:
a – 77 K
b – 293 K
c – 1273 K
Character of Formation of Hot Cracks in Welding 75
Fig. 4. Effect of temperature of alloy IN 738 prior to welding
on the mean length of cracks and zone width L
The examination of the structure of the zone adjoining the weld (Fig. 5a,
b) revealed precipitates of the finely dispersed γ'-phase, in contrast to
coarser precipitates of the γ'-phase contained in the base metal at some distance
from the fusion line (Fig. 5c).
Scanning electron microscopy proves differences in structure and sizes
between the γ'-phase in the base metal (as-received conditions) (Fig. 6b) to
those in the heat-affected zone adjacent to the weld (Fig. 6a).
It should be noted that changes in sizes of the γ'-phase lead to changes in
strength properties of the zone examined, including the surface (Fig. 7). It
can be seen from this figure that the hardness of the HAZ changes
depending upon the temperature of metal prior to welding. Maximal hardness
is achieved in welding of metal preliminarily cooled to a temperature
of –196 °C. Metal welded after preheating has a minimal hardness. Therefore,
the width of the zone with increased hardness as well as the width of
the zone wherein the cracks are formed also depend upon the initial temperature
conditions prior to welding.
A welded joint made on metal preliminarily cooled to –196 °C has a
minimal width of the zone with an increased hardness. These differences
can be explained in terms of the principles of variations in size of the
strengthening γ'-phase, depending upon the cooling rate within the temperature
range of the γ→γ' transformation.
76 Phenomena and Mechanisms
It was shown [1] that an increase in the cooling rate was accompanied
by a decrease in the size of the γ'-phase (Fig. 8). One might expect that a
change in the strength characteristics, including hardness, takes place in
this case.
Fig. 5. HAZ structure of heat-resistant nickel alloy IN 738:
a – general view
b – fragment 1
c – fragment 2
Fig. 6. Size of the strengthening γ'-phase in different zones
adjoining the alloy IN 738 weld:
a – zone of hot crack formation
b –base metal (as-received conditions)
Character of Formation of Hot Cracks in Welding 77
Fig. 7. Dependence of HAZ microhardness on the temperature
of the metal prior to welding:
1 – welding with preliminary cooling (-196 oC)
2 – welding at room temperature (20 oC)
3 – welding with preheating (1000 oC)
Fig. 8. Variations in mean size of the γ'-phase
during cooling from temperature T > Tsolv
78 Phenomena and Mechanisms
In a general form, the strength of precipitation-hardening alloys depends
upon the distance between the particles, their size and volume fraction [2],
and can be determined from the following Eq. 1:
d
c f
σ =σ 0 + ,
(1)
where
σ0
is the tensile strength of the matrix;
c is the constant that includes the Burgers vector and shear modulus
of the matrix;
f is the volume fraction of the precipitates;
d is the diameter of a particle.
This equation shows that the higher the volume fraction of the strengthening
phase and the smaller the size of a particle, the stronger the alloy.
Therefore, it can be concluded that the hot cracking zone adjoining the
weld metal has an increased hardness (strength) due to the γ'→γ→γ' transformations
(following the heating → cooling cycle) taking place under the
effect of the welding thermal cycle (Fig. 9). Here, differences in hardness
are caused by differences in the rate of metal cooling during welding, and
thus by differences in the occurrence of the γ→γ' transformation process
during cooling of the HAZ metal after the weld pool has moved in welding
direction. In a general form, the diagram of preferable formation of the
cracks in the weld zone of welded joints in nickel alloys with the γ'-phase
strengthening and its relationship with structural transformations in the
HAZ are shown in Fig. 10. Hot crack propagation along the grain boundaries
from zone 1 (Fig. 10) to the base metal is retarded by relaxation of local
stresses, owing to plastic deformation taking place at the crack mouth
(Fig. 11). It can be seen from the diagram in Fig. 11b that this is the case
for the state of plane deformation. Apparently, in this case the principles of
retardation of a crack can be described by conditions of formation of the
plastic deformation zones in metallic materials in terms of fracture mechanics.
The results obtained are confirmed by interference optical microscopy
of the surface of a hot crack mouth after welding of a nickel alloy
(Fig. 12). The figure shows the interference pattern of the crack mouth as
well as of a region of artificial deformation induced by a scratch on the
section surface. It can be seen that the course of the interference fringes a
region of the crack mouth and the scratch tends to move upward. As the
scratch forms a depression in the metal, the zone of the metal ahead of the
crack moves downward relative to the section plane, i.e. shrinkage of
metal takes place due to realization of the plastic deformation mechanism.
Character of Formation of Hot Cracks in Welding 79
Fig. 9. Temperature distribution in the HAZ of a welded joint in alloy IN 738 and
character of structural transformations in realization of the heating → cooling
welding cycle
Fig. 10. Diagram of the preferable formation of cracks in the HAZ of joints in
heat-resistant alloys with γ'-phase strengthening. Structural components in the
HAZ at the moment of high-temperature heating
80 Phenomena and Mechanisms
Fig. 11. Retardation of hot crack propagation in the HAZ due to the plastic deformation
mechanism:
a – appearance of the surface of metal at the hot crack mouth
b – diagram of distribution of plastic deformation in stress fields
Character of Formation of Hot Cracks in Welding 81
Fig. 12. Distribution of plastic deformations at the hot crack mouth
The investigation results suggest the following mechanism of hot cracking
in the HAZ during welding of nickel alloys with γ'-phase strengthening:
1. In welding of an alloy, heating of the HAZ metal causes structural transformations
of the strengthening γ'-phase to propagate into the γ-matrix.
These transformations take place at temperatures from about 700 °C to
Tsolv, depending upon the chosen alloying system for a metal.
82 Phenomena and Mechanisms
2. The γ→γ' transformation is accompanied by an increase in the metal
volume, as at increased temperatures the crystalline lattice parameters
for γ' are smaller than those for γ. This process is accompanied by the
development of intensive thermal deformation processes localized in a
narrow HAZ.
3. There is a high-temperature zone adjoining the weld, where the structure
of a nickel alloy remains in a fully austenitic state for a certain period of
time.
In this case a mechanism of enrichment of the austenite grain boundaries
with impurity elements (carbon, oxygen, sulphur, phosphorus and
other surface-active elements) takes place. Contribution to the embrittlement
processes by plastic deformation is realized by way of interaction of
mobile dislocations with impurity atoms and determined by the competing
temperature processes, such as changes in the energy of interaction of atoms
with dislocations (inverse temperature dependence) and the diffusion
mobility of these atoms (direct temperature dependence).
The temperature range in which a sufficiently high energy of bonding of
an impurity atom with dislocations is maintained and, at the same time, ensured
by its mobility, owing to increase in the diffusion parameters, is the
ductility-dip temperature range (DTR), as in this range an impurity element
may be captured by moving dislocations and transported to the
boundaries. The latter provides a change in the grain body to boundary
strength ratio and leads to the formation of a crack (DDC) along the grain
boundaries. As noted earlier, embrittlement of the grain boundaries takes
place in a zone with austenitic structure.
As the metal cools down, the process of precipitation of the γ'-phase occurs
in the bulk of grains. This is accompanied by the development and, as
the amount of the γ'-phase increases, enhancement of the process of retardation
of the dislocation movement as well as by an enrichment of the
boundaries with impurity elements and an increase in the HAZ ductility. In
this temperature range, the propagation of a crack is arrested due to realization
of plastic deformation at the crack mouth.
Conclusions
1. Nickel alloys with γ'-phase strengthening are characterized by the formation
of cracks along the grain boundaries in the heat-affected zone
during heating and cooling in a temperature range from 700 °C to temperatures
close to the melting point.
Character of Formation of Hot Cracks in Welding 83
2. It is shown that phase transformations by the γ+γ'→γ→γ'+γ mechanism
within the base metal zone near the fusion line as well as local deformation
along the grain boundaries occur during the welding process. Relaxation
of stresses takes place in a crack formed intergranularly (at the
crack mouth) at temperatures from 700 C to Tmelt.
3. The probable mechanisms of formation of cracks in the HAZ include:
• a process of thermal deformation loading accompanied by segregation
of impurity elements at the grain boundaries, followed by their melting
and cracking;
• a process of thermal deformation loading resulting in a substantial local
deformation within the ductility-dip temperature range (DTR), thus
leading to cracking;
• the formation of cracks in metals with a high degree of alloying with
surface-active elements, such as boron, may occur by the above two
mechanisms.
References
1. Haafkens MH, Mathey HG (1982) A New Approach to the Weldability of
Nickel-Base As-Cast and Powder Metallurgy Superalloys. Welding Journal,
vol 61, no. 11
2. 􀉆􀉚􀉫􀉥􀉟􀉧􀉤􀉨􀉜 􀉋􀈻 (1983) 􀉀􀉚􀉪􀉨􀉩􀉪􀉨􀉱􀉧􀉵􀉟 􀉫􀉬􀉚􀉥􀉢 􀉢 􀉫􀉩􀉥􀉚􀉜􀉵. 􀉆.: 􀉆􀉟􀉬􀉚􀉥􀉥􀉭􀉪􀉝􀉢􀉹:
192 􀉫
Contribution to HAZ Liquation Cracking
of Austenitic Stainless Steels
P. Bernasovský
Welding Research Institute – Industrial Institute of SR,
Bratislava, Slovakia
Abstract
The paper shows results of 30 experimental stainless steels which were
subjected to two laboratory liquation cracking tests. Namely the
Thermorestor–W test with temperature-strain simulation of welding cycles
(Japanese) and the LTP-1-6/TIG test (Russian) were employed. Preliminary
criteria of liquation cracking tests (3 degrees) are proposed. Comparison
of experimental results with the calculation indices L, ΔH and
ECr/ENi is presented. A parametric equation tgαK based on the simulation
test results is also proposed. Thanks to a vacuum chamber of the
Thermorestor-W simulator, the fracture surfaces of liquation cracks are
suitable for further study of metallurgical cracking. Examples of low
melting eutectic phases revealed by microfractographic analysis are
shown. The validity of calculation indices and the preliminary test criteria
of liquation cracking have to be further verified with the results obtained
on real joints.
Introduction
Austenitic steels already have a certain inherent susceptibility to hot crack
formation in welded joints (WJ), given by the nature of face-centered
cubic lattice. Therefore, it is important to eliminate the other possible unfavorable
factors (metallurgical, structural and technological ones) during
welding.
Evaluation of hot cracking susceptibility in base metals (BM), means
the evaluation of their resistance to liquation cracking and to ductility-dip
Contribution to Liquation Cracking of Austenitic Stainless Steels 85
cracking, which both are formed in the base metal heat affected zone
(HAZ) of welded joints.
The ductility-dip cracks, which are formed in the lower temperature
range (850–1100 °C), are usually related to decreased cohesion strength of
grain boundaries during their migration. However, ductility-dip cracking is
relatively uncommon. So this presentation is preferably devoted to liquation
cracks in austenitic stainless steels.
Liquation cracks are usually short, they reach only several grain diameters
or at maximum several millimeters, and in general they are of subsurface
character. Employing old conventional non-destructive techniques,
they could be sometimes hardly identified.
But, in applications where there is low defect tolerance (e.g. nuclear
power plant components), their occurrence is inadmissible. Even though it
is known that liquation fissures in austenitic steels affect only a little their
strength properties and toughness, they may under certain circumstances
reduce the fatigue, corrosion and creep properties of a welded joint. Their
effect on reheat crack initiation and stress corrosion cracking is also reported.
Fig. 1. Eutetic NbX on liquation crack surface in AKOR 2 steel
Liquation cracks are always intercrystalline. On fracture surfaces of
liquation cracks, we can find residues of solidified molten metal in the
form of the eutectic secondary phases (Fig. 1) or the round grains with
typical bridges between them (Fig. 2) or thermal facets (Fig. 3) which enable
distinction of liquation cracks from other lower temperature types.
86 Phenomena and Mechanisms
Fig. 2. Typical bridges between grains on the liquation fracture surface
of 17 259 steel
Fig. 3. Thermal facets on the hot tensile fractured surface of 17 349 (K) steel
Mechanism of Liquation Cracking
The formation of liquation cracks is conditioned by the presence of a
liquid film on grain boundaries during the cooling phase of the welding
thermal cycle when the thermal tensile stresses are induced. The height of
Contribution to Liquation Cracking of Austenitic Stainless Steels 87
temperature and strain gradient the so-called temperature strain gradient
[1] depends mainly on technological and structural factors.
The formation of a liquid film on grain boundaries is explained by several
processes:
a) Preferential melting of grain boundaries which can be formed also in
relatively pure alloys, especially when the dendritic segregation extends
the solidification interval. This is the case with some nickel alloys
and some austenitic stainless steels;
b) The segregation (diffusion) of impurities or deliberately added alloying
additions on the grain boundaries where they decrease the melting
temperature. These segregates can precipitate as secondary phases (e.g.
see Fig. 1) or they can remain in solid solution (Fig. 2);
c) The penetration of a liquid film formed from low melting precipitates
entrapped on the migrating grain boundaries in the overheated HAZ
[2];
d) Migration of surface-active elements from the weld metal (WM) along
grain boundaries into the adjacent HAZ where they form a liquid film
[3]. This mechanism does not depend on the chemical composition of
base metal.
The formation of liquation cracks depends on the deformability (hot
ductility) of metal and on the intensity of high-temperature strain of metal
in welding. Mathematical processing of a physical module of the liquation
cracking susceptibility is attractive, it was however hitherto impossible to
plot a suitable model including both high-temperature strain and hotdeformability
of metal. Problems arise especially with the expression of
deformability of metal in intricate non-equilibrium conditions which occur
in welding.
Effect of δ-Ferrite and of the Chemical Composition of
Steel on Liquation Cracking
δ-ferrite is the most significant factor for hot cracking elimination in austenitic
steels. Whereas the presence of δ-ferrite in weld metal at normal
temperature ought to be 2–7 %, in base metal even less, up to 1 % is sufficient
[4].
However, the primary ferrite segregated during solidification is decisive
from the viewpoint of cracking elimination. According to the pseudobinary
Cr-Ni diagram in Fig. 4 it is segregated on the right from point A through
the L → L + δ → L + δ + γ transformation or through L → L + δ trans88
Phenomena and Mechanisms
formation (right from point B). The steels which solidified on the left from
point A with primary segregated austenite are more susceptible to
cracking, because with falling temperature the solubility of nickel in δ-
ferrite is decreased and ferrite becomes unstable and is transformed to
austenite. At normal temperature, usually only about 1/10 of its content
remains preserved during solidification. Due to this fact the suitability of
the conventional Schaeffler´s diagram for assessing the cracking susceptibility
becomes questionable.
By projection of solidification modes into the Schaeffler diagram
(Fig. 5), the zone of austenitic and austenitic-ferritic steels can be divided
into characteristic bands with the aid of a, b, c lines [5]. For instance, the
band limited by a-b lines corresponds to solidification in the pseudobinary
diagram between points A-B. It can be seen that also purely austenitic
steels can solidify though δ-ferrite at normal temperature (hatched zone).
Knowledge of high-temperature transformations of δ-ferrite is significant
from the viewpoint of liquation cracking. If the underbead zone of the
heat affected zone in austenitic steel reaches again the ferrite zone, the
cracking susceptibility is decreased.
Residues of the so-called high-temperature δ-ferrite can then be observed
in the underbead zone, because at quick cooling-down it cannot be
retransferred completely (see Fig. 6 with 13 % of δ-ferrite, whereas only
0.35 % was present in the base metal).
Fig. 4. Pseudobinary Cr/Ni diagram for 69.7 % Fe
Contribution to Liquation Cracking of Austenitic Stainless Steels 89
Fig. 5. Schaeffler diagram plotting the solidification zones according to Fig. 4
Fig. 6. High-temperature 􀄯-ferrite in the underbead zone of 17 352 steel
The favorable effect of δ-ferrite on the elimination of hot cracks is explained
by several facts:
a) Higher solubility of impurities such as S, P and harmful elements
(e.g. Si) compared to that observed in austenite;
90 Phenomena and Mechanisms
b) Enlargement of the total grain boundary area;
c) Grain size refinement due to which, similarly as in b), the covering
of grain boundaries with molten phase is reduced;
d) Lowering of contraction stresses because ferrite has a lower coefficient
of thermal expansion;
e) Enlargement of the dihedral angle 􀄬, i.e. the angle of grain boundary
contact because the interphase energy σδγ < σγγ (σss is the interphase
energy between the solid-solid phase, σsl between the solidliquid
phase), [6].
2 1
2cos
s
ss
σ
σ
Θ = (1)
Even though the formation of liquation cracks is conditioned by the coaction
of metallurgical, structural and technological factors, the primary
factor is metallurgical, i.e. the proper susceptibility of steel.
Some metallurgical reasons such as impurities in steel (S, P…) are valid
in general (i.e. the case of extrinsic cracks). Other, the so-called intrinsic
cracks are caused by alloying elements (e.g. Nb, Ti, C, B …). Except for
the fundamental chemical composition of the steel, its manufacturing
process and deoxidation method etc. also affect its susceptibility to liquation
cracking. The enumeration of all possible metallurgical causes for liquation
cracking is as a fact impossible and each steel requires an individual
approach.
Fig. 7. Eutecticum Fe􀈖-M23C6 on the liquation fractured surface in 17 255 steel
Contribution to Liquation Cracking of Austenitic Stainless Steels 91
Only some of elements whose effect is already considered as more or
less generally known will be mentioned below.
Carbon acts as austenitizing element which decreases the content of δ-
ferrite in steel. For the creep-resistant steel type 17 255 (0.165 % C), the
connection between liquation cracking and carbide eutecticum formation
Fe γ - M23C6 (Fig. 7) was revealed. The solidus temperature of such eutecticum
is decreased down to 1300–1250 °C.
S and P impurities are, in general, considered to be unfavorable.
Therefore, some authors [10]require P < 0.01 % and S < 0.005 % in order
to eliminate the cracks in CrNi-steel type 25/20. Phosphorus is considered
to be more unfavorable than sulphur because sulphur can be eliminated by
manganese. Manganese binds sulphur to sulphides of α-MnS-type which
have higher melting points. These sulphides can substitute chromium up to
the chemical composition of Mn26Cr49. Chromium decreases the melting
point of these sulphides. A higher Mn-content in the steel is favorable because
the Mn/Cr-ratio in sulphides is increased. Phosphorus forms lowmelting
point phosphide of M3P-type with its eutectic temperature of 1060-
1100 °C. Higher P-contents can be eliminated only by δ-ferrite which has
an expressively higher solubility than austenite, e.g., for a temperature of
1365 °C: 1 %P in Feδ in comparison to 0.1 % P in Feγ [7].
Silicon is also in general considered to be undesirable because it forms a
silicate eutecticum. Its content in weld metal is limited to 0.5 % and from
the viewpoint of liquation cracking even a lower content of max. 0.3 % is
recommended.
A low content of titanium (up to 0.1 %) is considered as favorable, since
it increases the eutectic temperature of the phosphide eutecticum
(∼1090 °C). However, the stabilizing contents of Ti and Nb are unfavorable
because they form eutectic reactions between TiX and/or NbX and Feγ
(Ts = 1310–13500C). Moreover, titanium forms hexagonal sulphide τ -
Ti2S or Ti4S2C2 with a melting point of ∼1350 °C, which can also be the
reason for liquation cracking. If the stabilizing ratio is exceeded, the niobium
can also form the Laves phase Fe2Nb with a lower melting point.
From the viewpoint of liquation cracking, the excessive amount of eutecticum
can, on the contrary, have a favorable effect due to crack self-healing.
A similar case was encountered also with an experimental melt of
03Cr19Ni11B steels with a higher boron content (∼1 %). At a temperature
of 1050 °C the excess of eutectic phase (Fe, Cr)2 B - Feγ (Fig. 8) was
formed which assures liquation cracking resistance, but this steel has low
toughness at room temperature. The formation of low-melting Ni3S2 is not
likely in case of sufficient Mn-content in the steel.
92 Phenomena and Mechanisms
Fig. 8. Eutectic phase (Fe,Cr)2B-Fe􀈖 on the liquation fracture surface in
03Cr19Ni11B steel
Only the effect of several common elements in austenitic Cr-Ni steels
was mentioned, but as has already been pointed out, the problem of liquation
cracking is often more complex.
Test Material and Experiments
Table 1. shows the used experimental steels. All test materials are commonly
used steels made in the Czech republic. The steels designated as A,
B and D are titanium-stabilized. Low-carbon steels represent another experimental
group, among which G, I, J and I1 are nitrogen-hardened and B,
C and E are the strain-hardened ones. The steel designated as I has a higher
manganese content (2.7 %). The steels designated as N (02Cr 20Ni 22Mo
3Nb), O (02Cr 22Ni 20) and P (05Cr 21Ni 33Al 0,3Ti) serve for demanding
corrosive atmospheres. The steel designated as X is an experimental
melt of austenitic manganese steel and the steel Y is a martensitic chromium
steel, the only non-austenitic experimental steel. The effect of refining
processes, electroslag (ESR) and vacuum remelting (VR) was verified
for the steels Cl (17 249) and K (17 349). The sign of EOP means remolten
in an electric arc furnace.
To verify the resistance of steel to liquation cracking, two tests were
employed, namely the test with simulation of welding cycles and the VÚZLTP-
1-6/TIG test. Both are laboratory tests. The shop tests did not prove
to be reliable for the evaluation of liquation cracking.
Table 1. Chemical composition (wt%) of experimental steels
Des. Steel Melt h
[mm]
Asdelivered
Condition
C Mn Si P S Cr Ni Mo Ti Nb N Cu Al
A 17246 20 r.z. 0.068 1.6 0.46 0.032 0.007 17.7 11.1 0.97 0.08
B 17247 E0094 20 d.s. 0.06 1.32 0.75 0.025 0.014 18.31 10.16 0.43 0.09
C 17249 E5621 20 d.s. 0.03 1.25 0.31 0.032 0.018 18.07 11.74 0.05
D 17347 16 r.z. 0.098 0.79 0.45 0.034 0.019 16.22 11.2 1.62 0.48 0.13
E 17350 E7699 20 d.s. 0.03 0.66 0.31 0.037 0.021 17.59 15.0 2.60 0.13
F 17353 9316 15 r.z. 0.067 1.15 0.54 0.031 0.016 16.77 12.84 2.29 0.11
G 17359 E8916 25 r.z. 0.02 0.90 0.30 0.017 0.020 16.96 12.0 2.30 0.26 0.12
I 17259 E7341 26 r.z. 0.026 2.70 0.08 0.023 0.020 18.28 12.0 0.23
J 17360 E8882 26 r.z. 0.02 1.40 0.75 0.016 0.016 18.5 12.98 2.80 0.25 0.09
N AKOR 0-893 5
r.z. 0.02 0.43 8.5 0.015 0.01 20.25 22.7 3.50 0.8
O AKC2 0-688 5 r.z. 0.02 1.0 0.68 0.016 0.01 22.80 19.45 0.1
P AKR17 0-605 3 r.z. 0.05 1.2 0.6 0.021 0.015 21.0 32.80 0.5
X 17483 2679 50 r.z. 0.025 20.23 0.47 0.016 0.015 13.68 0.5 0.34 0.014
Y 17199 E6222 100 z. 0.2 0.7 0.09 0.014 0.006 10.45 0.15 0.94 0.08 0.07 V0.18 0.05
E1 17350 2309E 20 r.z. 0.02 1.21 0.36 0.033 0.016 17.2 13.52 2.78 0.05
E2 17350 – 20 r.z 0.024 1.29 0.33 0.024 0.022 17.94 14.43 2.90 0.03 0.012 0.09
H1 17352 95098E4 45 x 45 k. 0.05 1.0 0.48 0.027 0.018 17.8 12.42 3.12 0.32 0.042 0.17
H2 17352 95019E1 45 x 45 k. 0.05 1.31 0.4 0.032 0.018 16.54 12.23 2.55 0.033 0.034 0.15
H3 17352 94631E1 90 x 90 k 0.06 0.97 0.41 0.028 0.018 17.0 11.70 2.06 0.035 0.037 0.09
I1 17259 E7162 25 r.z. 0.034 1.26 0.26 0.020 0.012 17.45 10.35 0.16
C2 17249 – 14 r.z. 0.03 1.00 0.39 0.019 0.011 19.29 11.5 0.049
C1 17249 E9597 15 r.z. 0.03 1.24 0.38 0.020 0.014 17.32 11.6 0.05 0.048 0.05
C1O 17249 EOP ∅20 v.t. 0.03 1.35 0.46 0.014 0.014 18.12 10.95 0.04 0.034 0.13
C1E 17249 ETP ∅20 v.t. 0.05 1.29 0.32 0.017 0.010 17.36 11.35 0.040 0.07
C1V 17249 VOP ∅20 v.t. 0.03 1.20 0.50 0.016 0.013 17.85 11.45 0.036 0.10
K 17349 15 r.z. 0.02 1.10 0.26 0.022 0.015 16.30 13.57 2.30 0.040 0.05
KO 17349 EOP ∅20 v.t. 0.033 1.14 0.37 0.021 0.018 17.34 12.45 2.38 0.025 0.08
KE 17349 ETP ∅20 v.t. 0.055 1.10 0.027 0.023 0.010 16.60 12.41 2.10 0.023 0.08
KV 17049 VOP ∅20 v.t. 0.03 0.97 0.04 0.021 0.012 18.12 12.81 2.13 0.018 0.09
K1 17349 14 r.z. 0.02 0.88 0.21 0.021 0.018 18.15 13.8 2.32 0.044
Des. Designation, r.z. solution annealing, d.s. strain hardening, z quenched and tempering, k
forged, v.t. hot rolling, h thickness.
Contribution to Liquation Cracking of Austenitic Stainless Steels 93
94 Phenomena and Mechanisms
Fig. 9. Record of a hot cracking programme
The tests with simulation of welding cycles were performed on the
Japanese Thermorestor-W equipment. The simulation of underbead thermal
cycles with a cooling time Δt8/5 = 12 s was selected. In the course of
the cooling phase of the cycle the test bar 7 mm in diameter is subjected to
1–4 mm tensile strain at a strain rate of 1 mms-1 (Fig. 9).
Fig. 10. Dangerous zone of liquation crack initiation in AKC-2, AKRI-7 and
AKOR-2 steels evaluated by the welding cycle simulation test
Contribution to Liquation Cracking of Austenitic Stainless Steels 95
The test simulates a tack welded joint in which the stitch is suddenly
loosened after passing of the following run. The presence of cracks after
cooling down represents boundaries of the so-called dangerous zones for
liquation cracking (see an example in Fig. 10).
The calculated value of tg αK (mm/100 °C) is the test criterion, where
αK is the angle between the directive plotted from the solidus temperature
(of assessed DTA) to the boundary of the liquation crack zone and the abscissa
(see an example in Fig. 11). The susceptibility to cracking is inversely
proportional to the tg αK-value.
Fig. 11. An example of liquation cracking test results of four steels employing the
tg 􀄮K (mm/100 °C) criterion
In the VUZ-LTP-1-6/TIG test which is of Russian origin, specimens
(Fig. 12) remelted by TIG-welding within the brittle temperature range
(BTR) are subjected to tensile strain at different strain rates.
The maximum strain rate Vd (mms-1) at which no hot cracks are formed
in four repeated specimens is the criterion (see example in Fig. 13). The
blackened part of the circles is proportional to the hot crack size. The
higher the value Vd, the higher the resistance against hot crack formation.
Both tests are specified in the STN 05 1143 standard.
In the course of the investigation, the preliminary criteria of both tests
for evaluating high-alloy steels were determined (Table 2).
96 Phenomena and Mechanisms
Fig. 12. Sketch of the VÚZ-LTP-1-6 test specimen
Fig. 13. Result of the VÚZ-LTP-1-6/TIG test for AKRI-7 steel (Vd = 0.01 mms-1)
Contribution to Liquation Cracking of Austenitic Stainless Steels 97
Table 2. Preliminary criteria of hot cracking tests by simulation of welding cycles
(tg αK) and VÚZ-LTP-1-6/TIG (Vd)
tg αK
[mm/100 °C]
Vd
[mms-1]
Evaluation of steel
> 1.2 > 0.1 1st degree: steel resistant to hot crack
initiation without special conditions
during welding
0.7–1.2 0.033–0.01 2nd degree: steel partially susceptible
steel to hot crack initiation which
can, however, be prevented with the
use of specified conditions during
welding
< 0.7 <0.033 3rd degree: steel susceptible to hot
crack formation
Calculation of Hot Cracking Parameters
The calculable parameters L, ΔH [8] and ECr/ENi [9] for the expression of
the cracking susceptibility of austenitic steels are the most known from literature.
L = - 299% C + 8% Ni + 142% Nb – 5.5% δ2 – 105, (2)
if L > 0, the steel is susceptible to cracking.
ΔH = -700 + 17% Cr – 37% Ni – 117% Nb + 29% Mo + 188, (3)
if ΔH > 100, the steel is crack-resistant.
The parameter L expresses a crack length in the TIG-remelting test. The
parameter ΔH was proposed based on the hot ductility test and is only
valid for HAZ-cracking. The parameters are applicable to both stabilized
and non-stabilized steels.
ECr = %Cr + 1.37% Mo + 1.5% Si + 2% Nb + 3% Ti, (4)
ENi = %Ni + 0.31%Mn + 22%C + 14.2%N + Cu, (5)
if ECr/ENi <> 0.02 %, the steel is susceptible to cracking
if ECr/ENi 􀂕 1.5 and P + S 􀂔 0.02 – 0.05, the steel is crack-resistant.
The term ECr/ENi contains newly proposed chromium and nickel equivalents
which better express the chemical composition of the steel than the
known equivalents from the Schaeffler diagram. The ratio of ECr/ENi as a
fact considers the mode of primary solidification. If ECr/ENi < 1.5, the steel
98 Phenomena and Mechanisms
solidifies in the mode of austenite or austenite-ferrite. If ECr/ENi 􀂕 1.5, the
ferritic-austenitic or ferritic mode is concerned which indicates a resistance
of the steel to cracking at conventional levels of impurities.
Before using the Thermorestor-W and VÚZ-LPT-1-6 test equipment,
liquation cracking tests were carried out on the Russian IMET-1 equipment
which was adapted for operation in an inert atmosphere. The impact
tensile test (force induced by an electromagnet) was performed in the
course of the simulated thermal cycle of the overheated HAZ (controlled
by resistance heating and cooling gas). The nil ductility temperature designated
as Td was evaluated in the test as a lower temperature of the brittle
temperature range (BTR), see example of the evaluation in Fig. 14 (results
of the IMET-1 test are not a subject of this paper).
Fig. 14. IMET-1 test result of 03Cr19Ni11B steel
The solidus temperature assessed by DTA was considered as the upper
BTR temperature for the case of liquation cracking. However, it has been
found out that the dependence between BTR width and liquation cracking
susceptibility is not generally valid. In CrNi-steel with higher boron
addition (~1 %), the assessed Td is very low (Fig. 14) because at 1050 °C,
a high amount of boride eutecticum (Fe, Cr)2 B - Fe􀈖 is formed as it can be
seen at the fractured surface (Fig. 8). However, this material did not exhibit
any cracking in real joints because a high amount of the liquid phase
itself overflowed potential cracks (self-healing effect). As the IMET-1 test
always results in fracture, this effect could not be considered. This was one
of the reasons why we have subsequently ceased the IMET-1 test.
More objective results can be achieved on the Japan Thermorestor-W
simulator where the test specimens are loaded imposing 1 up to 4 mm
Contribution to Liquation Cracking of Austenitic Stainless Steels 99
additional strain at a strain rate of 1 mms-1 which more closely resembles
the real condition.
The advantage of the Thermorestor-W equipment is to be seen in the
fact that the tests are carried out in the shielding atmosphere of the vacuum
chamber, so that the high-temperature fractured surfaces remain free from
oxides and are therefore suitable for further metallurgical analyses to identify
the causes of the cracking susceptibility by SEM or TEM replicas as
shown in the figures of this paper.
Fig. 15. Eutectic carbon nitride on the liquation fracture surface (17 347 - D steel)
Test Results
The results of liquation cracking tests obtained by simulation of welding
cycles (tg αK values) and VÚZ-LPT-1-6/TIG tests (Vd values) are summarized
in Table 3. Except for a certain scatter, both methods are comparable.
The complex evaluation of the test series proves that the conventional
types of CrNi- and CrNiMo-steels (in Table 1 designated as class 17) can
be classified by the 2nd and partially 1st degree of cracking resistance.
The worst results were found for AKOR 2 and AKRI 7 steels which
have an unfavorable Cr/Ni < 1.0 ratio. Their primary solidification mode is
austenitic representing the 3rd degree of cracking susceptibility in the
VÚZ-LPT-1-6/TIG-remelting test. In the case of AKOR 2 steel, the liquation
fracture surface was covered by eutectic particles of niobium carbonitrides
(NbX).
100 Phenomena and Mechanisms
Regardless of their purity, these steels are constitutionally susceptible to
hot crack formation. Sporadic occurrence of microcracks, mainly in the
remolten steel, cannot always be excluded completely and these cracks
have to be evaluated by the admissibility criteria.
The melt of 17 347 steel (designated as D), which had a relatively high
carbon content (0.098 %), was evaluated by the simulation test to be at the
boundary of the 3rd degree. Frequent presence of eutectic constituents of
coarser (TiX) and thinner (τ - Ti4S2C2) morphology was observed on the
surface of liquation cracks, as shown in Fig. 15 and Fig. 16.
Fig. 16. Eutectic carbon sulphides Ti4S2C2 on the liquation fracture surface
of 17 347 steel
It is supposed that the decreased cracking resistance of this melt (D) is
caused by the formation of a low melting point eutecticum consisting of
TiX and Ti4S2C2 phases in the underbead zone. The tests have proved that
the majority of conventional austenitic steels classified by the 1st degree
(crack-resistant) contain a low amount of δ-ferrite (0.3–3.5 %) and/or a
lower amount of sulphur (∼ 0.01%).
The calculated parameters are summarized in Table 3. The most comparable
results from the verified calculation indices were attained for the
parameter L including also δ-ferrite.
Contribution to Liquation Cracking of Austenitic Stainless Steels 101
Table 3. Results of liquation cracking tests, calculated parameters and 􀄯-ferrite
measurements
Des. Steel Melt tg αk Vd
[mms-1]
L ΔH Ecr/
ENi
P+S
[%]
δmet
[%]
A 17246 1.5 0.15 -63.8 30 1.61 0.039 8.5
B 17247 E0094 1.25 0.15 -62.1 65 1.66 0.039 3.3
C 17249 E5621 2.0 0.13 -2.6 48 1.48 0.050 0.3
D 17347 0.7 0.06 -53.4 27 1.49 0.053 3.5
E 17350 E7699 1.18 0.08 23.9 13 1.35 0.058 0
F 17353 9316 0.95 0.07 8.4 16 1.39 0.047 1.3
G 17359 E8916 1.33 0.12 -2.0 85 1.24 0.037 0
I 17259 E7341 1.00 0.15 -1.23 36 1.16 0.04 0
J 17360 E8882 1.05 0.15 4.84 87 1.34 0.032 0
N AKOR2 0-893 0.59 0.01 188.0 -27 1.2 0.025 0
O AKC2 0-688 1.10 0.02 56.0 -158 1.16 0.026 0
P AKR17 0-605 0.83 0.01 172.2 -703 0.67 0.036 0
X 17483 2679 1.66 – -1892 384 1.91 0.036 31.1
Y 17199 E6222 0.91 – -34.0 293 2.40 0.020 –
E1 17350 2309E 1.17 0.07 -14.9 55 1.52 0.049 1.5
E2 17350 – 1.43 – 21.9 22 1.44 0.046 0.3
H1 17352 95098E4 0.74 – 13.4 82 1.56 0.045 0.22
H2 17352 95019E1 0.72 – 11.9 52 1.46 0.050 0.25
H3 17352 94631E1 0.91 0.06 10.8 58 1.50 0.046 0.35
I1 17259 E7162 0.83 – -12.0 78 1.29 0.032 0
C2 17249 – 1.66 – -5.4 69 1.52 0.030 0.50
C1 17249 E9597 1.17 – -3.2 33 1.38 0.034 0
C1O 17249 EOP 1.15 – -8.9 71 1.50 0.028 0.30
C1E 17249 ETP 2.25 – -4.7 28 1.32 0.027 1.00
C1V 17249 VOP 1.17 – -4.6 46 1.43 0.029 0.20
K 17349 1.00 – 9.5 15 1.32 0.037 0
KO 17349 EOP 0.95 – 3.9 68 1.52 0.039 0.30
KE 17349 ETP 3.33 – 2.8 33 1.34 0.033 1.20
KV 17049 VOP 1.05 – 5.9 62 1.54 0.033 0.30
K1 17349 0.87 – 11.30 39 1.42 0.039 0
Des. Designation.
The parameter ΔH is the most stringent in comparison to others, no steel
type was evaluated as resistant. The ratio of ECr/ENi expressing the mode of
primary solidification seems to be significant, too.
The assessed scatter of results (e.g. steels D) follows from the fact that
also other physical-metallurgical factors are involved which can at present
hardly be defined in the parametric equations. With the help of a regression
analysis of the welding cycle simulation test result, a parametric
102 Phenomena and Mechanisms
equation for tgαK in dependence on the chemical composition was
proposed:
tgαK = 4.166 + 361.30C – 4540.7C (P + S)
-117.27C (Cr/Ni+ 90.6 (P + S)Cr/Ni).
(6)
The equation is valid for chemical compositions of the austenitic steel
within the range as follows:
C = 0.02 – 0.07 %, Mn = 0.8 – 1.4 %, P + S = 0.025 – 0.050 % and
Cr/Ni = 1.12 – 1.72
The criteria of the parameter tgαK are the same as those given in Table2.
Effect of Welding Parameters on Liquation Cracking
The austenitic steels are especially susceptible to grain coarsening in the
overheated heat affected zone because they are monophase steels with a
low thermal conductivity coefficient. The grain size in the heat affected
zone depends on the size of the initial grain, on microstructural obstacles
slowing-down the grain boundary migration (δ-ferrite, segregation of
carbides, etc.), on the weld heat input, but also on the geometric factors
(thickness). The size of the initial grain is determined by the steel making
procedure. The presence of a carbide phase is mostly inadmissible from
the viewpoint of the required properties (corrosion resistance).
In general, lower weld heat input is preferred to restrict the grain size,
and thus to eliminate liquation cracking. Except for grain size reduction,
also a narrower heat affected zone, lower segregation degree and a lower
amount of the total molten phase are attained. Apart from a decrease in
heat input, it is also recommended to reduce the welding speed v. Lower
welding speed means lower thermal gradient of strain. In the Russian literature,
the nominal heat input is denoted as:
v
I E
v
q = ⋅ ,
(7)
i.e. the welding speed is considered separately; this is more practical
from the viewpoint of hot cracking sensitivity evaluation. The criterion of
the critical welding speed, i.e. the maximum speed which the welded joint
withstands without crack initiation (while preserving the bead crosssection)
is included in the Russian standard.
Preheating at low temperature, which several times proved to be suitable
to provide protection against solidification cracking due to lower thermal
stresses during cooling, has less effects on liquation cracking. Deposition
of several smaller beads or employing a pulse process is considered to be
Contribution to Liquation Cracking of Austenitic Stainless Steels 103
convenient for lowering the total stresses. An other possibility of eliminating
liquation cracking in the base metal is to use weld metal with a lower
melting point, because, if the weld metal solidifies later, the tensile stresses
cannot be transferred into the base metal.
Conclusions
Preliminary criteria of Thermorestor-W and VÚZ-LTP-1-6 tests (3 degrees)
are proposed. Experimental results have been compared with the
calculation indices (L, ΔH and ECr/ENi). An own parametric equation tgαK
based on the Thermorestor-W test are also proposed.
The majority of the examined austenitic stainless steels attained the 2nd
degree of cracking resistance, and partially the 1st degree, i.e. the liquation
cracking susceptibility would not restrict their weldability. The 2nd evaluation
degree means only a partial cracking susceptibility which can usually
be eliminated by specified welding conditions.
A resistance to liquation cracking (1st degree) of conventional austenitic
steels was found at low contents of δ-ferrite (0.3–3.5 %) and at a low content
of S (∼0.01 %). The austenitic steels with the unfavorable ratio of
Cr/Ni < 1 are susceptible to cracking (3rd degree).
The validity of calculation indices for the hot cracking susceptibility as
well as the preliminary test criteria have to be further verified by results
obtained from real joints.
References
1. Jakušin BF, Želev AN, Machnenko VI (1983) 2. simpozium SEV “Primenenie
matemati􀃾eskich metodov pri izu􀃾eniji svarijemosti”. Sozopol, Bulgaria
2. Tamura H, Watanabe T (1973) Trans of the Japan Welding Society, 2: 3–10
3. Medovar BJ (1964) Avtomati􀃾eskaja svarka, 6: 1–13
4. Perteneder E, Rabeusteiner G, Schabereiter H, Tösch I (1979) Schweißtechnik,
No 3
5. Matsuda F, Nakagawa S (1984) IIW Doc IX-1315-1984
6. Smith CS (1964) Metal Review, 9: 33
7. Kohira J (1973) Yosetsu Gijutsu, No 5: 41–47
8. Morishige N, Kuribayashi M, Okabayashi H (1979) IIW Doc IX-1114-1979
9. Kujanpää VP, Moisio T (1980) Conference Solid techn in the foundry and
cast house. University of Warwick, Couventury
10. Arata Y, Matsuda F, Nakagawa H, Katayama S (1978) Transaction of Japan
Welding Research Institute, No 2
Morphology of Hot Cracks
in Single-Phase Weld Metal
B. Yakhushin
Baumann University, Moscow, Russia
Abstract
Solidification and liquation cracking, representing different types of hot
cracking, have been investigated for austenitic stainless steels. Methods for
the experimental evaluation and for the prediction of solidification and
liquation cracking during single- and multi-layer arc welding are suggested.
Morphology of Solidification Cracks
Austenitic stainless steels which are often applied in safety-relevant constructions
or components often exhibit a low hot cracking resistance during
welding. Multiple thermal cycles affecting especially the grain boundaries
of the primary weld metal are typical of multi-layer arc welding techniques
and might even intensify the probability of hot crack occurrence.
Therefore investigations have been carried out concerning the testing as
well as the prediction of hot cracking for single- and especially for multilayer
welding. The filler material chosen for the investigations was
X5CrNiAMo 18 09 featuring a single-phase austenite microstructure up to
the melting temperature. For best possible reproducibility of hot cracking
during testing, the specimens were subjected to defined external loading
during welding.
For hot cracking experiments concerning the root weld, pairs of specimens
of 3 mm thickness were used which were welded end-to-end under
simultaneous straining longitudinally to the weld direction [1]. The welding
speed was varied over a wide range from 3 to 67 m/h. The welding
current was adapted to the welding speed to ensure equal penetration depth
of each weld. For all variations of welding parameters the detected hot
Morphology of Hot Cracks in Single-Phase Weld Metal 105
cracks were directed longitudinal to the welding direction, which is especially
characteristic of hot cracks in root welds. In order to minimize undefined
thermomechanical loadings during overwelding the specimen’s
edges, slanting run-on and run-off tabs were applied (Fig. 1).
Fig. 1. (􀉚) Outline of the specimen with a slanting run-on tab applied in the investigations
and (b) a photograph of the topography of cracks obtained in tests
The straining of the specimens was triggered when the electrode reached
the top of the run-on tab. By this special timing of the test procedure potential
hot cracking could be limited to the middle part of the specimen.
The period of post-weld straining was calculated as the time it took the
specimen to reach a temperature between 700 °C and 800 °C, representing
the lowest possible temperature of the brittle state.
Counterbalancing the opposite directed influences of arc current and
welding speed on the heat input, single-layer welds of identical height but
of various relative crystallite orientations in the plane of welding (Fig. 2)
were produced. For obtaining a functional dependency, the general criterion
R was applied:
R = q⋅vw. (1)
The variable q physically represents the heat input per second and vw is
the weld travel speed. The usually applied criterion q/vw varies only insignificantly
when different welding speeds are counterbalanced by other
welding parameters, especially if the cross-sectional area of the weld remains
unchanged, but the microstructure and particularly the resistance
against hot cracking vary greatly.
106 Phenomena and Mechanisms
Fig. 2. Changes of the crystallite orientation «θ/2» in the weld metal center and
resistance against hot cracking BM for the q⋅vw criterion depending on the welding
speed
Fig. 2 shows the influence of welding speed on the crystallization orientation
θ/2 and on the critical deformation BM per °C. Angle θ/2 represents
the direction of the solidification front measured at the surface on the weld
metal centerline.
The factor BM is calculated from the relation between the critical deformation
speed vcr and the cooling speed of the weld metal (vbtr) within the
brittle temperature range (BTR), according to the following equation:
BM = vcr/vbtr. (2)
Considering a constant heat input q, it can be seen by Fig. 2 that as the
welding speed increases, the angle of crystallite orientation passes a minimum
at a position where the critical strain rate BM has its maximum.
The values of R corresponding to the maximum peak of BM are nearly
identical for various steels and range at welding speeds of vw = 6–12 m/h.
Such an optimal constellation of welding speed and arc current is named
R0 and means that crystals with lateral facets occur in the center of a weld.
A slowdown of the welding speed, however, causes an unfavorable weld
metal microstructure, since the crystals in the centerline concur under an
obtuse angle. The corresponding decrease of hot cracking resistance can be
measured by a lower critical strain rate BM per °C. For values of R > R0,
Morphology of Hot Cracks in Single-Phase Weld Metal 107
crystals with front facets concur in the centerline of the weld leading also
to a measurable reduction of the critical strain rate.
For various steels the critical strain rate BM can be calculated from the
specific R values by the following equation:
BM = B0 − k ⋅ lg(R R0), (3)
whereby 􀈼0 represents the value of BM in the optimum mode from
which the slope function B(R) for R > R0 is subtracted. This means that the
critical strain rate BM is composed of a material-specific summand B0 and
a welding-specific summand represented by BW = k ⋅ lg (R/R0).
For the calculation of the material-specific summand B0, an equation
based on statistical results of numerous hot cracking experiments was developed:
[ ( ) ( )]7
B0 = − 3.779 + 2.331⋅ E Cr Ni + 0.4961⋅ FN − 47.33⋅ P + S ⋅10− , (4)
whereby E(Cr/Ni) represents the chromium-nickel ratio, FN is the ferrite
number and P+S represents the cumulated content of sulfur and phosphorus
in the material. It follows from Eq. 4, that B0 most significantly depends
on the content of detrimental impurities but also on FN and E
(Cr/Ni).
Regarding the technological and welding-related contribution to hot
cracking, the value of BW is significantly influenced by the size of the weld
pool and can be calculated according to the following equation for a typical
austenitic stainless steel:
BW = 3.04 − 0.04L + 0.19B − 0.23b , (5)
whereby L is the length of the weld pool, B is the width of the weld pool
at the top side and b is the width of the weld pool at the bottom side.
The lengths of the weld pools L could be determined after the instant
removal of the melt during welding by percussive impacts. According to
the results of a statistically analyzable number of experiments, the following
equation as an approximation for L was developed for steel
X5CrNiTi 18 10:
L = 4.28 + 0.14IW + 0.5VW +16d . (6)
In the Eqs. (5) and (6) the fact is taken into account that the most significant
parameters influencing BW are the length of the weld pool and the
width of the weld pool at the bottom side of the root weld. Under the condition
of b converging to 0, a flat crystallization pattern changes to a threedimensional
pattern resulting in a reduction of unfavorable angles between
crystallites and in a significant increase of the hot cracking resistance.
108 Phenomena and Mechanisms
Fig. 3. Delineation of thermal interactions and the probable position of liquation
cracks after three passes welded with a single-phase austenitic filler material
Morphology of Liquation Cracks
Liquation cracking has especially to be considered during multi-layer filler
welding [3]. Liquation cracks usually appear as fine fissures in the previous
layers of a weld and they generally do not cross the fusion line of two
weld layers (Fig. 3). The histograms (Fig. 4) of 793 liquation cracks show
that the liquation cracks are for the most part smaller than 0.5 mm.
They form up under the thermomechanical influence of the subsequent
layer of the weld. Liquation cracks can run along migrated grain
boundaries (Fig. 5).Liquation cracking as described above is assumed to
result from high-temperature creep during temperature-activated relaxation
of residual stresses accumulated by the previous weld pass(es).
A technological-based test procedure was developed reproducing such
cracks under real welding conditions (Fig. 6). The operation principle is
that the first layer of the weld is produced without any external deformation
whereas during welding of the second layer the specimen with both
layers is deformed by external loading. To exclude hot crack formation in
the top layer, filler materials with increased technological crack resistance
were applied [4].
Morphology of Hot Cracks in Single-Phase Weld Metal 109
Fig. 4. Histograms of (􀉚) liquation crack lengths and (b) frequency of cracks not
crossing the fusion line; total number of analyzed hot cracks in multi-layer welds
(above 100 passes): 793
110 Phenomena and Mechanisms
Fig. 5. A liquation crack in solid state crossing a crystallite, formed at the last
stage of migration in the new position of grain boundary
Fig. 6. Delineation of a test procedure for liquation cracking by repeated arc heating.
Specimen is bent during welding of the second layer, so that the first layer
represents the test weld and the second layer represents the repeated arc heating
If liquation cracking in the bottom layer of the weld is initiated, a critical
deformation rate B2 can be defined representing the hot cracking resistance
of the weld metal in solid state (Fig. 7). Highest values of B2, which
means the hot cracking susceptibility is lowest, can be found in welds with
5–7 % δ-ferrite, which conforms to observations in practice.
In order to analyze the temperature range in which liquation cracks are
formed, welded specimens were subjected to simulated weld cycle reheating
and simultaneously exposed to different stresses (Fig. 7).
Morphology of Hot Cracks in Single-Phase Weld Metal 111
Fig. 7. Resistibility of high-alloy steels to crystallization and liquation cracking
The effects of high-temperature creep and of the temperatures at which
subsequent brittle failures occurred are inversely proportional to the
applied stresses.
By measuring the necking (reduction of the cross sectional area) of the
specimen during heating, the brittle temperature range (BRT) was found at
around 900–1100 °􀉋 (Fig. 8), i.e. in the liquation range (BRT2). When the
stress loading of the specimen was further reduced, brittle failures occurred
in the crystallization-liquation temperature range (BTR1) resulting from
grain boundary melting.
112 Phenomena and Mechanisms
Fig. 8. Test device for the measurement of ductility and plasticity, respectively, of
welded specimens (2) during simulated thermal weld cycle under external loading,
induced by the machine parts (8)(7)(6)(5)(4)(1). Part (3) represents the sensor for
the measurement of the specimen necking
An increase of the deformation speed within BTR2 did not influence the
brittle failure. For specimens containing δ-ferrite, brittle failure occurred
only within BTR1. Specimens exhibiting a recrystallized base metal according
to the tested grades also showed no brittle failure related to BTR2.
The results confirm the above-stated assumption that the basic reason
for the existence of BTR2 is the non-equilibrium (weld metal) microstructure.
The weld metal microstructure is supersaturated with vacancies
which grow by deformation and which provide high speed diffusion processes
during repeated arc heating.
The second reason for the existence of BRT2 is the shrinkage of the
metal during cooling at speeds causing high-temperature creep. Such kind
of deformation is determined by partial dislocations merging and creeping
onto grain boundaries together with impurities which have a lower diffusion
mobility in comparison to the elements of the base material. Segregations
of this kind on grain boundaries cause grains to slide, leading to brittle
failure of a solid-phase weld. By a comparison of the research results it
was found that alloys exhibiting a BTR2 are characterized by a high level
of the stacking fault energy, γ, responsible for their significant tendency to
high-temperature creep. The stacking fault energy is calculated by the
equation:
γ = −67 + 4.75%Cr + 2%Ni + 0.5%Mn − 43.3%N − 5%Mo (7)
It was furthermore found that the critical deformation rate B2 within
BTR2 is proportional to stacking fault energy [5]. The stacking fault
Morphology of Hot Cracks in Single-Phase Weld Metal 113
energy is recommended to be considered as a general function of the
chemical composition of the weld metal.
The second factor which determines the formation of hot cracks within
the BTR2 is the thermal cycle during the repeated arc welding. The period
of the heating and period of the BTR2-state can be taken as the criterion
for the deformation rate B2. By adding a composition-specific term and a
term considering the characteristic of the thermal cycle with linear dependencies,
a basic dependence of B2 on the chemical composition and the
mode of welding was developed:
􀂸 􀂸 􀂹
􀂷
􀂨 􀂨 􀂩
􀂧 −
+ 􀂸 􀂸􀂹
􀂷
􀂨 􀂨 􀂩
􀂧
=
0
0
2
1
t
t t
B n m i
γ
,
(8)
whereby γ is the stacking fault energy, t0 and ti describe the time within
BTR2 during welding in the optimum and in the investigated modes, respectively.
m and n are factors of proportionality.
From Eq. 8 it follows that for increasing 􀈼2 within BTR2, it is necessary
to alloy a filler material with elements, as for instance nitrogen, molybdenum
lowering the stacking fault energy and also increasing the cooling
rate, thus reducing the period within crack-critical high-temperature states.
An increase of the welding speed should also be avoided, since this might
cause hot cracking within BTR1.
It is quite effective to apply low welding speeds (10–20 m/h) and to intensify
the cooling of the weld metal by internal and/or external heat
drainage and by electromagnetic agitation of the weld pool. It was also observed
that alloying the weld metal with certain elements, for example
with titanium, leads to contrary results of BTR1 and BTR2. Within BTR1,
titanium leads to grain refinement, and within BTR2, it is an element, that
increases the tendency to segregation. These results emphasize the need to
determine Bcr separately for BTR1 and BTR2.
Furthermore, the prevention of solidification and liquation hot cracking
can be achieved by creating the volumetric crystallization pattern in the
root weld and by accelerating weld metal cooling using external or internal
heat drainage [6].
Relation between Crystallization and Liquation Cracks
The described method reveals for the first time that plastic deformation of
the weld metal within BTR1, which does not cause hot cracking but reduces
the plasticity by over 50 %, also increases the tendency to hot crack114
Phenomena and Mechanisms
ing within BTR2 (Fig. 9). This phenomenon apparently occurred during
the tests carried out with the PVR method.
Metallographic investigations showed that plastic deformation within
BTR1 leads to an increase of migrations, segregations, linearity and sensibility
to pickling of grain boundaries, which also indicates an increased
tendency to high-temperature creep. It was found that the measured value
of high-temperature deformation within BTR1, i.e. after the end of crystallization,
has a significant relevance to many other characteristic material
properties, for example the resistance to low-cycle fatigue, the resistance to
heat loadings and to corrosion. It is proposed to explain this phenomenon,
which we named technological inheritance, by the same summation of
stacking defects by which the existence of two different BTRs is explained.
Fig. 9. Influence of plasticity exhaustion in BTR1 according to diagram (Fig.6) on
the index of resistibility to liquation hot cracking within BTR2
Morphology of Hot Cracks in Single-Phase Weld Metal 115
Conclusions
The suggested test device allows it to determine the tendency to liquation
hot cracking in solid-phase weld metal for repeated high-temperature cycles
more precisely than other test procedures.
A high tendency of hot cracking in austenitic steels exhibiting a small
temperature range of solidification results from merging of the solidification
and the liquation temperature ranges of brittleness.
A higher susceptibility to hot cracking in austenitic welds at high welding
speeds results from the formation of a flat crystallization pattern and
the concourse of crystallites under an obtuse angle in the center of the
weld, thus reducing the plasticity within both BTR1 and BTR2, respectively.
References
1. Jakushin BF (1969) Evaluation of technological durability depending on
modes of welding. Welding manufacture 1
2. Jakushin BF, Gadzhiev MN, Gritsenko AI, (1991) Calculated-experimental
method of determining weld metal resistibility to hot cracking. Welding
manufacture 4
3. Jakushin BF, Misjurov AI (1983) Technological durability of multilayered
welds with stably austenitic structure. Automatic welding 6
4. GOST of the USSR 26.389 – 84 (1985) Test methods for resistibility to hot
cracking during welding by fusion. Publishing house of committee of standards
5. Vishnjakov JD (1970) Stacking defects in crystal structure. M Metallurgy:
216
6. Makarov EL (1991) Weldability of materials: Reference book. M Metallurgy
7. Rabesteiner G, Tosch J and Schabereiter H (1983) Hot cracking problems in
different fully austenitic weld metals. Welding Journal, vol 62-1: 21–27
II Metallurgy and Materials
The Effect of Silicon and Iron on the Weldability
of Ni-Co-Cr-Si HR-160® Alloy
I.S. Maroef1, M.D. Rowe2, G.R. Edwards3
1The Netherlands Institute for Metal Researches, Delft, Netherlands
2Haynes International Inc., Kokomo, USA
3CWJCR, Colorado School of Mines, USA
Abstract
Solidification cracking of HAYNES HR-160®1 alloy was investigated with
an emphasis on the interactive effects between silicon and iron concentrations.
The focus on silicon was relevant to its major role in the sulfidation
and oxidation at elevated temperatures, while the focus on iron was to explore
the potential use for over-lay welding onto alloyed steels. Five compositions
of the alloy were investigated, covering a [2 x 2] matrix of low
and high concentrations of silicon and iron, in addition to the commercial
alloy. To find a relationship between thermodynamic properties, microstructures,
and weldability, several supporting analyses were performed.
These analyses were differential thermal analysis and electron microscopy
(chemical analysis), as well as fractographic investigations. Within the
tested levels of alloying addition, both silicon and iron additions proved to
increase the susceptibility of the alloy to solidification cracking, with silicon
having the predominant influence. The detrimental effect that silicon
caused was the increase in the solidification temperature range of the alloy,
which in turn increased the brittle temperature range of the alloys. On the
other hand, iron has the tendency to decrease the threshold strain to cracking,
even though it narrowed the brittle temperature range of the alloy for a
fixed concentration of silicon.
1 HAYNES and HR-160 are trademarks of Haynes International Inc.
120 Metallurgy and Materials
Introduction
The purpose of this study was to investigate the effect of silicon and iron
on the weldability of HAYNES HR-160® alloy. HR-160 alloy is a solid
solution strengthened Ni-Co-Cr-Si alloy. The alloy is designed to resist
corrosion in sulfidizing and other aggressive high temperature environments.
Silicon is added (~2.75 wt.pct.) to promote the formation of a protective
oxide scale in environments with low oxygen activity. HR-160
alloy has found applications in waste incinerators, calciners, pulp and paper
recovery boilers, coal gasification systems, and fluidized bed combustion
systems.
HR-160 alloy has been successfully used in a wide range of welded applications.
However, the alloy can be susceptible to solidification cracking
under conditions of severe restraint. A previous study by DuPont, et al. [1]
showed that silicon promoted solidification cracking in the commercial
alloy. In earlier work conducted at Haynes, and also from published work
by DuPont, et al., it was recognized that silicon segregates to the terminal
liquid, creating low melting point liquid films on solidification grain
boundaries. Solidification cracking has been encountered when using the
alloy as a weld overlay on steel, and when joining HR-160 plate in a thickness
greater than 19 mm with matching filler metal. The effect of silicon
on the weldability of HR-160 alloy has been well documented, but the
effect of iron is not well understood. Prior experience at Haynes has indicated
that iron may be detrimental to the solidification cracking resistance
of the alloy. Iron does not segregate to the terminal solidification product
in nickel-base alloys, as does silicon [2], but iron may have an indirect or
interactive influence on weldability. A set of alloys covering a range of
silicon and iron contents was prepared and characterized to better understand
the welding metallurgy of HR-160 alloy.
Materials and Experimental Procedures
Materials
For this study, four HR-160 type experimental alloys with different target
levels of silicon (2.3 and 3.9 %) and iron (0.10 and 4.0 %) were prepared.
The experimental heats represent a [2 x 2] matrix of high and low silicon
and iron. The heat 8727-7-7506 was included in this study to represent the
standard commercial alloy. The chemical compositions of the experimental
and commercial alloys are given in Table 1.
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 121
Table 1. Chemical compositions of the HR-160 type experimental and commercial
alloys (wt.pct.)
Haynes Intl.
Heat #
EN2799-9-
0894
EN2999-0-
0896 8727-7-7506 EN2899-9-
0895
EN3099-9-
0897
Nominal
Si & Fe 2.3% Si 2.3% Si 2.8% Si 4.0% Si 4.0% Si
Content 0.10% Fe 4.0% Fe 0.10% Fe 0.10% Fe 4.0% Fe
Source Commercial Exp. Exp. Exp. Exp.
Element
C 0.062 0.061 0.05 0.062 0.065
Cb <.01 0.01 <.050 0.01 0.01
Co 30.01 29.99 30.1 29.99 29.74
Cr 27.61 27.88 27.80 27.60 27.68
Fe 0.08 4.01 <0.10 0.09 3.96
Mn 0.55 0.53 0.46 0.54 0.52
Ni 38.65 34.40 37.11 36.72 33.04
P 0.002 <.002 0.002 0.005 0.006
S <.001 <.001 0.007 <.001 <.001
Si 2.34 2.24 2.81 4.09 4.11
Ti 0.5 0.49 0.46 0.5 0.44
The experimental alloys were vacuum melted, cast into electrodes, then
electroslag remelted into 102-mm-diam. ingots. The ingots were forged to
44.5 mm, then further hot-rolled to 12.5-mm plate. The materials for varestraint
testing were prepared by hot rolling at 1021 °C to 3-mm plate. The
alloys were annealed at 1021 °C for 20 minutes and subsequently water
quenched. Sigmajig specimens were taken from the hot rolled 12.5 mm
(0.5 in.)-thick plate to be hot rolled to 2 mm (0.080 in.), then cold rolled to
1 mm (0.04 in.), followed by an anneal at 1100 °C.
Weldability Tests
Longitudinal varestraint tests were undertaken with welding parameters
listed in Table 2. Each experimental alloy was tested at four augmented
strain levels (nominal values of 1.0 to 3.5) in triplicate. The total crack
length (TCL) and maximum crack distance (MCD) were taken as indicators
of the cracking susceptibility. The maximum crack distance, MCD, is
defined as the perpendicular distance from the fusion boundary to the farend
tip of the crack. Cracks were measured in the as-welded condition with
a stereoscopic microscope and a filar eyepiece. An average of three timetemperature
profiles were collected from a representative sample
122 Metallurgy and Materials
(commercial HR-160 alloy), specifically at the trailing edge of the fusion
line of the weld pool center. The temperature measurements were accomplished
by manually harpooning pure Pt/ Pt-13 % Re type R thermocouple
wires to the weld pool during GTA welding. Before harpooning, the wires
were kept trailing near the welding arc for several seconds to accumulate
enough heat and thus minimize the thermal inertial problem in the temperature
measurement. To approach the condition that occurs in the varestraint
test, a copper block was placed underneath to support part of the
length of the sample. GTA welding started at the unsupported part of the
sample and thermocouple harpooning was executed when the weld pool
traveled on the copper-block supported part. The average cooling rate, over
the temperature range of 1400 and 1000 °C and at the trailing edge of the
weld pool, was found to be 200 ± 30 °C/s.
Sigmajig testing of 50 mm × 50 mm coupons was also performed. As an
indicator of cracking susceptibility, a threshold stress for crack initiation
was measured. To determine the threshold stress, the sigmajig test was performed
over a range of applied stress, and the threshold stress was taken as
the minimum stress to produce cracking. Hot cracks that formed in the fusion
zone were measured under a stereoscopic microscope.
Table 2. Welding parameters and specimen dimensions used in weldability tests
Varestraint Test Sigmajig Test
Specimendimensions (mm) 25 × 152 × 3 50 × 50 × 1
Current (Amp.) 70 20
Travel speed (mm/s) 1.9 14.8
Arc length (mm) 2.4 2.4
Shielding gas (L/s) argon 0.28 0.28
Type of electrode 2 % thoriated 2 % thoriated
Electrode include angle 60o 60o
Arc length (mm) 1 1
Weld configuration Autogenous Autogenous
Microstructural Characterization
Metallographic samples were taken from welded specimens to analyze resulting
microstructures. Samples were electrolytically etched at 6 V using
a solution of 5 g oxalic acid dissolved in 95 ml of hydrochloric acid. The
weld microstructures were initially analyzed using light optical microscopy.
The volume percent of interdendritic eutectic constituent was estimated
by measuring its area fraction on a metallographic cross-section,
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 123
using image analysis software. The average of 20 fields analyzed at 200x
magnification was reported.
Selected fracture samples were analyzed with the scanning electron
microscope for a better picture of the dendrite structures that were responsible
for reduction or increase in cracking susceptibility. For chemical
analysis of selected alloys, electron probe X-ray micro-analysis (EPMAWDS)
was conducted with a JEOL JXA 8900R, at electron beam energy
of 15 keV and current of 50 nA, as well as a focused beam of 6 nm resolution.
The measurements used Kα lines for Ni, Co, Cr, Si, Ti, Fe, and
Mn. All samples for EPMA-WDS analysis were not epoxy-mounted, but
were polished flat with 0.25 μm alumina slurry, then ultrasonically cleaned
in acetone prior to analysis. The target outcome of the WDS measurement
was an accurate assessment of the partitioning coefficient. Under an assumption
of ideal condition satisfying the Scheil equation (no solid backdiffusion),
the apparent partitioning coefficient (k = kCo/Co = Cs/Co) of an
element of interest can be estimated from the ratio of the concentration at
the dendrite core (Cs) to that of the bulk material surrounding the core (Co).
Six-spot readings from nearby dendrite cores were collected in search of
the three lowest concentrations of the segregating element. The three
values were then averaged for use in the estimation of the apparent partitioning
coefficient. The bulk composition was assessed in the weld metal
and at the same location as that of the spots, covering an area of approximately
1600 μm2. In average, about 20 dendrites were covered across this
area of measurement. The area for both spot and bulk analyses was selected
to be the weld metal adjacent to the HAZ of the sample, where primary
dendrite morphology is relatively clear and where the cooling rate
near solidification temperature was expected to be the highest.
Differential Thermal Analysis
Differential thermal analysis (DTA) was performed with a Netsch Thermal
Analyzer STA 409, using alumina as a reference. This analysis was
intended to provide some thermodynamic basis to explain the trends in
solidification cracking susceptibility of the different alloys. As a parameter
for solidification susceptibility, the solidification temperature range (STR)
assessed from DTA was determined. Alumina crucibles and lids, designed
for differential scanning calorimetry (DSC), were used to increase sensitivity
of the analysis. During initial analyses (DTA1), a sample mass of
150 ± 20 mg was found sufficient for easy detection of reaction peaks. Onheating
rate was determined to be 0.083 °C/s (5 °C/min.) to obtain near
equilibrium solidus and liquidus temperatures. On the other hand, a rate of
124 Metallurgy and Materials
0.33 °C/s (20 °C/min.) was selected for on-cooling analysis. In this way,
reaction temperatures associated with non-equilibrium solidification can be
measured. To detect the liquidus more accurately during cooling cycle, the
analyses were repeated (DTA2) with an initial cooling rate of 0.05 °C/s to
1250 °C, followed by a final cooling rate of 0.33 °C/s. This final cooling
rate was prescribed to allow easy detection of the terminal liquid reaction
temperatures. The sample mass for DTA2 was determined to be
300 ± 30 mg.
Results
Weldability Data
The maximum crack distance (MCD) and the corresponding total crack
length (TCL) obtained from the longitudinal varestraint test at four levels
of augmented strain are shown in Fig. 1 and Fig. 2 , respectively. A decrease
in the value of the cracking index for strains above 2.5 %, was evident.
However, such reductions were not considered as an improvement of
weldability.
Fig. 1. Longitudinal varestraint data. Average maximum crack distance (MCD)
vs. augmented strain
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 125
Fig. 2. Longitudinal varestraint data. Average total crack length (TCL)
vs. augmented strain
Hence, an average value of crack lengths or distances was taken as the
measure of susceptibility at strain levels of 2.5 % and 3.5 %. Many possible
factors might have contributed to such reductions, such as change of
restraint or increase in HAZ hot cracking, with increasing level of strain.
To get a better interpretation of the effect of silicon and iron addition on
the weldability of the experimental HR-160 alloys, the values of MCDs
and TCLs above 2.5 % are further summarized as a function of silicon
concentration in Fig. 3 and Fig. 4, along with the threshold stress obtained
from the sigma-jig test. Data points of iron-rich samples are filled with
gray color to show the effect of iron for a given silicon concentration.
The summary plot in Fig. 3 showed that both the peak values of MCDs
(at 2.5-3.5 % strain) and sigmajig threshold stresses were in a remarkably
good agreement, indicating the predominantly detrimental effect of silicon
over that of iron. Both types of data showed that the cracking susceptibility
increased with an increase in silicon content up to approximately the
commercial HR-160 alloy’s concentration (2.3 wt.pct. Si).
Then, the susceptibility leveled off at higher concentrations of silicon,
within this test matrix. When represented by the peak MCDs and sigma-jig
threshold stresses, the weldability of the alloys appeared to be un-affected,
if not improved, by the addition of iron. However, caution is necessary because
detail in the varestraint data is not shown in such a summary plot.
126 Metallurgy and Materials
Fig. 3. The maximum crack distance (MCD) and sigmajig threshold stress
as a function of silicon concentration
Fig. 4. The total crack length (TCL) and sigmajig threshold stress
as a function of silicon concentration
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 127
Recall that for any of the high iron-containing alloy, the MCD value,
and thus the cracking susceptibility, was always slightly higher at small
levels of augmented strains. Testing condition at low levels of augmented
strain in a varestraint test is thought to resemble an actual welding restraint
severity closer than does a test at higher augmented strain level. It is worth
mentioning that the local strains n front of the crack can be very large,
even at low levels of augmented strain.
Despite the good agreement between MCDs and sigmajig threshold
stresses, the authors considered not to rely too heavily on the sigma-jig
data to support the analysis presented in the following discussions. The
main reservation of using the sigma-jig was the low level of stresses measured
during the testing. The sigmajig test has been shown to be marginally
adequate at low values of threshold stress. Published experimental work
[3], supported by finite element calculations, revealed that “free-stress”
(zero preset loading) specimens could crack during welding solidification,
but resisted cracking at moderate levels of pre-set loading.
It was found that, due to the peculiarity of the sigmajig set-up, the freestress
condition lacks sample restraint. During heating, the sample can easily
expand in all directions, causing significant tensile stress in the transverse
direction during the cooling cycle. A minimum level of preset load is
required to prevent such a free expansion. Obviously, such a limitation
could be overcome by using thicker samples. However, because sigma-jig
testing was originally designed for thin sheet materials, production of
thicker samples was not considered in the initial set-up of the study.
In contrast to MCD data, the total crack length (TCL) in Fig. 4 showed
that cracking susceptibility continuously increased with an increase in silicon,
within the concentration established by the test matrix. There was no
measurable effect of iron, at any particular silicon concentration, to the
values of TCL. The predominant effect of silicon was even more obvious
when one considered these TCL data.
Quantitative Metallography
Fig. 5 shows the volume fraction of interdendritic eutectic, estimated from
its area fraction, as a function of the silicon concentration, both from the
varestraint and the sigmajig test samples. The non-metallic phase that
forms the eutectic has been identified by Dupont as the silicon-rich G
phase [(Ni, Co)16(Ti, Cr)6(Si)7], which is known to form during solidification
of the HR-160 alloy [1]. In this study, characterization of the eutectic
phases was done only qualitatively (SEM-EDS) because, within the range
of composition varied in this study, no change in eutectic solidification
128 Metallurgy and Materials
Fig. 9. The quantity of interdendritic eutectic product in the weld metal
of HR-160 type alloys with different silicon and iron concentrations,
as measured from varestraint and sigmajig specimens
product was anticipated. The amount of eutectic product appears to be
directly proportional to the silicon content of the alloy.
The comparison of the eutectic amount and morphology of the weld
metals studied can be seen from Fig. 6. The two low-Si HR-160 alloys
clearly produced the lowest amount of eutectic upon welding. Their
eutectic phases were droplet-shaped and completely isolated. The low-Si,
high-Fe alloy produced a notably higher amount of eutectic than did the
low-Si, low-Fe alloy. In the commercial HR-160 alloy, the amount and the
liquid film network of interdendritic eutectic product are visibly less than
in the two high-Si experimental alloys. Between the two high-Si alloys, the
liquid film network of the eutectic product was roughly similar. It was recognized
that the solidification structure is complex, and that more than one
cross-sectional view would be necessary for one to compare the quantity
and distribution of the liquid film network created by thermally cycling a
given material. The ratio of TCL to MCD represents the average number
of cracks occur at a given level of strain. As generally observed, the number
of cracks in this study increased with increasing level of strain. One
possible reason is that the MCD values did not increase linearly with
strain, but even leveled off at 2.5 % strain. Therefore, as long as weak or
brittle parts of the sample were still available, additional cracks would be
initiated to accommodate the global strain imposed by the set-up.
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 129
Fig. 10. Microstructures of longitudinal varestraint test specimens
(a) 2.0-wt.-pct.-Si, 0.1-wt.-pct.-Fe HR-160 experimental alloy
(b) 2.0-wt.-pct.-Si, 4.0-wt.-pct.-Fe HR-160 experimental alloy
(c) 2.8-wt.-pct.-Si, 0.1-wt.-pct.-Fe HR-160 commercial alloy
(d) 4.0-wt.-pct.-Si, 0.1-wt.-pct.-Fe HR-160 experimental alloy
(e) 4.0-wt.-pct.-Si, 4.0-wt.-pct.-Fe HR-160 experimental alloy
Accordingly, as shown in Fig. 7, the larger the volume fraction (fe) of
terminal liquid eutectic, the larger was the value of TCL. A structure with
a large fraction of interdendritic eutectic phase contained many locations
with poor coherency, which acted as easy crack initiation sites to accommodate
a global augmented strain.
Fig. 11. The total crack length (TCL) and the volume fraction
of terminal liquid eutectic (fe), as a function of the silicon concentration
130 Metallurgy and Materials
Fig. 12. The maximum crack distance (MCD) and the volume fraction
of terminal liquid eutectic (fe), as a function of the silicon concentration
In other words, larger TCL did not necessarily imply that the alloy is
more brittle, but there were higher density of potential crack initiation
sites. As a comparison, Fig. 8 shows that the MCD values did not depend
on fe after the silicon concentration exceeded 2.8 wt.pct.
Scanning Electron Fractography
Details of the fracture surface of the three most susceptible alloys (the
commercial alloy and the two high-Si experimental HR-160 alloys) are
shown from Fig. 9 to Fig. 11. The fracture surfaces shown were documented
from specimens, varestraint tested at 1.5 %. strain. The fractographs
shown were located immediately under the weld bead surface. The
location of each surface was not necessarily nearest to the fusion line;
hence the selected fracture surfaces may not represent cracking at identical
temperatures.
In the fracture surface of the commercial HR-160 alloy, as shown in
Fig. 9, classic dendritic structure is shown as evidence of solidification
cracking. The dendrites appeared to be predominantly columnar, with
moderate protrusions of secondary dendrite arms. Some plasticity, manifested
as slip lines, is notable on the dendrite surfaces at high magnification.
The slip lines were not necessarily involved in the cracking initiation
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 131
process, but were most likely a result of crack opening displacement to accommodate
the augmented strain.
Fig. 17. SEM fractographs of commercial HR-160 weld metal
The fracture surface of the high-Si, low-Fe alloy is shown in Fig. 10.
The primary dendritic columns are easily identified, one separated from
another by a relatively long span of secondary dendrite arms. The surface
revealed an apparently preferential growth of secondary arm dendrite
growth in two dimensions.
Fig. 18. SEM fractographs of 4.0-wt.-pct.-Si and 0.1-wt.-pct.-Fe HR-160 type
experimental weld metal
132 Metallurgy and Materials
Fig. 19. SEM fractographs of 2.3-wt. pct.-Si and 4.0-wt. pct.-Fe HR-160 type
experimental weld metal
In contrast to fracture surfaces in the high-Si, low-Fe alloy, the fracture
surface of the high-Si, high-Fe alloy (Fig. 11) revealed a more refined
dendritic and tortuous structure. Such a three dimensional shape was not as
well revealed in the cross-sectional micrographs presented previously in
Fig. 6d. Some bridging, which appeared as brighter colored islands, was
evident on the dendrite surfaces. At first, these islands gave an impression
that they were the eutectic microconstituents, but back-scattered imaging
did not distinctively show that they contained a high concentration of low
atomic elements, such as silicon.
Partitioning Coefficients of Alloying Elements
Susceptibility to solidification cracking of a weld metal is frequently related
to the quantity of the interdendritic eutectic product and the solidification
structure itself. Both are governed by microsegregation, which itself
depends on both the partitioning coefficient (k) of the particular elements
in the alloy besides the solidification temperature range. Values of k for
major elements can be approximated by dividing the measured dendrite
core’s composition (Cs) by the bulk alloy’s composition (Co); k = kCo/Co =
Cs/Co. The method assumes that dendrite tip curvature effects and solidstate
diffusion are negligible, that thermodynamic equilibrium is maintained
at the solid/liquid interface, and that diffusion of elements in the
liquid is very fast [4]. According to this definition, an element with a low k
value segregates strongly to the terminal liquid phase.
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 133
Table 3. Apparent values of initial partitioning coefficients, k, of selected elements
in two high-Si and commercial HR-160 alloys
4.0 Si – 0.1 Fe
HR-160
4.0 Si – 4.0 Fe
HR-160
Commercial
Element HR-160
C0 Cs C0 Cs C0 Cs
Iron 0.08 0.08 4.10 4.42 0.11 0.11
Nickel 37.92 36.98 32.90 31.75 35.75 35.10
Chromium 26.57 27.23 29.57 29.67 31.33 31.18
Cobalt 30.98 32.30 29.06 31.11 29.05 30.68
Silicon 3.56 2.74 3.73 2.52 2.81 2.11
Titanium 0.44 0.25 0.46 0.18 0.47 0.26
Manganese 0.50 0.42 0.46 0.34 0.58 0.54
4.0 Si – 0.1 Fe
HR-160
4.0 Si – 4.0 Fe
HR-160
Commercial
Element HR-160
k (+/-)of k k (+/-)of k k (+/-)of k
Iron 1.00 0.03 1.08 0.02 1.00 0.02
Nickel 0.98 0.02 0.97 0.03 0.98 0.05
Chromium 1.02 0.05 1.00 0.05 1.00 0.03
Cobalt 1.04 0.04 1.07 0.02 1.06 0.04
Silicon 0.77 0.04 0.68 0.03 0.75 0.03
Titanium 0.57 0.02 0.39 0.04 0.55 0.02
Manganese 0.84 0.03 0.74 0.02 0.93 0.02
Cs concentration at the dendrite core (at zero fraction of solid), C0 bulk composition
(all in wt.pct).
The apparent k values for selected elements were calculated for the two
high-Si and the commercial HR-160 alloys are listed in Table 3.
Table 4. Apparent values of initial partitioning coefficients, k, HR-160 commercial
alloy and HR-160 overlay weld on 2.25Cr-1Mo steel (Dupont [1, 2])
Commercial HR-160
(DuPont)
Weld Overlay
Element (DuPont)
C0 Cs k C0 Cs k
Iron 32.91 37.53 1.14
Nickel 37.90 36.50 0.96 25.83 23.79 0.92
Chromium 27.64 28.00 1.01 19.22 19.02 0.99
Cobalt 30.54 32.90 1.08 20.01 21.54 1.08
Silicon 2.60 1.85 0.71 1.93 1.02 0.53
Titanium 0.48 0.21 0.44 0.34 ND NA
Cs concentration at the dendrite core (at zero fraction of solid), C0 bulk composition
(all in wt.pct.), ND not detected.
134 Metallurgy and Materials
As a comparison, Table 4 includes the apparent k values assessed by
DuPont [1, 2] from both the commercial HR-160 alloy and weld overlay of
HR-160 superalloy on 2.5Cr-1Mo steel. The addition of iron to the high-Si
HR-160 alloy increased the partitioning coefficient of silicon. With even
higher iron content, as in the weld overlay data of DuPont, it was shown
that the k value of silicon was further decreased.
Differential Thermal Analysis
As a typical DTA thermogram, Fig. 12 exhibits the DTA1 results for the
commercial HR-160 alloy, containing both the on-heating and on-cooling
thermogram curves. During heating, the sample exhibits a solidus temperature
of approximately 1300 °C (determined as the onset of a detectable difference
in the endothermic part of the curve) and reaches the liquidus temperature
at 1370 °C (determined as the peak of the endothermic reaction).
The near equilibrium melting range was 70 °C, similar to the result reported
earlier by DuPont [1]. Upon solidification in alumina crucible, the
alloy exhibited a large exothermic peak at an under-cooled temperature of
1340 °C, which corresponded to the formation of the primary dendrites.
Such an undercooling also occurred at the analyses for the other HR-160
experimental alloys. Solidification was completed with a terminal liquid
exothermic reaction at 1150 °C.
Fig. 23. DTA thermogram of the commercial HR-160 alloy during melting at
0.083 °C/s and solidification at 0.33 °C/s
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 135
Table 5. Results of Differential Thermal Analysis and Apparent Solidification
Temperature Ranges
Alloy On
Heating
Liquidus
(°C) a
On
Cooling
Liquidus
(°C) b
Terminal
Liquid
(°C) c
Solid.
Temp.
Range
STR1 (°C)
Solid.
Temp.
Range
STR2 (°C)
2.3%Si,
0.1%Fe 1378 ± 5 1350 ± 5 1280 ± 10 98± 10 70± 10
2.3%Si,
4.0%Fe 1376 ± 5 1360 ± 5 1272 ± 10 96± 10 88± 10
2.8%Si,
0.1%Fe 1371 ± 5 1364 ± 5 1152 ± 5 219 ± 5 212 ± 5
4.0%Si,
0.1%Fe 1340 ± 5 1330 ± 5 1125 ± 5 215 ± 5 205 ± 5
4.0%Si,
4.0%Fe 1341 ± 5 1330 ± 5 1130 ± 5 211 ± 5 200 ± 5
STR1 On Heating Liquidus Temp. – Terminal Liquid Temp., values from DTA1,
STR2 On Cooling Liquidus Temp. – Terminal Liquid Temp., values from DTA2.
a values were assessed from DTA1 with heating rate of 0.083 °C/s to 1500°C and
cooling rate of 0.33 °C/s.
b values were assessed from DTA2 with heating rate of 0.33 °C/s to 1500 °C and
initial cooling rate of 0.05 °C/s to 1250 °C, followed by final cooling rate of
0.33 °C/s for detection of terminal liquid temperature.
c values for terminal liquid temperature was an average of values assessed with
DTA1 and DTA2.
To better approximate weld metal solidification (epitaxial growth), the
solidification temperature ranges 1 (STR1) for the alloys were taken as the
difference between the on-heating liquidus temperature and the on-cooling
terminal liquid reaction temperature. As a complementary data, the solidification
temperature ranges 2 (STR2) were also calculated, from the difference
between the on-cooling liquidus and the terminal liquid temperatures,
assessed during DTA2 analyses described in the experimental procedure.
For the HR-160 commercial alloy, the STR1 and STR2 were 219 °C and
212 °C, respectively. Table 5 lists the reaction temperatures and the STR
values of the alloys.
Discussion
The following discussion is intended not only to explore the effect of composition
on the solidification cracking susceptibility, but also to relate the
results to established hot tearing criteria. Factors that influence solidi136
Metallurgy and Materials
fication cracking susceptibility have been taken into account in different
theoretical formulations, although no single one has satisfactorily described
the complete picture of the mechanism involved. Two main
schools of thought on this subject are: 1) those of the structural considerations,
such as the Clyne-Davis concept [5] and 2) those of the rheological
considerations, such the Feurer’s competitive processes between liquid
back-filling and shrinkage [6]. The validity of each criterion appears to depend
strongly on the type of process (welding or casting) and the mechanical
testing applied (stress or strain based).
However, it is generally accepted that a solidifying metal starts to be
vulnerable to solidification cracking when coherency is reached [7]. Coherency
is defined as the condition when adjacent dendrites start to form
solid/solid contacts.
A common way to analyze data obtained from a strain-based testing,
such as the varestraint test, is to plot the augmented strain as a function of
the cracking temperature range. This plot is usually termed the ductility
curve. Such a plot for the data assessed in this study is shown in Fig. 13
below. The temperature range for cracking is deduced from the maximum
crack distance, taking the temperature of the fusion line as the on-cooling
liquidus of the alloy, as assessed by the DTA2 analysis. When considering
the epitaxial growth during weld metal solidification, estimating the fusion
line temperature from the on-cooling liquidus determined by DTA2 analysis
(slow cooling rate) would be quite realistic. The temperature at the
crack tip was estimated from knowledge of the cooling rate, over the temperature
range of 1400 to 1000 °C, and at the trailing edge of the weld pool
(an average of 200 ± 30 °C/s). Weldability of the alloy is then measured by
two parameters: the brittle temperature range (BTR), and the threshold
strain to initiate cracking. BTR is defined as the maximum range of cracking
temperature covered by the ductility curve. An excellent weldability is
then characterized by short BTR and high threshold strain to cracking.
In this study, the threshold strain to initiate cracking (an alternate ductility
indicator) could not be accurately quantified because strain blocks that
could give lower levels of strains were not available. To highlight the trend
in the threshold strain to cracking, the cracking temperature range at 1 %
augmented strain (CTR1%) of the alloys will be used for comparison purposes.
It was assumed that a large value of CTR1% would tend to yield a
low value of threshold strain to cracking, by extrapolation of the ductility
curves.
As shown in Fig. 13, the two low-Si HR160 alloys clearly outperformed
the other alloys, as signified by their short BTRs, as well as by
their short CTR1%. If not for the tendency to yield a lower apparent
threshold strain to cracking, the low-Si, high-Fe HR160 alloy would cerThe
Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 137
tainly be categorized to be as resistant to cracking as its low-Fe sister.
Both the commercial and the high-Si low-Fe HR-160 alloys had similar
values of BTR and CTR1%. Both alloys’ BTRs were also the widest
among the alloys studied. In contrast to these two later alloys, the high-Si,
high-Fe HR-160 alloy yielded a BTR about 30 °C shorter, but cracked at a
higher value of CTR1%. In other words, by extrapolation of the ductility
curve, the high-Si, high-Fe HR-160 alloy tended to yield a lower apparent
threshold strain than do the commercial and the high-Si, low Fe HR-160
alloys.
Fig. 24. Ductility curve (response) for the various HR-160 alloys upon rapid
straining during the longitudinal varestraint testing
With respect to hot tearing criteria, the varestraint test may pose some
limitation in contrasting the rheological process (liquid back-filling)
among different alloys. As highlighted by Matsuda [8], the major part of a
solidification crack in the varestraint test propagates rapidly through a
rigid structure, with only dispersed and droplet-like terminal liquid.
Accordingly, the varestraint test emphasizes the structural factors of solidification
cracking as opposed to the rheological factors, particularly at
high strain levels. However, at low strain levels, the rheological process
may play a bigger role, and may better demonstrate the beneficial effect of
back-filling by a weld metal with a large volume fraction of terminal liquid
(beginning from 10 to 20 pct).
138 Metallurgy and Materials
The BTR values, assessed at high augmented strain levels, were predominated
by the region of fast crack propagation. In this mode of fracture,
the structural factors played a big role. Therefore, the material parameters
that influenced the structure formation must be explored to
elucidate the role of chemical composition on the solidification cracking
susceptibility. Among the many materials parameters, the predominant
effect of k has been well recognized [9].
To focus on the role of k, specifically kSi, the two high-Si alloys were selected
for a comparison study because their values of STR were similar.
Comparison of the solidification structures among the alloys of interest
from cross-sectional metallography was not successful, especially when
the columnar growth followed the weld pool in a curved trajectory. For
this purpose, the solidification structures revealed by the fracture surfaces
in Fig. 10 to Fig. 11 were considered as alternative choices to assess signatures
of growing dendrite structures. A revealing effect produced by the
two fracture surfaces, was a decrease in kSi (from 0.78 to 0.68) led to an increase
in tortuosity of the dendrites. It follows that an increase in tortuosity
implied an increase in coherency of the structure, as the density of
solid/solid bridging during solidification should also be increased. The observation
that the BTR value of the high-Si, high-Fe Hr-160 alloy (smallest
kSi) was considerably higher than that of the high-Si, low-Fe alloy appeared
to be due to a possible increase in load bearing capacity when coherency
was increased. Although the high-Si, high,-Fe HR-160 alloy was superior
with respect to BTR, an explanation for why this alloy performed poorly
with respect to the trend of threshold strain to cracking (higher value of
CTR1%) remains elusive. Further study is needed to clarify whether it was
the rheological factor (degree of back-filling) or the structural factor (degree
of coherency) predominant in causing the significant reduction of the
threshold strains, as the iron concentration was increased.
Fig. 14 shows the relationship between BTR and two materials parameters;
namely the silicon partitioning coefficient (kSi) and the solidification
temperature range (STR). Without the presence of iron, it was clear that
BTR increased with increasing STR, although the function was not linear.
The addition of iron then made the change in the value of BTR unpredictable.
There also appeared to be a relationship between kSi and BTR, for
the three alloys for which the data were available. The relationship was
also independent of the composition of the alloys. However, more experimental
data is necessary to confirm this apparent correlation between kSi
and BTR. The eutectic volume fraction was not included in the figure because
it has been found to lack in correlation with either BTR (MCD), or
with kSi. This observation may change if another testing methodology is
used, such as the Houldcroft or circular-patch test.
The Effect of Silicon and Iron on the Weldability of Ni-Co-Cr-Si HR-160® 139
Fig. 25. Relationship between brittle temperature range (BTR) and silicon partitioning
coefficient ksi, as well as the solidification temperature range (STR)
Conclusions
1. Within the alloying levels tested, both silicon and iron additions proved
to increase the susceptibility of the alloy to solidification cracking, with
silicon having the predominant influence.
2. At a fixed iron concentration, the addition of silicon increased the brittle
temperature range of the HR-160 alloys.
3. The solidification temperature ranges of the HR-160 alloys depended
upon the silicon content, with iron having negligible measurable effect.
4. At a fixed silicon concentration, the addition of iron decreased the brittle
temperature range. However, such an apparently beneficial effect was
compromised by a concurrent increase in the cracking temperature
range at 1 % augmented strain. It is anticipated that iron had the tendency
to yield a reduction in the threshold strain to cracking.
5. In high-Si HR-160 alloys, the addition of iron decreased the partitioning
coefficient of silicon.
140 Metallurgy and Materials
Acknowledgement
This research was conducted at the Colorado School of Mines and Haynes
International. The financial support of the Fossil Energy Materials Program,
U.S. Department of Energy, under Contract Number 19X-SW314C,
is gratefully acknowledged. The authors also gratefully acknowledge the
support of Dr. Stan David, Materials Division, ORNL and the technical assistance
of Ms. Brenda Mulac, formerly of CSM. Ing. J. Kiersch and Ing.
C. Kwaakernaak, from the Surface and Interface Research group – Delft
University of Technology, the Netherlands, are also gratefully acknowledged
for their help on the EPMA-WDS measurement. The support from
the Netherlands Institute for Metal Researches (NIMR) that enabled the
first author to complete the study was highly appreciated.
References
1. DuPont JN, Micheal JR, Newbury BD (1999) Welding Journal 81: 408–415
2. Dupont JN (1997) J. Materials Science 32: 4101–4107
3. Feng Z, Zacharia T, David SA (1997) Welding Journal 79: 470–483
4. Brody HD, Flemmings MC (1966) Trans. AIME 236: 615–623
5. Clyne TW, Davies GJ (1975) British Foundry 68: 238–244
6. Feurer U (1976) Giessereiforschung, 28: 75
7. Borland JC (1960) British Welding Journal 7: 508–512
8. Matsuda F, Nakagawa H, Sorada K (1982) Trans. JWRI, 11(2): 67–77
9. Flemming MC (1974) Solidification Processing. McGraw Hill, New York
The Influence of Different Nb-Contents on the Hot
Cracking Susceptibility of Ni-Base Weld Metals
Type 70/20
R. Vallant
Institute for Materials Science, Welding and Forming,
Graz University of Technology, Austria
Abstract
Metal-cored wires and flux-cored wires for GMAW of different Nbcontents
have been tested concerning their hot cracking susceptibility. Using
tensile test specimens and the PVR-test, an optimum content could be
found, to cause low hot cracking susceptibility comparable to other welding
consumables on the market. This improvement was due to the formation
of NbC and by keeping the basicity index at a definite value, whereby
the required mechanical-technological values could be fulfilled.
Introduction
The background of this investigation was the development of a Flux-cored
wire (FCW) of type Ni-base 70/20, whereby the arc stability and the slag
viscosity (weldability in horizontal position and position vertical up, bead
covering and slag detachability) were optimized. However, the required
fracture strain in Table 1 could not be reached, because of the appearance
of hot cracks.
From the historical point of view the Ni-base welding alloy type 70/15
(SG-NiCr15FeMn, alloy 182) was developed in North America and was
standardized in AWS A5.14.
However it came out, that this welding alloy was susceptible to hot
cracks and to subsequent stress corrosion cracking (SCC), when welding
thick plates, especially in reactor constructions.
142 Metallurgy and Materials
Hence the Ni-base welding alloy type 70/20 (SG-NiCr19Nb, alloy 82)
was developed in Europe, which has considerable lower hot cracking
susceptibility (HCS) and susceptibility to SCC than type 70/15. From there
it appeared, that no solid wires type 70/15 are available, as the higher alloyed
type 70/20 meets its requirements. Consequently the similar base
material alloy 600 is usually welded with “overalloyed” welding rod or
solid wire, however welding with MMA-Electrode of type 70/15 is also
permitted [1].
Table 2 shows the standard All-weld metal composition of Ni-base
70/20, Ni-base 70/15 and base metal alloy 600 as well.
The function of Nb in Ni-base weld metals and stainless weld metals as
well, is to stabilize the C by forming NbC. By that the formation of Cr-
Carbides (sensitization) is avoided and the susceptibility for intergranular
corrosion should be lowered.
Additionally the very stable NbC should also improve the creep strength
of the weld metal for high temperature applications. For the high cooling
speeds in welding the Nb-content has to be distinctly higher than the
stoichiometric content 8 x % C. The weld metal has to be so-called “overstabilized”.
Unlike the base metal a stabilization using Ti is not advisable,
because of the flashing of Ti in the welding arc, due to the high affinity to
O [2].
In the similar base metal alloy 600 the formation of NbC seemed to correlate
with increased protection from intergranular corrosion IGC, where a
suitable addition of Nb/C>30 had a positive influence [3]. However, increasing
the Nb content for alloy 690 welds, resulted in decreased IGC resistance
within the interdendritic spaces [4].
The stated Nb-contents of Ni-base welding alloy types 70/20 (bare rods
and weld metal) in European and AWS standards are 1.5–3 and 2.0–3.0 %
respectively. The maximum carbon content is stated to 0.10 and 0.05 % respectively
[5 – 9]. The objective of this investigation was to evaluate the
influence of Nb within and out of this tolerances, see Table 2.
Table 1. Standard requirements for the mechanical values of Ni-base 70/20 and
70/15 All-weld metal
Standard Abbreviation Rp0.2 [MPa] Rm [MPa] A5 [%]
DIN 1736-2 NiCr20Nb 360 600 30
DIN 1736-2 NiCr15FeMn 360 600 30
AWS A5.11 ENiCrFe-3 – 550 30
AWS A5.14 ERNiCr-3 – 550 30
Table 2. Standard all-weld metal composition of Ni-base 70/20 and 70/15, bare electrodes and similar base metal.
Kurzzeichen –DIN EN Norm/
Classification-AWS Spec.
W.-Nr./
UNS No.
C Si Mn P S Cr Ni Nb Fe Ti Mo Cu Othersa
min 14 72 6 NiCr15Fe,
alloy 600
2.4816
N06600 max 0.08 0.5 17 10 0.3 0.5 0.3 Al
min 14 72
Base metal
LC-NiCr15Fe,
alloy 600 L
2.4817
N06600 max 0.03 0.5 17 0.3
min 2.5 18.0 65.0 1.5 NiCr20Nb DIN 1736 Draht-,
Bandelektroden 2.4806 max 0.05 0.5 3.5 0.015 22.0 3.0 3.0 0.8 0.5
min 5.0 13.0 61.0 1.0 2.0 NiCr15FeMn DIN 1736
Schweißgut Stabelektrode 2.4807 max 0.10 1.0 10.0 0.015 17.0 3.5 9.0 1.0 0.5 0.3 Ta
min 2.5 18.0 67.0 2.0 ERNiCr-3 AWS A5.14 Bare
Electrodes and Rods N06082 max 0.10 0.5 3.5 0.03 0.015 22.0 3.0 3.0 0.75 0.5 0.5
min 5.0 13.0 59.0 1.0 ENiCrFe-3 AWS A5.11 Weld
Metal SMAW-Electrode W86182 max 0.10 1.0 9.5 0.03 0.015 17.0 2.5 10.0 1.0 0.5 0.5
min 2.0 18.0 63.0 1.5 NiCr20Mn3Nb
prEN ISO 14172,
Schweißgut Stabelektrode
6077 max 0.10 0.5 6.0 22.0 3.0 4.0 0.5 2.0 0.5
min 5.0 13.0 60.0 1.0 2.0 NiCr15Fe6Mn
prEN ISO 14172, Schweißgut
Stabelektrode
6182 max 0.10 1.0 10.0 17.0 3.5 10.0 1.0 0.5 0.3 Ta
min 2.5 18.0 67.0 2.0 NiCr20Mn3Nb
prENISO 18274,
Draht-, Bandelektroden
6082 max 0.10 0.5 3.5 22.0 3.0 3.0 0.7 0.5
min 15.0 70.0 1.5 5.0
Bare Electrodes/ All-weld metal
NiCr16Fe8Nb
prEN ISO 18274,
Draht-, Bandelektroden
6062 max 0.08 0.3 1.0 17.0 3.0 9.0 0.5
Ni-base Type 70/20 Ni-base Type 70/15.
a Sum must not exceed 0.5 wt%.
The Influence of Different Nb-Contents 143
144 Metallurgy and Materials
Experimental Procedures
Self-Loaded Hot Crack Testing: Hook Cracks on Tensile Test
Specimens
The evaluation of the so called Hook cracks on tensile specimens according
to DIN 32525 was recommended as a self-loaded hot crack testing
method in 1990 [10] (Fig. 1). Hot crack investigations with this method
were already performed in earlier decades [11, 12]. As described, with this
simple method only distinctions can be made between lower or higher
HCS of weld metals, i.e. the preliminary selection of weld metals. For detailed
investigation this method is not suitable, as solidification cracks and
liquation cracks can not be kept apart [12].
Fig. 1. Hook cracks on the surface of a tensile test specimen (SC and LC widened
by loading)
For the evaluation of the HCS of ductile weld metals Pohle [13] has
suggested to use the reduction of area over the number of so called Hook
cracks found on the surface of broken tensile specimens by using a magnifier
(6–10 fold).
The Testing program I is shown in Table 3: The tensile test specimens
were made out of All-weld metal (AWM), acc. EN 1597-1 and weld metal
of a Plate Deposit (PD).
The Influence of Different Nb-Contents 145
Fig. 2. (a) Cross section of All-weld metal AWM, 6 layers - 2 runs each and (b)
weld metal of Plate Deposit PD, 8 layers - 2 runs each. Doted circles: Diameter of
tensile specimen acc. to DIN 32525 (thread M16)
The weld metal of PD is partially hindered in shrinkage by the plate thickness
and the underlying layers, and therefore a higher HCS can be expected
than for AWM (Fig. 2).
The fabrication of the weld metals was performed using metal-cored
wires (MCW) and FCW respectively. Cross sections of them are shown in
Fig. 3. Additional weld metals were produced with solid wire (SW),
MMA-Electrode and FCW on the market. The applied welding procedures
were GMA-welding for FCW (shielding gas 82Ar/18CO2) and for MCW
(50Ar/50He), with direct current electrode positive (DCEP). The SW Nibase
70/20 was welded using pulsed current (50Ar/50He).
Fig. 3. Cross sections of Testing wires Ni-base type 70/20 ∅ 1.2 mm: (a) Metalcored
wire MCW No. 3–10 and (b) Flux-cored wire FCW No. 11–13 in Table 3
Table 3. Nb- and C-content of the investigated welding consumables and heat input for manufacturing weld metals Ni-base 70/20
and 70/15 (Testing program I)
Consumable
type
Solid
wire
MMAElectrode
Metal-cored wires MCW Flux-cored wires FCW
Market products Testing wires Testing wires Market products
Alloy type Ni-base 70/20 Ni-base 70/20 Ni-base 70/20 Ni-base 70/20 Ni-base 70/15
Alloy No. 1 2a 2b 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18a 18b
Nb [%] in
Filling 2.5 n.k. 0.0 1.0 2.0 3.0 3.5 1.0 2.0 3.0 n.k. 2.4 n.k. n.k. 1.9
C [%] in
Filling ≤0.03 n.k. 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 n.k.
Nb/C-ratio 83 n.k. 0 20 40 20 60 30 70 35 20 40 60 n.k.
Heat input
AWM
[kJ/cm]
12.2 9.6–10.3 12 13 13 9.5 12.2 9.5 15 9.7 9.5 9.3 9.3 9.5 10.4 8.5 9.1 –
Heat input
PD [kJ/cm] 13.4 9.6–
10.5
5.6–
6.2 8.6 8.7 8.4 8.3 8.2 7.9 8.2 7.9 6.5 6.5 6.3 – – – – 13.0 7.8
n.k. not known, – no weld metal manufactured.
146 Metallurgy and Materials
The Influence of Different Nb-Contents 147
External-Loaded Hot Crack Testing: PVR-Test
For the development and quantification of hot crack susceptible fully austenitic
weld metals the controlled deformation crack test or controlled flat
tension test, PVR-test for short, was developed and employed the first time
more than two decades ago [14, 15, 16] (Fig. 4).
Fig. 4. PVR-test, scheme and test procedure to determine the critical tension speed
vcr initiating the 1st hot crack transvers to welding direction; U Magdeburg
The PVR-test belongs besides the Modified Varestraint-Transvarestraint-
Test (MVT) and Hot tension test (using the Gleeble® physical simulator)
and others, to the external loaded hot crack testing methods [17].
The PVR-test was created to evaluate the effects of the welding procedure
on hot cracking [2]. Only a single test specimen is required to determine
the HCS of a base metal and weld metal of the applied welding process.
The ordinary test procedure of the PVR-test in Fig. 4 uses flat specimens
with the dimension 40 x 10 x 300mm, clamped into a special tension fixture,
lengthened in a horizontal servo-hydraulic system of the test equipment.
The welding process with a constant welding speed is superposed by a
linearly increased tension speed vPVR in welding direction. The standard
test procedure of the PVR-test is carried out with tension rates linearly in148
Metallurgy and Materials
creasing from zero to 60 mm/min, during bead on plate TIG-welding with
Argon shielding gas, applying two types of heat input per unit length about
7 and 10 kJ/cm (Iweld=180/220 A, Uweld=12/14 V, vweld=19 cm/min) [18].
The critical tension speed vcr or critical elongation speed (CES) is the
test criterion for the PVR-test. CES corresponds to the first hot crack, detected
visually at a magnification of 40. It can be determined for each of
the hot crack types: Solidification cracks SC, liquation cracks LC and ductility
dip cracks DDC (Fig. 5).
Fig. 5. Solidification cracks SC, liquation cracks LC and Ductility dip cracks
DDC on a PVR-specimen
The success to assess all hot cracking types depends on the superposition
of the local thermal cycle beside the bead with the tension rate. In addition,
different indices for the critical tension speed vcr are possible, dependent
on the size and number of visible cracks. For example: vcr1 is the
critical tension speed for the first microscopically visible hot cracking feature.
vcr3 describes the first three hot cracks per 10mm of weld bead and
vcr9 determines the first nine hot cracks per 10mm of weld bead. The modelling
of the PVR-test has shown the correlation between test criterion, hot
cracking theory by Prokhorov, and its applicability [19].
The PVR-test I was for to quantify the HCS of MMA-Electrode Ni-base
type 70/20 ∅ 4 mm (No. 2) and the basic FCW Ni-base type 70/15 (No.
18), using two different heat inputs from the TIG-arc for remelting the
PVR-weld metal (6.8 and 9.7 kJ/cm). To evaluate the HCS of the optimized
FCW (No. 12b) in comparison to FCW market products (No. 14,
The Influence of Different Nb-Contents 149
15, 16) the PVR-test II was performed. The final PVR-test III was to compare
it’s HCS to MMA-Electrode ∅5mm (No.2’), Solid Wire (No.1), and
similar base metal alloy 600, see Table 4.
Table 4. Parameter for manufacture of PVR-specimens with different welding
consumables (PVR-test I, II, III)
PVR-Tes I PVR-Test II PVR-Test III
Consumable
type
MMAElectrode,
∅
4 mm
Flux-cored
wire, ∅
1.2 mm
Flux-cored wire, ∅
1.2 mm
MMAElectrode
∅ 5 mm
Solid
wire, ∅
1.2 mm
FCW,
∅
4 mm
Alloy type Ni-base
70/20
Ni-base
70/15 Ni-base 70/20 Ni-base 70/20
Alloy No. 2 18 16 14 12b 15 2’ 1 12b
PVRspecimen
No.
N 30 N 31 N 32 N 33 1 2 3 4 P 29 P 30 P31
Heat Input
[kJ/cm] 6.4 6.4 10.5 9.0 8.4 8.5 8.9 9.3 5.5 7.6 7.4
Number of
layers 4, á 5 runs
4, á
3
runs
4, á
4
runs
4, á 4 runs 4, á 4 runs
The standard heat input in the PVR-test was 6.8 kJ/cm, except PVRspecimen
N31 and N33 having 9.7 kJ/cm.
Evaluation of the Slag Basicity
The basicity of the solidified slag was determined by EDX-analysis of slag
micro-sections, using the modified basicity index B.I. by Bauné et al. [20]
in Eq. 1 below.
( )[ ]
[ .%]
.%
0.5 ( )
. . 0.5
2 2 3 2 2 2 5
2 2 2 2 3 2 3
at
at
SiO Al O TiO ZrO Nb O
B I CaF CaO MgO Na O K O MnO Fe O Cr O
+ × + + +
= + + + + + × + + (1)
This basicity approach is a modification of Tuliani’s expression [21],
using mole fractions (atomic %) for the oxides-concentration in the solidified
slag. Their concentration was estimated by stoichiometric calculations.
However two adaptions were carried out: For Ni-base slags Cr2O3
und Nb2O5 had to be added in Eq. 1, this was accomplished using the optical
basicity Λ of these oxides: As Cr2O3 has the same recommended optical
basicity like Fe2O3 (Λ=0.69), it was added to the transition oxides
Fe2O3 and MnO (Λ=0.95) in the dividend of Eq. 1. Nb2O5 (Λ=0.61) was
added to the amphoteric oxides Al2O3, TiO2 (Λ=0.65) and ZrO2 in the denominator
[22, 23, 24].
Table 5. Analysis of All-weld metals AWM manufactured with different welding consumables.
Ni C Nb Fe Cr Mo Si Mn Ti P S O N
1 SW 70/20a 73.0 0.006 2.72 0.37 20.2 0.00 0.11 3.15 0.33 0.005 0.003 0.010 0.019
2 MMA 70/20 68.5 0.027 2.19 3.17 18.52 1.25 0.39 5.33 0.06 0.008 0.006 0.055 0.023
3 MCW 70/20 73.3 0.015 0.04 2.42 20.48 0.09 0.41 3.20 0.01 – 0.005 0.093 0.015
4 MCW 70/20 73.0 0.017 1.04 2.45 19.60 0.09 0.38 3.38 0.02 – 0.005 0.100 0.016
5 MCW 70/20 72.0 0.020 2.01 2.04 19.90 0.10 0.38 3.45 0.03 – 0.005 0.098 0.017
6 MCW 70/20 72.4 0.092 2.24 2.62 19.40 – 0.36 3.04 – <0.005 0.005 0.033 0.019
7 MCW 70/20 71.5 0.021 3.03 2.03 19.55 0.10 0.36 3.27 0.04 – 0.005 0.086 0.018
8 MCW 70/20 70.9 0.093 3.32 2.70 19.61 – 0.37 3.01 – – 0.007 0.032 0.020
9 MCW 70/20 71.2 0.020 3.43 1.88 19.33 0.12 0.36 3.39 0.04 – 0.006 0.110 0.019
10 MCW 70/20 69.9 0.092 3.66 2.77 19.61 – 0.37 3.10 – – 0.007 0.028 0.021
11 FCW 70/20 73.6 0.022 1.00 1.97 19.17 0.08 0.34 2.57 0.08 – 0.003 0.072 0.018
12 FCW 70/20 72.7 0.024 2.00 1.65 19.29 0.08 0.35 2.79 0.09 – 0.003 0.068 0.019
13 FCW 70/20 71.2 0.027 3.14 1.66 19.66 0.09 0.34 2.84 0.10 – 0.004 0.086 0.018
14FCW 70/20 73.3 0.030 2.36 1.68 19.27 – 0.50 3.00 – <0.005 0.004 0.054 0.011
15 FCW 70/20 – 0.063 2.63 2.13 21.52 – 0.22 3.18 0.36 0.003 0.005 0.071 –
16 FCW 70/20 72.3 0.031 2.28 1.25 20.70 – 0.36 2.83 – <0.005 0.002 0.078 0.011
17 FCW 70/15 66.6 0.024 1.72 8.22 16.22 – 0.54 6.80 – – – 0.064 –
18 FCW 70/15 76.5 0.023 1.47 1.04 13.74 – 0.36 6.53 0.08 <0.005 0.004 0.049 –
– not analyzed.
a Analysis Solid wire.
150 Metallurgy and Materials
The Influence of Different Nb-Contents 151
Results and Discussion
Analysis of All-Weld metal
Except the Nb-contents of the consumables No. 3 and 4 MCW 70/20 and
No. 11 FCW 70/20, all of them are within the tolerances of Ni-base weld
metal 70/20 from SMAW-Electrode (prEN ISO 14172) and bare electrodes
and rods (DIN 1736, AWS A5.14) respectively. Only the Fe-content of
weld metal No. 18 FCW 70/15 of approx. 1 % is below the min. standard
content of 2%Nb in Ni-base 70/15 from SMAW-Electrode (DIN 1736),
see Table 5, compare Table 2.
For the optimization of the manufactured Flux-cored wires the alloy No.
12 FCW 70/20 was chosen as a basis (2 % Nb in filling). By increasing the
Nb-content to about 2.5 % and making some variations mainly in Fe, Mn
and Si-content, the required value for the fracture strain of 30% could be
reached.
Taking into consideration the requirements for the arc stability and slag
viscosity of the FCW, the final product was found to be 12b FCW 70/20,
see Table 6.
Table 6. Analysis of All-weld metal AWM manufactured with optimized Fluxcored
wires
FCW
70/20
Ni C Nb Fe Cr Si Mn P S O N
12 72.7 0.024 2.00 1.65 19.29 0.35 2.79 – 0.003 0.068 0.019
12a 73.2 0.027 2.37 2.01 19.57 0.37 2.45 <0.005 0.004 0.086 0.018
12b 72.7 0.042 2.46 1.71 19.83 0.29 3.01 <0.005 0.004 0.063 0.013
12c 71.8 0.077 2.58 3.53 19.50 0.33 2.54 <0.005 0.004 0.068 0.015
12d 72.4 0.030 2.44 2.09 18.43 0.34 2.80 <0.005 0.006 0.060 –
– not analyzed
Mechanical Values of Weld Metals
With increasing Nb-content up to 3 % in the weld metal, higher fracture
strain and tensile strength was measured for FCW and MCW as well. At
3.5 % Nb the fracture strain for weld metal of MCW becomes lower again,
see Table 7 and Fig. 6. With increasing Nb-content from 0 to 3 % in AWM
152 Metallurgy and Materials
Fig. 6. Fracture strain A5 of All-weld metal AWM and weld metal of Plate
Deposit PD, manufactured with Metal-cored wire (a) and different welding
consumables (b)
and PD of low carbon MCW a strong improvement of the fracture strain
is found (7/4 to 42/28 %).
The Influence of Different Nb-Contents 153
For weld metals with high C-content between 2.2 % and 3.6 % Nb no
significant change in the fracture strain can be observed (Fig. 6a).
With increasing Nb-content from 1 to 3 % in AWM and PD manufactured
with No. 11, 12 and 13 FCW a strong improvement of fracture strain
can be found, too. Two of five FCW from market are below the required
30 % fracture strain, whereas No. 18 FCW 70/15 seems to be susceptible
to higher welding heat input. The Solid wire and the MMA-Electrode, for
lower and higher heat input as well, give very good A5-values (Fig. 6b).
With increasing Nb-content from 1 to 3 % in the AWM and PD produced
with No. 11, 12 and 13 FCW a strong improvement of the fracture
strain can be found, too. Two of five commercial FCW are below the required
30 % fracture strain, whereas No. 18 FCW 70/15 seems to be susceptible
to higher welding heat input.
The Solid wire and the MMA-Electrode (for lower and higher heat input
as well) give very good A5-values (Fig. 6b). With some variations in the
metallurgical system of No. 12 FCW(Table 6), the fracture strain could be
improved to about 40 % (12b–12d), so the required standard value of
A5 > 30 %, could easily be fulfilled (Table 8).
Reduction of Area and Hook Cracks
The differences in the mechanical values (Table 7, Fig. 6) was due to the
appearance of Hook cracks, which have been evaluated over the reduction
of area (magnification 8 x), as suggested by Pohle [13] (Fig. 7).
The AWM and PD of MCW with high-carbon content (HC) No. 6 (6A,
6P) with 2.2 % Nb and 0.09 % C reached the highest RoA (47/39 %) and
the fewest Hook cracks (6/31) as well. HC MCW No. 8 and 10 with
3.3/3.7 % Nb and 0.09 % C give comparable good results too.
Among weld metals with low-carbon content (LC) MCW No. 9 and 7
with 3.4/3.0 % Nb and 0.02 % C give the best values for AWM (9A, 7A).
The Plate Deposit PD has the highest RoA and lowest no. of cracks for No.
7 and 5 (7P, 5P) (Fig. 7a).
The weld metals of No. 1 Solid wire type 70/20 (1A, 1P) and No. 17
FCW type 70/15 (17A) show superior ductility, i.e. RoA more that 50 %,
whereby only a few Hook cracks appeared. Astonishingly the Plate Deposit
DP of the basic No. 18 FCW (18bP) produced with ~8 kJ/cm heat input
show a RoA of 51 %, having 102 small Hook cracks.
The same PD 18aP welded with 13 kJ/cm show a distinctly lower RoA
(28 %), but lesser cracks too (33). From this it can be expected that the
weld metal is sensitive to overheating, compare Table 3.
Table 7. Mechanical values of All-weld metal AWM and weld metal of Plate Deposit PD (Testing program I)
Consumable type Solid
wire
MMAElectrode
Metal-cored wires MCW Flux-cored wires FCW
Market products Testing wires Testing wires Market products
Alloy type Ni-base 70/20 Ni-base 70/20 Ni-base 70/20 Ni-base 70/20 Ni-base 70/15
Alloy No. 1 2a 2b 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18a 18b
AWM 407 419 335 365 383 431 400 437 409 451 396 426 430 414 428 423 387 – Rp0.2
[MPa] PD 358 386 417 318 362 390 406 411 420 386 431 355 376 414 – – – – 325 345
AWM 660 676 381 446 588 686 666 707 684 712 416 616 674 655 587 635 633 – Rm
[MPa] PD 651 668 679 337 424 558 676 664 676 633 687 407 525 644 – – – – 458 560
AWM 38 37 7 10 24 35 42 36 42 30 4 25 29 40 16 28 46 –
A5 [%]
PD 46 43 44 4 8 17 26 28 21 22 22 9 16 30 – – – – 17 31
AWM 51 43 23 26 36 47 41 42 45 36 21 33 38 46 32 37 55 – RoA
[%] PD 58 49 48 17 23 29 39 32 32 28 35 18 29 39 – – – – 28 51
RT 178 129 105 86 111 94 125 78 106 71 137 146 113 142 99 146 154 –
Av[J]
-196 °C 179 124 90 104 97 83 117 69 94 60 142 134 101 134 84 140 138 –
Rp0.2 Yield Strength, Rm Tensile Strength, A5 Fracture strain, RoA Reduction of Area, Av Charpy-Impact Value.
154 Metallurgy and Materials
The Influence of Different Nb-Contents 155
Fig. 7. Number of Hook cracks against RoA [%] of All-weld metal AWM (A) and
weld metal of Plate Deposit PD (P) manufactured with Metal-cored wire (a) and
different welding consumables (b)
For FCW testing wires the RoA increases significantly with the Nbcontent
(1.0, 2.0, 3.1 %), like the fracture strain (Fig. 6). The lowest
amount of Hook cracks appeared for No. 13 All-weld metal (13A).
156 Metallurgy and Materials
The No. 2 MMA-Electrode shows also very good ductility and just a
few Hook cracks. Besides it no deterioration of the values using ~10 kJ/cm
welding heat input (2aP) compared to ~6 kJ/cm (2bP) for Plate Deposit PD
can be observed (Fig. 7b).
Table 8. Heat Input for the manufacture of AWM and PD, mechanical properties
and no. of Hook cracks of optimized FCW (Testing program II)
Consumable type Flux-cored wires FCW
Testing
wire Optimized testing wires
Alloy type Ni-base 70/20
No. 12 12a 12b 12c 12d
AWM 9.3 10.0 9.6 8.5 Heat Input 5 9.8
[kJ/cm] PD 6.5 9.5
Yield Strength AWM 426 418 416 403 423
Rp0.2 [MPa] PD 376 358 380 362 363
Tensile AWM 616 521 669 654 669
Strength Rm
[MPa] PD 525 436 641 628 648
Fracture strain AWM 25 12 43 41 39
A5 [%] PD 16 8 32 37 37
Reduction of AWM 33 10 22 41 28
Area [%] PD 29 11 31 36 40
No. of Hook AWM 40 34 4 2 1
Cracks PD 48 50 92 40 49
A satisfying optimization of the manufactured No. 12 FCW could be
reached by setting the Nb-content to about 2.5 % and the C-content in
AWM to about 0.05 %. Other important factors are the ratios of Nb/Si and
Ni/Cr/Fe, as well as the Mn- and Fe-content [25, 26].
According to the investigation of Dupont [27] a higher C-content lowers
significantly the solidification interval. Herewith the no. of Hook cracks
could be decreased strongly from 40 of No. 12 FCW to 4 cracks of No.
12b FCW at high RoA of almost 50 % (Table 8).
The AWM of No. 12a FCW was somehow out the tolerances of this
metallurgical system, i.e. slightly lower/higher C-/Si-content, compare
The Influence of Different Nb-Contents 157
Table 6, what can be seen at the lower Basicity Index (BI=0.59) which is
shown in Fig. 10.
Fig. 8. Fracture surface of tensile test specimens from AWM of No. 11 (a), 12 (b)
and 13 (c) FCW from Stereo microscope (left) and SEM-fractographs in BSE
mode (right)
158 Metallurgy and Materials
Fractography of All-Weld Metal from FCW
For the necessary development and optimization SEM-investigations of
the fracture surface from the AWM tensile test specimens No. 11, 12 and
13 FCW were performed. As the ductility and tensile strength was
improved as well (Table 7), it could be expected to find a finer subgrain
structure, i.e. smaller dendrite spacings, what could be confirmed.
The All-weld metal of No. 11 FCW (11A) shows brittle fracture behavior,
i.e. due to the intercrystalline cracks leading to fracture, a typical
fibrous structure can be seen in the stereo microscope (Fig. 8a left). In the
SEM-fractograph plane structures with partially melted zones appear
(Fig. 8a right). All-weld metals of No. 12 and 13 FCW (12A, 13A) show a
microstructural ductile fracture and the bright precipitations found in the
SEM were analyzed to be NbC (Fig. 8b and 8c).
All three weld metals show oxidic inclusions of different size and shape,
marked with “Ox.” The All-weld metal of the optimized No. 12a and 12b
FCW 70/20 show totally different fractographs, despite the similar chemical
analysis in Table 6: 12a shows partially melted zones and a crack surface
topography as well. On the contrary 12b shows a distinctive dimple
fracture (Fig. 9a and 9b).
Fig. 9. SEM-fractographs of tensile test specimens AWM FCW 12a (a) and 12b
(b) in SE mode
Table 9. Basicity Index BI of the slag, O-content and Charpy-Impact Values Av of All-weld metal AWM (Testing program I)
Solid
wire
MMAElectr.
Metal-cored wires MCW Flux-cored wires FCW
Consumable
type Market
products Testing wires Testing wires Market products
Alloy type Ni-base 70/20 MC Ni-base 70/20 FC Ni-base
70/20
FC Ni-base
70/20
FC
Ni-base
70/15
Alloy No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18a
BI slag
[at.%/at%] – n.a. – – – – – – – – 0.72 0.76 0.71 n.a. 0.91 0.81 n.a. 2.02
O-content
[ppm] 100 550 930 1000 980 330 860 320 1100 280 720 680 860 540 710a 780 640 490a
RT 178 129 105 86 111 94 125 78 106 71 137 146 113 142 99 146 154 n.a. Charpy-
Impact
Value
Av [J] -
196°C 179 124 90 104 97 83 117 69 94 60 142 134 101 134 84 140 138 n.a.
n.a. not analyzed, – no slag.
a Analysis from other experiment.
The Influence of Different Nb-Contents 159
160 Metallurgy and Materials
Fig. 10. Correlation of Basicity Index BI to the O-content in All-weld metal and
the Hook cracks for optimized No. 12a - 12d FCW
Slag Basicity and O-Content
The basicity of the solidified slag was determined by the modified basicity
index by Bauné et al. [20, 28] in Eq. 1. Herewith a good correlation could
be achieved with the O-content and the impact values for the Testing wires
FCW No. 11, 12, 13 (Table 9). By this means the relatively small variations
of metals (1, 2, 3 Nb) in the filling could be investigated.
For the optimized FCW No. 12a-12d also a good correlation was found:
Decreasing O-content in All-weld metal with higher BI of the solidified
slag and decreasing Hook cracks (Fig. 10).
PVR-Test Results
The results of PVR-test I in Fig. 11 show a reduction in the critical elongation
speed (CES), i.e. higher HCS for Solidification cracks SC and Liquation
cracks LC as well, when using higher heat input of the TIG-arc. That
applies to No. 2 MMA-Electrode Ni-base type 70/20 ∅ 4 mm and the
The Influence of Different Nb-Contents 161
Fig. 11. Evaluation of the PVR-test I: MMA-Electrode ∅4mm Ni-base 70/20 and
basic FCW Ni-base 70/15 show comparable low Hot crack susceptibility HCS to
Solidification and Liquation cracks SC/LC
basic No. 18 FCW Ni-base type 70/15, too. The reason for this is the
bigger size of the mushy zone at 9.7 kJ/cm [19]. Apart from that the weld
metals have comparable low HCS, like it was estimated in the RoA-Hook-
Fig. 12. Evaluation of the PVR-test II: The optimized 12b FCW 70/20 (3) show
comparable HCS to Solidification, Liquation and Ductility Dip Cracks SC / LC /
DDC (TIG arc 6.8 kJ/cm) to FCW on the market
162 Metallurgy and Materials
cracks diagram, Fig. 7, 2aP/2bP/18bP. Ductility Dip cracks DDC could not
be found in these weld metals.
The optimized weld metal of FCW No. 12b show comparable CES to
the FCW 70/20 market products for each of the hot crack types (Fig. 12),
but it is quite lower than the CES of No. 2 MMA-Electrode ∅ 4 mm and
the basic No. 18 FCW 70/15 in Fig. 11. A further comparison shows, that
the CES of 12b is far away from that of the No. 1 Solid wire. It is just
comparable to No. 2 MMA-Electrode ∅5mm, concerning the vSC / vMicro-SC
criterion. Still these values are quite below the similar base metal alloy 600
(Fig. 13).
Fig. 13. Evaluation of the PVR-test III: MMA, SW, FCW
and Base metal (TIG arc 6.8 kJ/cm)
Conclusions
Nb-additions to weld metals Nibas-70/20 can reduce the hot crack susceptibility
HCS strongly, because of the formation of NbC, what makes finer
subgrains in the weld metal. By that also the fracture strain and the tensile
strength is improved.
The tested weld metals of Plate Deposit PD show mostly more Hook
cracks than All-weld metal AWM. Thus it should be a useful method for
pre-selecting a weld metal concerning its HCS. For the optimization of
slag bearing consumables the determination of the Basicity Index of the
solidified slag can be a support to find the right metallurgical system.
The Influence of Different Nb-Contents 163
Higher heat input in the manufacture of the weld metals as well as for
the remelting in the PVR-test is increasing the HCS strongly. The HCS of
weld metals from FCW 70/20 in the PVR-test is distinctly higher than for
weld metals of MMA-Electrode and Solid wire. The HCS of weld metals
Ni-base 70/15 for the investigated products was lower than for 70/20.
In general the HCS of similar base metal alloy 600 is still much lower
than of the weld metals.
Acknowledgments
The author would like to thank Prof. Herold, Mrs. Streitenberger and Mr.
Pchennikov from the Institute of Joining and Beam Technology at TU
Magdeburg for the performing and evaluation of the PVR-tests. Likewise
Mr. Tösch and Mr. Klagges, Böhler Schweißtechnik Austria and Mr.
Heinemann, UTP-Schweißmaterial for their expert advices, as well as
Prof. Cerjak, head of the institute, for the possibility to carry out this investigation.
References
1. Heinemann J, Kilian R Personal communication
2. Folkhard E (1984) Welding Metallurgy of Stainless Steels. Springer Wien
New York
3. Yamanaka K et al. (1993) Development of High Corrosion Resistant Alloy
600 with Niobium Addition. Sumitomo Search 51: 1–15
4. Lee HT, Kuo TY (1999) Effects of niobium on microstructure, mechanical
properties, and corrosion behavior in weldments of alloy 690. Science and
Technology of Welding and Joining, Vol 4, No 4: 246–256
5. DIN 1736 (Aug 1985) Schweißzusätze für Nickel- und Nickellegierungen
6. DIN EN ISO 18274 (Aug 2000) Draht- und Bandelektroden, Drähte und Stäbe
zum Lichtbogenschweißen für Nickel und Nickellegierungen
7. AWS A5.14 Specification for Nickel and Nickel-Alloy Bare Welding Electrodes
and Rods
8. DIN EN ISO 14172 (Mai 2001) Umhüllte Stabelektroden zum Lichtbogenschweißen
von Nickel und Nickellegierungen
9. AWS A5.11/A5.11M-97 Specification for Nickel and Nickel-Alloy Welding
Electrodes for Shielded Metal Arc Welding
10. Deutscher Verband für Schweißtechnik e.V. (Sept 1990) Heißrissprüfverfahren
mit selbstbeanspruchenden Proben. Merkblatt DVS 1004 Teil 3. DVS
Verlag Düsseldorf
164 Metallurgy and Materials
11. Kauhausen E, Kaesmacher P (1955) Bestimmung der Warmrißbeständigkeit
von Schweißelektroden. Schweißen und Schneiden 7, H 2: 400–404
12. Düchting W (1978) Einige Betrachtungen zur Heißrißprüfung von Schweißzu-
satzwerkstoffen. Schweißen und Schneiden 30, H 12: 498–502
13. Pohle C (1985) Möglichkeiten für die Beurteilung der Heißrissanfälligkeit von
austenitischem Schweißgut. Schweißen und Schneiden 37, H 2: 55–60
14. Folkhard E et al (Nov 1977) Der PVR-Test, ein neues Verfahren zur Ermittlung
der Rißsicherheit von Schweißwerkstoffen mit hoher quantitativer Aussagekraft.
Sonderdruck aus Jubiläumsschrift "50 Jahre BÖHLERSchweißtechnik"
15. Rabensteiner G, Tösch J, Schabereiter H (Jan 1983) Hot Cracking Problems
in Different Fully Austenitic Weld Metals. Welding Journal: 21-s–27-s
16. Kobayashi T et al (1986) Hot Cracking Susceptibility Evaluation of Austenitic
Stainless Steel Weld Metals. The Japan Welding Engineering Society, IIW
Doc II-1068-86/IX-1395-86
17. Deutscher Verband für Schweißtechnik eV (Nov 1996) Heißrissprüfverfahren
mit fremdbeanspruchten Proben. Merkblatt DVS 1004-2. DVS Verlag
Düsseldorf
18. Herold H, Pchennikov A, Streitenberger M (2003) Current problems in hot
cracking research described on the example of PVR test. Materials Science
Forum Vols 426–423: 4093–4098
19. Herold H, Streitenberger M, Pchennikov A (2001) Hot Cracking Theory by
Prokhorov and Modelling of the PVR-test. Welding in the World 45, 3/4: 17–
22
20. Baune E, Bonnet C, Liu S (March 2000) Reconsidering the Basicity of a
FCAW Consumable-Part 1: Solidified Slag Composition of a FCAW Consumable
as a Basicity Indicator. Welding Journal: 57-s–65-s
21. Tuliani SS, Boniszewski T, Eaton NF (1969) Notch Toughness of commercial
submerged arc weld metal. Welding and Metal Fabrication 37, 8: 327–339
22. Mills KC (1995) Slag Atlas. VDEh 2nd Edition, Verlag Stahleisen GmbH,
Düsseldorf
23. Sommerville ID, Yang Y (2001) Basicity of metallurgical slags. The AusIMM
Proc 306, 1: 71–77
24. Lebouteiller A, Courtine P (1998) Improvement of a Bulk Optical Basicity
Table for Oxidic Systems. Journal of Solid State Chemistry 137: 94–103
25. Pease G R (1957) Metallkundliche Vorgänge beim Schweissen von Nickel
und Nickellegierungen. International Nickel, Zürich 1957
26. Anik S, Dorn L (1983) Metallphysikalische Vorgänge beim Schweißen von
Nickelwerkstoffen - Einfluss der Werkstoffzusammensetzung. Schweißen und
Schneiden 35: 445–450
27. DuPont N, Robino CV, Marder AR (1998) Solidification and Weldability of
Nb-Bearing Superalloys. Welding Journal 10 (1998)
28. Baune E, Bonnet C, Liu S (March 2000) Reconsidering the Basicity of a
FCAW Consumable – Part 2: Verification of the Flux/Slag Analysis Methodology
for Weld Metal Oxygen Control. Welding Journal: 66-s–71-s
Hot Cracks as Stress Corrosion Cracking
Initiation Sites in Laser Welded
Corrosion Resistant Alloys
K. Stelling, Th. Böllinghaus, M. Wolf, A. Schöler, A. and A. Burkert,
B. Isecke
Federal Institute for Materials Research and Testing, Berlin, Germany
Abstract
Although hot cracks at weld surfaces of corrosion resistant alloys (CRAs)
might provide access to respective aggressive media and thus, might provide
the conditions for a local potential and pH drop, the interaction between
hot cracking and corrosion phenomena has not been investigated up
to the present. This particularly concerns the initiation of stress corrosion
cracking inside such crevices.
As a first approach to clarify the influence of hot cracks on corrosion resistance,
considering particularly stress corrosion cracking, hot crack afflicted
laser welds of three different CRAs have been subjected to a series
of different corrosion tests.
Besides the fact that the Drop Evaporation (DE) Test represents a more
realistic procedure than standard immersion tests it turned out that hot
cracks can obviously represent an initiation site of stress corrosion cracking
in the austenitic stainless steel AISI 309 (German No. 1.4828). The
susceptibility of such hot crack afflicted laser welds to stress corrosion
cracking (SCC) significantly depends on temperature and chloride concentration
of the test environment In comparison, the high nitrogen austenitic
stainless steel AISI S 34565 (German No. 1.4565) did not exhibit any
stress corrosion cracking associated with hot cracking. This material exhibited
pitting corrosion and stress corrosion only in the base material. The
Alloy 50 (German No. 2.4850), a Nickel based material, was susceptible to
pitting corrosion, but showed no tendency towards stress corrosion cracking
at all.
166 Metallurgy and Materials
Introduction
Laser welding is increasingly applied to fabrication joining of components
made of corrosion resistant alloys (CRAs), for instance in pipe manufacturing
for offshore or chemical industry. As a particular advantage of the
low heat input associated with laser welding, the risk of heat-induced precipitation
and respective depletion of alloying elements is reduced at considerably
lower residual stresses as compared to arc welding processes. On
the other hand, most CRAs exhibit an austenitic microstructure providing
large solidification intervals with respective phase segregation and, consequently,
the conditions for the appearance of hot cracking during welding.
Fig. 1. Change of the primary solidification mode in austenitic stainless steels near
the eutectic rim
As a special effect it has to be considered that a group of widely applied
stainless steels with a chemical composition located at the primary δ-ferrite
side of the respective constitutional diagram (Fig. 1a) and which are considered
as non-hot cracking sensitive, might undergo a primary solidification
mode change to austenite caused by kinetic undercooling and also
caused by epitaxial effects during rapid welding procedures (Fig. 1b). It
has widely been proven in previous studies that laser welding of these materials
increases the risk of hot cracking [1, 2].
Hot Cracks as Stress Corrosion Cracking Initiation Sites 167
Fig. 2. Metastable ferritic crystallization of equiaxed dendritic morphologies in an
austenitic stainless steel [2] produced by the following laser welding sequence:
1 –Welding with high nitrogen additions in the laser process gas to shift the solidification
mode from primary ferrite to primary austenite, 2 – Remelting the primary
austenitic solidified weld at high welding speeds to shift the primary solidification
mode back to the ferrite side by kinetic undercooling
As a particular item, by respective nitrogen additions evidence has been
produced for the first time that rapid welding processes might not only
lead to a primary solidification
mode change from ferrite to austenite, but also vice versa from austenite to
ferrite (Fig. 2), up to the present only known from non-chambered undercooling
and quenching experiments, respectively [2, 3, 4].
Regarding the service behavior of components made of CRAs it has to
be made sure that welds do not represent preferential sites for corrosion attack.
Weld defects like undercut notches have thus to be avoided. Hot
cracks as particular weld imperfections in austenitic CRAs cannot always
168 Metallurgy and Materials
be detected during non-destructive testing before a welded component is
applied to service operation.
There are only a few studies on the service behavior of hot crack
afflicted welded components and, in particular, no investigations have
been undertaken as to whether hot cracks might represent preferential sites
for corrosion attack in welded CRAs, except a brief study about intergranular
corrosion starting at hot cracks in nickel-base alloys [5].
However, since austenitic CRA components are most frequently subjected
to chloride environments, stress corrosion cracking represents the
most safety-relevant failure type. Hot cracks opened to the weld surface
might particularly provide the conditions for local pH and potential drops
and thus, might represent respective stress corrosion cracking initiation
sites.
The present study has been performed, in order to gain some insight into
such phenomena. Hot cracks have been introduced into laser welds of
three different types of austenitic CRAs using a special designed test
facility. After loading by a screw bolt the specimens have subsequently
been subjected magnesium chloride immersion tests at temperatures of up
to 140 °C. In order to simulate high salt concentrations as they prevail in
the splash water zone in marine environments, the more realistic Drop
Evaporation (DE) Test has also been carried out at medium temperatures
of up to 80 °C. Additionally, potentiodynamic polarization experiments
were performed in order to investigate the critical corrosion potentials and
repassivation characteristics.
Materials and Experimental
As an austenitic stainless steel known to have a tendency to change the solidification
mode during rapid laser welding, an alloy AISI 309 (1.4828)
was selected for the present study. Additionally, a high nitrogen austenitic
stainless steel AISI S 34565 (1.4565) and a nickel-base material trade
named Alloy 50 (2.4850) were tested, which are known to be susceptible
to hot cracking for any fusion welding technique.
All experiments were carried out using sheet material of 2 mm and
3 mm thickness, respectively. The chemical compositions of the materials
were determined by spectroscopic analysis. They are summarized in the
Tables 1 to 3.
Welding was carried out using a CO2-laser with a maximum power of 6
kW. The hot cracks were produced by the Laser Specific Hot Cracking
(LSHC) Test, a BAM-designed reference procedure (Fig. 3).
Hot Cracks as Stress Corrosion Cracking Initiation Sites 169
Table 1. Chemical composition of the steel AISI 309 in wt%
Cr Ni Mo N C S P Mn Nb Si
19.92 11.08 0.273 0.033 0.052 0.003 0.025 0.693 0.011 1.97
Table 2. Chemical composition of the steel AISI S 34565 in wt% (Fe bal.)
Cr Ni Mo N C S P Mn Nb Si Ti
23.4 17.1 4.2 0.46 0.029 0.003 0.021 5.6 0.041 0.148 0.017
Table 3. Chemical composition of the Alloy 50 in wt% (Ni bal. to ca. 50 wt%)
Cr Fe Mo C S P Mn Nb Si Ti
19.2 14.71 10.33 0.007 0.002 0.008 0.341 0.294 0.070 0.014
By wrapping a thin plate of the respective material around a 72 mm diameter
mandrel during laser welding, hot cracks were generated by mechanical
straining of the specimen during the solidification process.
Fig. 3. Illustration of the Laser Specific Hot Cracking (LSHC) Test
The hot cracking resistance of different materials during laser welding is
ranked by plots of the applied strain versus laser power, for instance [4].
Hot cracks can reproducibly be introduced into the laser welds by the
LSHC test, if the welding parameters, i. e. welding speed vs, laser power
PLaser, and focal point position are adjusted to the test parameters, i.e. deformation
speed vd, total strain, and distance of the weld pool from the
bending line.
Four different specimen conditions were differentiated and are summarized
for the various materials in Table 4.
170 Metallurgy and Materials
Table 4. Tested materials and respective specimens
Specimen AISI 309 AISI S 34565 Alloy 50
State 0 – not welded,
plane sheet material
X X X
State 1 – welded,
without any hot cracks
X X X
State 2 – welded, with numerous
transverse cracks in the weld metal
X
State 3 – welded, with a longitudinal
crack in the weld centerline
X X X
Hot cracks produced in the AISI 309 steel by the primary solidification
mode change are shown in Fig. 4. The different cracking characteristics
occurred as a consequence of variations in the deformation speed vd during
the hot cracking test. Table 5 summarizes the different parameters applied
to generate reproducibly longitudinal hot cracks.
Transverse hot cracks could only be introduced in the AISI 309 steel
and only a few specimens of this type (State 2) have been investigated for
comparison.
After the hot cracking test procedure the specimens were pickled and ultrasonically
cleaned. For the corrosion tests, the pre-bent LSHC Test
specimens were loaded with the aid of a screw bolt. Thus, the surface was
exposed to tensile stress (in addition to the residual stresses caused by the
welding process).
Fig. 4. Hot cracks in AISI 309 laser welds (a) State 2, top view (vd = 9 mm/min,
vs = 5000 mm/min, PLaser = 2.6 kW), (b) State 3, top view (vd = 22.5 mm/min, vs =
5000 mm/min, PLaser = 2.6 kW)
Hot Cracks as Stress Corrosion Cracking Initiation Sites 171
Table 5. Test parameters for introducing hot cracks of State 3
Material
Welding
speed vs
[mm/min]
Laser
power
PLaser [kW]
Deformation
speed vd
[mm/min]
Total
strain
[%]
Distance laser
beam – bending
line [mm]
AISI 309 5000 3.0 15 4.0 0.98
AISI S 34565 5000 2.5 40 2.7 1.20
Alloy 50 2000 2.5 52 4.0 1.55
Fig. 5a illustrates the dimensions of the U-bend specimens while Fig. 5b
shows their exposition to the Drop Evaporation Test. To ensure complete
passivation of the materials, the specimens were exposed to air for one
week before starting the corrosion tests.
Ahead of the stress corrosion cracking tests, potentiodynamic polarization
experiments in artificial seawater were performed using a scan rate of
0.5 mV/s, in order to investigate whether the hot crack afflicted welds exhibit
different pitting and repassivation characteristics in comparison to the
non-cracked welds and the base materials. To determine the respective
critical pitting and crevice corrosion potentials, the experiments were carried
out three times for each specimen condition.
Fig. 5. Stress corrosion cracking test specimens, (a) Specimen dimensions,
(b) Exposure to the Drop Evaporation Test
For the Immersion Tests, the bend specimens were exposed to a MgCl2-
solution by covering the welded area with 10 mm of the electrolyte. The
concentrations of the MgCl2-solutions varied from 10 % to 40 %, the applied
testing temperatures ranged from 80 °C to 140 °C. The test period
was also varied from 720 h up to 1700 h depending on the tested material.
For the Drop Evaporation Test , the U-bend specimens have been
mounted on a support inside a closed thermostatic box above a saturated
172 Metallurgy and Materials
MgCl2-solution. Thus, the desired temperature and constant air humidity of
approximately 35 % was kept to ensure a constant atmosphere. Three
drops of a saturated MgCl2-solution were applied to every respective Ubend
specimen (Fig. 5b).
Due to the expected different corrosion resistance, the three materials
were tested at different temperatures, i. e. the steels AISI 309 and AISI S
34565 were exposed at 30 °C and 50 °C, respectively, while the Alloy 50
was tested at 70 °C. After every 1000 h period, one U-bend specimen of
each alloy was taken out of the test environment. The longest test period
was 5000 h in each case.
After the Immersion Test and the DE Test, the surface of the specimens
was inspected using a stereo microscope. Subsequent, cracked samples
were also subjected to metallographic and SEM investigations.
Results and Discussion
Potentiodynamic Tests
For a first insight into the corrosion behavior of the base materials, potentiodynamic
reference measurements were carried out with the steel
AISI 309 in comparison to the Alloy 50 at different test temperatures in artificial
seawater.
Fig. 6. Current density-potential plots with dU/dt = 0.5 mV/s, Steel AISI 309, base
material, artificial seawater
Hot Cracks as Stress Corrosion Cracking Initiation Sites 173
Fig. 7. Current density-potential plots with dU/dt = 0.5 mV/s, Alloy 50, base material,
artificial seawater and borate buffer solution
As shown by the two characteristic current density potential plots in Fig.
6, the critical potential of the steel AISI 309 is reduced from values in the
range between 290 mV and 320 mV to values of about 70 mV by increasing
the temperature from 20 °C to 60 °C.
A comparison of Fig. 7 with Fig. 6 clearly shows that the critical potential
of Alloy 50 is not as temperature dependent as that of the austenitic
stainless steel. At 20 °C, the critical potential ranged at about 1000 mV,
whereas from 40 °C to 80 °C the value leveled at about 900 mV. At 20 °C
in a borate buffer solution (pH 6.4), the Alloy 50 showed an even higher
corrosion resistance. Some temporary increase of the current density might
be attributed to growth and reshape mechanisms inside the passive layer.
Additional potentiodynamic experiments were performed with various
specimens of the same materials including reversed polarization in order to
evaluate the repassivation behavior of hot crack afflicted welds as compared
to the crack-free welds and the base materials.
The current density-potential plots in Fig. 8 exhibit no significant differences
regarding the critical potential of the steel AISI 309 in all four states.
The critical values still ranged between 270 mV and 320 mV. However, a
significant difference was found with the reversed polarization.
Whereas the specimens of State 0 and State 1 showed a similar repassivation
behavior with maximum current densities of about 100 μA/cm²,
repassivation was retarded at the specimen of State 2 where the maximum
current density was up to 170 μA/cm². An even higher current density of
320 μA/cm² can be observed for the specimen with a longitudinal solidification
crack (State 3).
174 Metallurgy and Materials
Fig. 8. Current density-potential plots of different specimen types with
dU/dt = 0.5 mV/s, Steel AISI 309, base material, artificial seawater
This indicates that an increasing internal surface of hot cracks might reduce
the repassivation capacity of the laser welded steel AISI 309 in aqueous
environments.
Fig. 9 shows the current density-potential curves of similar experiments
carried out with Alloy 50 at 60 °C, due to its already observed higher pitting
corrosion resistance. With reversed polarization, a negative hysteresis
was measured in every case which generally indicates a remarkably good
repassivation behavior of this material.
Fig. 9. Current density-potential plots of different specimen types with
dU/dt = 0.5 mV/s, Alloy 50, base material, artificial seawater
Hot Cracks as Stress Corrosion Cracking Initiation Sites 175
A slight difference in the repassivation behavior might be deduced from
the curves between the base material and the welded specimens. But no
difference of the current density-potential plot of the State 3 specimen
could be observed as compared to the crack free weld. Thus, hot cracking
does not seem to have an influence on the local corrosion and repassivation
behavior of this alloy.
Immersion Tests
The results from the Immersion Tests are summarized in Table 6. The high
nitrogen steel AISI S 34565 showed no corrosion attack in the 20 % MgCl2
solution at temperatures up to 110 °C. Even under aggravated conditions
(1700 h in 40 % MgCl2-solution at 140 °C test temperature), no indications
of stress corrosion cracking were found in the Alloy 50 specimens. Only
sporadically, pitting occurred in the vicinity of the weld. However, the
steel AISI 309 definitely exhibited stress corrosion cracking in the
Immersion Tests.
Table 6. Results of the Immersion Tests
Material
Temperature
range
[°C]
MgCl2 concentration
[%]
Test duration
[h] Results
AISI 309 80–120 10–40 720 Various SCC
AISI S
34565 80–110 20 1300 No corrosion
attack
Alloy 50 90–140 10–40 720 and 1700 No SCC
but pitting
As shown by the cracking tendency plot for the welded specimens of
this material in Fig. 10, stress corrosion cracking generally appeared at
temperatures above 80 °C and at MgCl2 concentration above 25 %. It also
turned out that longitudinal hot cracks in the laser welded alloy AISI 309
represent a preferential site for stress corrosion cracking. At 80 °C in the
MgCl2 solution, stress corrosion cracking started at hot cracks and increasing
the temperature above 100 °C in the 30 % MgCl2 solution caused complete
cracking of the specimens (Fig. 10).
176 Metallurgy and Materials
Fig. 10. Boundary between completely cracked specimens with the respective
failure time and specimens with beginning stress corrosion cracking in the weld,
AISI 309, State 3, maximum period of the Immersion Test: 720 h
At test temperatures of 120 °C and in 40 %-MgCl2-solution, failure was
observed already after 80 and 90 h, respectively.
The micrographs in Fig. 11 show two cross sections of two AISI 309
specimens, State 3, which exhibit stress corrosion cracking without complete
fracture of the specimen. This confirms that the stress corrosion
cracking process is initiated at the tip of respective hot cracks.
It also seems that the hot cracks must have certain dimensions in width
and in depth to provide access of the electrolyte and to establish aggressive
Fig. 11. Stress corrosion cracking of AISI 309 State 3 specimens after the
Immersion Test (a) at 80 °C in 30 % MgCl2, (b) at 105 °C in 15 % MgCl2,
after a test period of 720 h
Hot Cracks as Stress Corrosion Cracking Initiation Sites 177
crevice like conditions in terms of a pH and potential reduction.
Also the direction of the hot cracks towards the tensile stress orientation
seems to exert an influence on the stress corrosion behavior. This assumption
is substantiated by the fact that State 2 specimens did not exhibit stress
corrosion cracking in the weld, but in the base material at the same exposure
conditions as applied to the respective State 3 specimens.
Stress corrosion cracking initiated at longitudinal hot cracks predominantly
appears in a transgranular mode. This has been confirmed by respective
SEM investigations of completely cracked specimens (Fig. 12).
The fracture topography of the hot crack tip exhibits smooth molten dendrite
surfaces and can clearly be distinguished from the transgranular
cleavage mode of stress corrosion cracking.
Fig. 12. Fracture topography at the hot crack tip of completely cracked U-bend
specimens, AISI 309, State 3, (a) in 41 % MgCl2 at 120 °C, time to failure 80 h
(b) in 40 % MgCl2 at 120 °C, time to failure 100 h
Drop Evaporation Test
As confirmed by the results listed in Table 7, also in the Drop Evaporation
Test stress corrosion cracking of the steel AISI 309 was found which had
been initiated at hot cracks in State 3 specimens. Pitting was observed on
all specimens. Stress corrosion cracking also occurred in the hot crack-free
welded specimens. But in contrast to the State 3 specimens it was localized
in the base material (Fig. 13a).
Thus, the welds themselves do not represent preferential sites for stress
corrosion cracking. In other words, hot cracks have to be present in such
welds to initiate stress corrosion cracking. In this context, it can only be
emphasized that although pitting occurred in the vicinity of the hot crack
178 Metallurgy and Materials
Table 7. Results of the Drop Evaporation Tests with AISI 309 specimens
Exposure
time at 30 °C
AISI 309,
State 0
AISI 309,
State 1
AISI 309,
State 3
1000 h No corrosion
attack
Pitting in the vicinity
of the weld
Pitting in the vicinity of
the weld
2000 h Pitting Pitting in the vicinity
of the weld No corrosion attack
3000 h No corrosion
attack
Pitting in the vicinity
of the weld
SCC initiated at hot
cracks, pitting in the vicinity
of the weld
4000 h Pitting No corrosion attack Pitting in the weld and in
the vicinity of the weld
5000 h Pitting SCC in the base
material
SCC initiated at hot
cracks, pitting in the vicinity
of the weld
afflicted laser welds, stress corrosion cracking was initiated inside the longitudinal
hot cracks, as shown in Fig. 13b. Regarding this material, the
electrochemical conditions inside hot cracks are more critical than inside
the pits which gives further evidence of the severity of hot cracks acting as
stress corrosion cracking initiation sites.
In contrast to the Immersion Tests, the AISI S 34565 specimens (Table
8) exhibited pitting at 50 °C in the DE Test with nearly no exception. The
pits occurred in the base material and also in the vicinity of the weld.
Fig. 13. Stress corrosion cracking of AISI 309 in the Drop Evaporation Test
(a) State 1, exposure time: 5000 h, (b) State 3, exposure time: 3000 h
Hot Cracks as Stress Corrosion Cracking Initiation Sites 179
Table 8. Results of the DE Test with AISI S 34565 specimens
Exposure
time at 50 °C
AISI S 34565,
State 0
AISI S 34565,
State 1
AISI S 34565,
State 3
1000 h Pitting Pitting in the vicinity
of the weld No corrosion attack
2000 h Pitting Pitting in the vicinity
of the weld
Pitting in the vicinity
of the weld
3000 h Pitting
Pitting in the vicinity
of the weld and
in the weld
Pitting in the vicinity
of the weld and
in the weld
4000 h SCC initiated at pits
in the base material
SCC initiated at pits
in the base material
Pitting in the vicinity
of the weld and
in the weld
5000 h SCC initiated at pits
in the base material
SCC initiated at pits
in the base material
SCC initiated at pits
in the base material
Fig. 14a shows exemplarily a strong localized corrosion attack in the
weld and in the base material. Stress corrosion cracking initiated at such
pits (Fig. 14b) was observed in the specimens of State 0 and State 1 after
3000 h and 4000 h, whereas the specimen of State 3 exhibited stress corrosion
cracking only after a 5000 h period. This already indicates that, longitudinal
hot cracks do not represent a preferential site for stress corrosion
cracking in these welds as compared to such pits, in contrast to the AISI
309 steel. Stress corrosion cracking usually appeared at pits in the base
material near the weld region.
Fig. 14. Alloy AISI S 34565, State 3, subjected to the DE-Test (a) pitting attack in
the weld and in the base material, exposure time: 3000 h, (b) pitting induced stress
corrosion cracking in the base material, exposure time: 4000 h
180 Metallurgy and Materials
Table 9. Results of the DE Test with Alloy 50 specimens
Exposure
time at 70 °C
Alloy 50,
State 0
Alloy 50,
State 1
Alloy 50,
State 3
1000 h No corrosion attack No corrosion attack No corrosion attack
2000 h No corrosion attack No corrosion attack No corrosion attack
3000 h Pitting Strong pitting in the
weld
Slight corrosion attack
in the hot crack
4000 h Pitting Pitting in and in the
vicinity of the weld
Slight corrosion attack
in the hot crack
5000 h Pitting Strong pitting in the
weld
Slight corrosion attack
in and in the
vicinity of the weld
The U-bend specimens of the Alloy 50 exposed to the DE Test at 70 °C
exhibited no stress corrosion cracking in the base material and also not in
the weld region (Table 9). A slight brown coloration of the drops on the
welded specimens of State 1 and State 3 was observed after exposure times
of 2000 h, and after 3000 h, an accumulation of corrosion products was
detected even in the base material specimens.
The pitting attack appeared stronger in the crack-free welded specimens
and indicates that the hot cracks might represent a preferential site for
localized corrosion in the alloy.
Fig. 15. U-bend specimens, Alloy 50 (a) State 1, localized corrosion attack of the
weld and the base material, exposure time: 3000 h,
(b) State 3, only slight corrosion attack, exposure time: 5000 h
Hot Cracks as Stress Corrosion Cracking Initiation Sites 181
Some evidence to this is given by the observation of a slight corrosion
attack in the longitudinal hot cracks of specimens subjected for 3000 h and
4000 h to the DE Test. Fig. 15a shows a severe localized corrosion in the
weld and in the base material of a State 1 specimen after 3000 h. In the
specimens of State 3, only slight corrosion attack was found even after a
5000 h test period (Fig. 15b).
Conclusions and Perspectives
From the first approach to elucidate the interaction between hot cracking
and localized corrosion, in particular, stress corrosion cracking, by using
laser welded bend specimens in chloride environments, the following conclusions
can be drawn:
1. In conjunction with respective lower repassivation capabilities, an enhanced
susceptibility to stress corrosion cracking of hot crack afflicted
AISI 309 laser welds was evidently shown by the Immersion Test in
MgCl2-solution and confirmed by the more realistic Drop Evaporation
Test. Especially longitudinal hot cracks represent an initiation site for
subsequent stress corrosion cracking, whereas numerous small hot
cracks perpendicular to the welding direction do not seem to affect the
susceptibility to stress corrosion cracking. The geometry as well as the
orientation of the hot cracks to the stress state of the specimen thus significantly
affect the initiation of stress corrosion cracking.
2. The influence of hot cracks on the corrosion resistance and, in particular,
on the susceptibility to stress corrosion cracking strongly depends
on the chemical composition, since the higher alloyed materials
AISI S 34565 and Alloy 50 did not exhibited any stress corrosion cracking
initiated at hot cracks. In contrast to the steel AISI 309, pits in the
steel AISI 34565 represent a more favorable site for stress corrosion
cracking than hot cracks. Also the Nickel base Alloy 50 partially
showed severe pitting in the Drop Evaporation Test, but no stress corrosion
cracking at all.
As a future perspective, also microstructural effects have to be investigated
to clarify why hot cracks represent preferential stress corrosion
cracking initiation sites in the metastable primary austenitic solidified steel
AISI 309, but not in the two stable alloys. Further research work will also
be performed to identify precisely the stresses and strains at the tip of hot
cracks and how hot cracking might affect subsequent stress corrosion
cracking in real restrained component welds.
182 Metallurgy and Materials
References
1. Schobbert H, Böllinghaus Th, Wolf M (2003) Hot Cracking Resistance of Laser
and Hybrid Welded Austenitic Stainless Steels. In: 6th International Trends
in Welding Research Conference Proceedings, Pine Mountain, GA, ASM International,
pp 76–81
2. Schobbert H, Wolf M, Böllinghaus Th (2004) Metastable Primary Solidification
Modes in the Fe-Cr-Ni System During Laser Welding. In: Solidification
and Crystallisation, Dieter M. Herlach, pp 216–226
3. Volkmann T, Löser W, Herlach DM (1997) Nucleation and Phase Selection in
Undercooled Fe-Cr-Ni Melts: Part II. Containerless Solidification Experiments.
Metall Mat Trans A 28A: 467–469
4. Böllinghaus Th, Schobbert H (2003) Laser Weld Metallurgy of Austenitic
Stainless Steels. In: Conference Papers, Stainless Steel World 2003, Maastricht,
pp 243–249
5. Herold H, Streitenberger M, Spieler S (1999) Verhalten von Heißrissen unter
Korrosionsbeanspruchung. DVS-Bericht 204, pp 66–72
III Modeling and Simulation
Simulating and Predicting Weld Solidification
Cracks
Y. Wei1, Z. Dong1, R. Liu2, Z. Dong2, Y. Pan2
1 State Key Laboratory of Advanced Welding Technology Production,
Harbin Institute of Technology, Harbin, China
2 Harbin Welding Institute, Harbin, China
Abstract
Weld solidification cracks result from the competition between the material
resistance and driving force to the cracks. The material resistance,
namely the ductility curve in BTR, was obtained by in-situ observation and
measurement of local strain at the trail of weld pool using a CCD-camera
in the Trans-Varestraint Test. The driving force, i.e. mechanical strains
against temperatures at the trail of a weld pool, was modeled with the finite
element method (FEM) in three steps. Firstly, thermal distributions in
both 3 mm and 10 mm thickness welds were modeled by two-dimensional
and three-dimensional thermal models, respectively. Secondly, the
strain/stress distributions arising during welding of 3 mm stainless steels
were simulated by a two-dimensional model on the basis of the simulated
thermal distributions. Thirdly, the driving force behind weld solidification
cracking was determined from the simulated thermal cycle (temperature
against time) and from the mechanical strain against time in the weld center
line. Furthermore, a computer system was developed on the basis of
two-dimensional simulation which provides a simple way of conducting
complex work for simulating and predicting weld solidification cracking.
Introduction
The weld metal solidification cracking occurs when the mechanical driving
force exceeds the material resistance. The fundamental parameter representing
the material resistance is its ductility curve in the brittle temperature
range (BTR) in the course of weld metal solidification, while the
186 Modeling and Simulation
mechanical strain accumulated in the BTR is considered as the mechanical
driving force. To obtain the ductility curve in the BTR, a variety of
weldability tests have been advised [1–5]. Among them, the Varestraint-
Test and the Trans-Varestraint-Test have been widely used to evaluate
quantitatively the solidification cracking susceptibility of different materials
by measuring their CTS (Critical Strain rate for Temperature drop),
BTR, the plastic deformation ability – the solidification ductility curves in
BTR and some other quantitative parameters such as crack length and
crack number.
However, the strains obtained from these tests are the average bending
strains on the surface of workpieces determined by the curvature radius of
the die block, while a solidification crack is caused by the local strain at
the trail of a weld pool. Consequently, it is necessary to measure the local
strains at the trail of weld pool so as to get a more accurate ductility curve
in BTR to predict weld solidification cracks.
Meanwhile, studies on the driving force, namely the mechanical strain
in BTR, are less frequently undertaken owing to the complex solidification
process involved in welding. With the rapid development in both computational
and numerical methods, a great deal of progress has been accomplished
in computational modeling of welding-related thermal and mechanical
responses [6–10].
However, most of them focus on the residual stress and distortion of
welded structures in which assumptions and simplifications on the material
behavior at elevated temperatures are often made. Such treatments can
provide results accurate enough for residual stress and distortion analyses,
but obviously not sufficiently accurate for computing the transient thermal
stress/strain fields in the vicinity of a weld pool.
To solve the problem, Z. L. Feng, W. P. Liu and S. Q. Guo had conducted
related researches concerning aluminium [11–14] and D. Dye has
also made profound research on superalloys [15]. Furthermore, weld
solidification crack predicting and simulation not only deal with measuring
ductility curves but also with simulating thermal distributions and with
strain distributions as well which is usually difficult for welding engineers
to perform.
This paper mainly deals with the methodology of simulating and predicting
solidification cracking of stainless steels [16–19]. The work includes
measurement of local strain around weld pool during the Trans-
Varestraint-Test in combination with a CCD-camera to get a more accurate
ductility curve, simulation of welding thermal/strain/stress fields to get the
driving force, and development of a computer system for simulating and
predicting weld solidification cracks.
Simulating and Predicting Weld Solidification Cracks 187
Measurement of Material Resistance
Testing Device and its Basic Principle
In the early 1980ies, Japanese scientists had measured the local strain with
the MISO method and indicated that there was strain concentration at the
trail of the weld pool [20].
In order to obtain accurate material resistance, they measured the local
strain through in-situ observation with a CCD-camera during conducting
the Trans-Varestraint-Test. The test was carried out with a HHPL-1 Trans-
Varestraint Testing Machine constructed by the Harbin Welding Institute.
Fig. 1 shows the device used in the testing.
Fig. 1. Testing device used for measuring the material resistance against weld solidification
cracking
As shown in Fig. 1, the device is composed of a CCD-camera, an image
treatment card, a xenon lamp, a focus lamp, an image treatment computer
system and the Trans-Varestraint Testing machine that is also illustrated in
Fig. 2a.
The workpiece is shown in Fig. 2b and 2c.There are two identified lines
along each side of the weld which are 5 and 6 mm away from the weld
center line, respectively, as shown in Fig. 2c. During welding, hard particles
were put inside the molten pool.
188 Modeling and Simulation
Fig. 2. Trans-Varestraint-Test and samples used in the testing
The changed distance of the identified lines and the hard particles will
be used to calculate the parameters representing the susceptibility to weld
solidification cracking as shown in Fig. 3a–c.
Fig. 4a and 4b show the weld appearances recorded with the CCDcamera
before and after welding. According to Fig. 3a and Fig. 4a, the
multiple of images can be calculated which is useful for calculating strains
arising during testing.
Fig.5 shows the recorded process of crack initiation and development at
the trail of weld pool during loading.
With the devices shown in Fig. 11, the following parameters are
measured and calculated on the basis of Fig. 3 and Fig. 5.
Simulating and Predicting Weld Solidification Cracks 189
Fig. 3. The principle for measuring parameters of the weld solidification cracking
susceptibility
Fig. 4. The workpiece before welding and the weld after welding recorded with a
CCD-camera
Fig. 5. Crack initiation and development recorded with a CCD-camera
Nominal Strain. The strain level applied to the specimen is determined by
the curvature radius of the die block as shown in Fig. 2a. As it is a pure
bending process during which the nominal strain is calculated by Eq. 1.
100%
2
= ×
R
h
ε n
(1)
190 Modeling and Simulation
In this formula, ε
n represents the nominal strain of specimen in %, h is
the thickness of plate in mm and R the curvature radius of block in mm.
Average Strain across the Weld Pool. During welding, the strength and
deformation ability of the weld pool differ from those of the cold metal
around it so that its strains should be different, too. Average strain across
the weld pool takes the form as expressed by Eq. 2.
100%
1
1
'
1 ×
−
=
L
L L
ε a
(2)
The parameter ε
a represents the average strain across the weld pool %
and L1 is original distance between two identified lines at each side of
weld in mm where L1’ is the distance between two identified lines at each
side of weld during or after bending in mm.
Local Strain around the Trail of Weld Pool. Solidification cracks usually
occur at the trail of the weld pool at which the strain is also different
from the average strain ε
a. In order to calculate the local strain around the
trail of the molten metal, that is around the cracks, hard particles have been
put inside the weld pool and two nearest points around the trail of weld
pool were chosen to assess the local strain that takes the form as given in
Eq. 3.
100%
0
0
'
0 ×
−
=
L
L L
εl
(3)
In this formula, ε
l represents the local strain around the trail of the molten
metal in %, while L0 is the original distance between two identified marks
around the trail of the weld pool in mm and L0’ is the distance between two
identified marks around the trail of the weld pool during or after bending,
mm.
Critical Strain at the Trail of Weld Pool. Critical strain is used to indicate
the strain when a crack is initiated during loading which represents the
minimal ductility of weld metal as formulated by Eq. 4.
100%
0
0
"
0 ×
−
− = L
L L
c l
ε (4)
Where εl-c is the critical strain at the trail of the weld pool, in %, L0 is the
original distance between two identified points around the trail of the weld
Simulating and Predicting Weld Solidification Cracks 191
pool in mm and L0″ is the distance between two identified points when
cracks are initiated in mm.
Concentration of Strain at the Trail of Weld Pool. The ratio of ε
l and ε
n
is used to illustrate the concentration of strain at the trail of weld pool
which takes the form as expressed by Eq. 5.
n
l
ε
ε
η =
(5)
Test Materials and Results
Two kinds of carbon steels were tested with the method described above
and appropriate parameters representing the susceptibility to weld solidification
cracking were measured and calculated. The chemical compositions
of the carbon steels are shown in Table 1 and the test results are shown in
Table 2.
Table 1. Chemical compositions (wt.%) of carbon steels used in the testing
C Mn Si S P
B 0.19 0.54 0.27 0.039 0.041
D 0.16 0.20 0.46 0.015 0.011
Referring to Table 2, the following explanations can be given:
• εl-c (critical local strain when cracks are initiated) is nearly constant for
the same steel with similar welding conditions which demonstrates that
the critical local strain is repeatable, and which further proves that the
method is practical. The critical strain around the cracks for steel B is
1.88 % to 2.47 % and the average is 1.91 % while εl-c of steel D is from
2.94 % to 3.98 % and the average is 3.46 %.
• The local strain around crack tips, ε
l,, is much higher than the average
strain represented by block curvature ratios which illustrates that there is
strain concentration around the trail of the weld pool. The average strain
concentration is about 2.2 for steel B, that is, η = 2.2.
• As strain concentration exists at the trail of a weld pool, the ductility
curves obtained with the common Trans-Varestraint-Test should be
modified by a strain concentration coefficient, η. Fig. 6 shows both the
original and modified ductility curves of steel B.
192 Modeling and Simulation
Table 2. Tested materials and calculated results of in-situ Trans-Varestraint-Tests
εn
εa
εl
η εa-work- a-c εl-c
piece
number
Block
number
Images
/frame [%]
B-17 U 11 4.12 6.96 7.17 1.74 1.65 1.94
B-19 U 12 4.16 6.69 7.61 1.83 1.69 2.24
B-20 U 14 4.00 9.04 8.22 2.06 2.35 2.02
B-22 N 9 3.26 7.13 4.97 1.53 2.18 2.33
B-26 S 8 2.60 5.45 7.34 2.82 2.26 2.47
B-27 S 2.55 1.76 2.22
B-28 S 9 2.58 6.21 6.37 2.47 2.28 1.88
B-33 H 7 2.04 5.83 5.79 2.83 1.81 1.88
D-36 U 12 3.81 9.41 2.05 3.98
D-37 U 12 3.80 9.07 1.99 3.76
D-38 U 13 3.67 9.34 2.98 3.42
D-40 U 13 3.83 8.50 2.37 3.62
D-43 U 12 3.73 8.45 2.16 3.85
D-46 U 12 3.98 7.79 1.81 3.43
D-63 U 11 3.47 8.52 2.46 3.56
D-66 U 12 3.30 9.39 2.04 3.29
D-69 U 13 4.14 6.09 2.17 3.55
D-47 H 7 2.09 4.45 2.12 3.19
D-60 H 7 1.88 5.62 1.69 2.94
εa-c critical ε
a when cracks are initiated.
Fig. 6. Comparison between original and modified ductility curves of steel B
Simulating and Predicting Weld Solidification Cracks 193
As shown in Fig. 6, the modified ductility curve of steel B moves up and
its CTS becomes greater compared with the original one, which demonstrates
that the ductility curve from the ordinary Trans-Varestraint-Test
should be modified when used in predicting weld solidification cracks. The
method described in the paper provides an effective way to get more accurate
parameters representing the susceptibility to weld solidification cracking
and founds basis for predicting them, too.
Numerical Models for Simulating Temperature Fields
Description of the Problem
As the mechanical strain is mainly caused by the transient non-uniform
heating and cooling at the weld and near-weld position, it is first of all
necessary to analyze the thermal distribution in the vicinity of the weld
pool. The problem of interest is outlined as follows:
• Material: austenitic stainless steels, SUS310 and SUS316, are analyzed.
• Welding process: tungsten inert gas welding (TIG) is simulated since it
has been used in most of the weldability tests related to solidification
cracks, which makes the results comparable.
• Numerical analytical method: the finite element analysis is used for
modeling both full penetration welding on 3 mm plate and bead-on-weld
on 10 mm plate.
• Coordinate system: the coordinate system is fixed on the workpiece, the
X-axis is along the centerline of the plate and the arc starts at point (X0,
0, 0), the Y-axis is along the edge of the plate and the Z-axis is along the
plate thickness, as shown in Fig. 7.
Fig. 7. The coordinate system
194 Modeling and Simulation
The Relevant Factors and their Treatments in Thermal
Computational Modeling
Arc Heat Input
For TIG-welding of 3 mm plate, a two-dimensional heat flux disk with a
radially symmetric welding torch is used.
Gaussian distribution is assumed to apply to the surface of the workpiece.
If the electrode moves along the X-axis and in positive direction of
the X-axis (Fig. 7), the Gaussian distribution of the arc energy input takes
the following form:
[ ]
􀂽 􀂾 􀂿
􀂭 􀂮 􀂯
− − +
= = − 2
2 2
0
2
3
0
3 ( )
( , , ) 2 3 exp
2
b b
r
r
arc r
x vt x y
r
q x y z q e b VI
π
η ,
(6)
where V, I, v are the arc voltage, current and travel speed, respectively,
η is the arc efficiency and rb is the characteristic dimensional distribution
parameter that defines the region in which 95 % of the heat flow is deposited.
For three-dimensional simulation of thermal fields for welding 10 mm
plate, the double surface ellipsoid Gaussian heat model has been proven to
be a more realistic representation of the heat source, of which the power
density distribution is shown in Fig. 8 [21].
Fig. 8. Double ellipsoid Gaussian distribution model of heat source
With the energy input rate Q in W, the fractions of the heat deposited in
the front and rear ellipse, ff, fr,, the parameters of double ellipsoid Gaussian
distribution, wx1, wx2, wy and the moving coordinates x’, y’, z’ the power
density distribution can be written by the following equation [21]:
Simulating and Predicting Weld Solidification Cracks 195
􀂰􀂰
􀂿
􀂰 􀂰
􀂾
􀂽
− <
= − >
) 0
2 2
exp(-
2
) 0
2 2
exp(-
2
( , , , )
,
2
,2
2
2
,2
2
,
2
,2
2
1
,2
1
, , , ,
x
w
y
w
x
w w
f Q
x
w
y
w
x
w w
f Q
Q x y o t
x y x y
r
x y x y
f
π
π
(7)
However, the above Gaussian surface heat flow model ignores the digging
action of the arc that transports heat well below the surface. Another
double ellipsoid source was proposed by Goldak et al. [22] as shown in
Fig. 9. The power or heat flow distribution is Gaussian along the longitudinal
axis. The front half of the source is the quadrant of one ellipsoidal
source while the rear half is the quadrant of another ellipsoidal source. The
power density distribution inside the front quadrant becomes:
2 2 2 2 2
1
3 2 3 3
1
6 3
( , , , ) f e x a e y b e z c
a bc
f Q
q x y z t = − − −
π π
.
(8)
Similarly, for the rear quadrant of the source the power density distribution
inside the ellipsoid becomes:
2 2 2 2 2
2
3 2 3 3
2
( , , , ) 6 3 r e x a e y b e z c
a bc
q x y z t = f Q − − −
π π
.
(9)
In Eqs. 8 and 9, four characteristic lengths must be determined which
physically correspond to the radial dimensions of the molten zone. The parameters
a, b, c can have different values in the front and rear quadrants,
since they are independent. In this study, two kinds of double ellipsoid
heat source were used to simulate the three–dimensional thermal distributions,
the former is called Gaussian model and the later is called double ellipsoid
model in this paper.
Fig. 9. Double ellipsoid heat source configuration
196 Modeling and Simulation
Fluid Flow in the Weld Pool
The fluid flow in the weld pool has a great effect on its temperature distributions,
as the flow speeds the heat conductivity. Research in modeling
heat and fluid flow in weld pools has been very active and much progress
has been made in the past [23–26]. Instead of explicit simulation, the effect
of the fluid flow in the weld pool on the overall temperature distribution of
the workpiece will be considered in the heat transfer model by means of
effective thermal properties such as the effective thermal conductivity.
Such an approach has been used by many investigators in the past and the
effectiveness of the approach has been well demonstrated [27–28]. Fig. 10
shows the effective thermal conductivity for different temperature ranges
of the stainless steels SUS310 and SUS316.
Fig. 10. The effective thermal conductivity for different temperature ranges
Latent Heat of Fusion
Previous studies have indicated that the latent heat of fusion highly influences
the shape and size of the weld pool as well as the temperature distributions
in the vicinity of the weld pool. Therefore, the latent heat of fusion
should be considered in this study.
Two approaches have been considered in the past to include the effect of
latent heat of fusion in the welding heat transfer analysis. The first one essentially
assumes that solidification occurs at one constant temperature.
Simulating and Predicting Weld Solidification Cracks 197
The release or absorption of the latent heat is modeled by application of a
constant specific heat of the metal.
Such a treatment fails to recognize the fact that solidification of an alloy
under most practical circumstances, such as welding and casting, takes
place over a temperature range encompassing liquidus and solidus temperature
of the alloy. In the second approach, a complex differential equation
has been adopted to calculate the latent heat release for binary Al-Cu
alloy [29]. Such computational approach is not suitable for ternary or multicomponent
alloys such as SUS310 and SUS316 studied in this thesis.
Fig. 11. Metallographic sections showing solid fractions of SUS310 stainless steel
at different temperatures
198 Modeling and Simulation
It is known that the release rate of latent heat is in direct proportion to
solid fraction, as described by the following equation:
t
f
q H s
i ∂
∂
= Δ,
(10)
where qi is the release rate of the latent heat, fs is the solid fraction, and
ΔH is the latent heat of fusion per volume.
By means of liquid quenching tests, fs, the solid fraction has been measured
from which the release rate of the latent heat can be obtained.
The experiment has been carried out in the following procedures.
First of all, put a stainless steel cube, 10 × 10 × 10 mm3 in volume, in a
crucible, and heat until it reaches a temperature a little higher than the liquidus
temperature at which it is kept for 5 minutes.
Next, cool it to a given temperature in the range between liquidus and
solidus and keep this temperature constant also for 5 minutes.
Then, take the sample out of the crucible and put it into the coolant
where it is quenched. Last, corrode the quenched cube and measure the
solid fraction with a metalloscope.
By this way, a series of solid fractions of the alloys SUS316 and
SUS310 are measured at different temperatures between solidus and liquidus.
The experimental results are demonstrated in Fig. 11.
Fig. 12. The linearizing curves of solid fractions
Simulating and Predicting Weld Solidification Cracks 199
Fig. 11 shows the solid fractions of SUS310 stainless steel at different
temperatures in the range from solidus to liquidus. The measured solid
fractions were linearized between the different temperatures as shown in
Fig. 12.
Material Properties
The physical properties of SUS310 stainless steel are shown in Table 3.
Both the thermal conductivity and the specific heat are dependent on temperatures.
However, the mass density remains constant all the time.
Table 3. Relationship between physical properties of SUS310 stainless steel and
temperatures
Temperature
[°C]
Thermal
conductivity
[W/mK]
Specific heat
[J/m3K]
Latent heat
release rate
[J/m3K]
Mass
density
[kg/m3]
20–1250 15.013
+1.363×10-2T
1250–1340 2.58×106
1340–1375 1.325×107
1375–1400
56.8
+1.98×10-2T
4.109×106
+1.138×103T
6.496×107
>1400 84.52 5.7026×106
8000
The Governing Equation
Based on the discussion above, the basic transfer models for 3 and 10 mm
plate, respectively, are assumed to be two-dimensional and threedimensional
heat conduction models with prescribed heat flux moving
along the weld, and simulated welding arc, convection and radiation heat
loss from the top, bottom and edge (for 10 mm plate only) surfaces of the
workpiece.
Because of the minor thickness of the 3 mm workpiece, the heat loss
from its edges is ignored. Therefore, all boundary conditions, i.e. the surface
heat loss and the heat flux from the welding arc, can be included in
the governing heat diffusion equation in the terms of internal heat generation
or loss.
In the heat transfer process, heat is lost from the surfaces in the form of
convection and radiation, which occur on all surfaces of the plate except
the plane of symmetry. The symmetry surface is defined as being subject
to adiabatic boundary conditions.
200 Modeling and Simulation
According to Newton’s law, the heat flux from a solid in contact with a
flowing gas or liquid is proportional to the difference between the
environment temperature T∞ and the surface temperature T through a coefficient
of convective heat transfer hc: qc = hc (T–T∞).
The heat radiated per unit area and time, by a heated body, is according
to Stefan-Boltzmann’s law proportional to the fourth power of the difference
between the environment temperature T∞ and the surface temperature
T through the radiation coefficient σε: qy = σε (T4–T∞
4).
In accordance with the coordinate system given in Fig. 7, the governing
equation for two-dimensional simulation can be written as:
Q
y
k T
x y
k T
t x
c T +
∂
∂
∂
+ ∂
∂
∂
∂
= ∂
∂
ρ ∂ ( ) ( ) .
(11)
For three-dimensional simulation, the governing equation takes the
form as expressed by Eq. 12:
Q
z
k T
y z
k T
x y
k T
t x
c T +
∂
∂
∂
+ ∂
∂
∂
∂
+ ∂
∂
∂
∂
= ∂
∂
ρ ∂ ( ) ( ) ( )
(12)
i
arc
i
c r arc q
H
q
H
T T
H
q h T T
H
q
H
q
H
Q q + +
−
−
−
= − − + + = − 2 ( ∞ ) 2σε ( 4 ∞4 ) , (13)
where ρc is the voluminal specific heat, H is the thickness of the workpiece,
and Q is the rate of total internal heat generation that consists of four
terms: surface convection heat loss, qc; surface radiation heat loss, qr; heat
flux input from the arc as described in Eqs. 6 or 7 or 8 and 9, and the release
rate of latent heat, qi, as described in Eq. 10.
Analysis of Simulating Thermal Distributions
The Finite Element Implementation
The finite element implementation of the mathematical models above can
be divided in three parts: geometrical discretization of the workpiece, consideration
of the thermophysical properties and imposition of boundary
conditions.
To discretize the 3 mm workpiece, four-node elements are used, and
smaller elements are adopted near the weld to account for possible severe
temperature gradients in those regions.
Due to symmetry along the weld line, half of the 10 mm plate is discretized
with eight-node isoparametric hexahedron elements.
Simulating and Predicting Weld Solidification Cracks 201
The FEM mesh is shown in Fig.13. Smaller elements were used near the
fusion zone and the adaptive mesh technology was used in and near the fusion
zone.
Fig. 13. Finite element mesh used for temperature calculations
In the study, the materials are assumed isotropic and homogeneous, but
with temperature dependent properties. The treatment of boundary conditions
is fairly simple.
The arc energy input and surface heat losses due to convection and radiation
are converted into a single internal heat generation term in this
work as described above.
Analysis of Simulating Results of Two-Dimensional Plate
The Influence of Arc Efficiency. As the arc efficiency highly affects the
arc energy input in the workpiece, it will have an influence on temperature
fields. Generally speaking, the arc efficiency is 0.65~0.70 for the TIGwelding
process.
To compare the influences of the arc efficiency, temperature fields with
different arc efficiencies, 0.65 and 0.70 respectively, were calculated as
shown in Fig. 14.
Fig. 14 demonstrates that the arc efficiency has a great effect on the
thermal distributions and isotherm shapes as well.
In this research work, 0.70 is finally used as arc efficiency in the calculation,
as its corresponding temperature field coincides better with the
experimental results.
202 Modeling and Simulation
Fig. 14. The effects of the arc efficiencies on the temperature distributions
The Influence of the Characteristic Dimensional Distribution Parameter.
The characteristic dimensional distribution parameter, rb, defines the
region in which 95 % of the heat flux is deposited. According to previous
studies, rb can be 2.5 mm or 3.0 mm for the TIG welding process when the
electrode diameter is 2.5 mm. Thermal distribution simulation, as shown in
Fig. 15, was performed for rb equal to 2.5 mm and 3.0 mm, respectively.
As illustrated in Fig. 15, the value of rb affects the thermal distributions.
In this study, 3.0 mm is chosen for the calculation, of which the results are
in better conformity with the experimental results [24].
Fig. 15. The effects of the characteristic radial dimensional distribution
parameter rb on temperature fields
Simulating and Predicting Weld Solidification Cracks 203
The Influence of Thermophysical Material Properties. Although both
SUS310 and SUS316 are austenitic stainless steels, their chemical compositions
and mechanical properties are much different. However, their thermophysical
properties are a little different at elevated temperatures such as
solidus, liquidus and thermal conductivity. The thermal distributions for
the two materials are shown in Fig. 16.
Fig. 16. The effects of the thermophysical material properties
on temperature fields
As indicated in Fig. 16, the isotherms in longitudinal direction are a bit
different while other parts are nearly the same. It may therefore be presumed
that the stress/strain distributions will not have much difference too.
In other words, the reason why SUS310 is more susceptible to solidification
cracking than SUS316 is mainly because the resistance to crack of
SUS310, which is determined by metallurgical factors, is lower than that
of SUS316.
Analysis of Simulating Results of Three-Dimensional Plate
Temperature Distributions of Two Heat Sources. Three-dimensional
numerical simulation of the welding process is developed with the model
above. The work is accomplished with commercial software MARC. Welding
is assumed to be finished in 75 s and the cooling time is 200 s. The heat source
is accomplished by a specially defined user subroutine. Fig. 17 illustrates the
three-dimensional temperature distribution of a double ellipsoid model.
The comparison between the temperature distributions of the double ellipsoid
model and those of the common Gaussian model are shown in Fig. 18.
204 Modeling and Simulation
Fig. 17. The three-dimensional temperature distributions of two double ellipsoid
model
Fig. 18. The temperature distributions of different heat models
As indicated in Fig.18, the different heat source models have an important
influence on the temperature distributions. The results have been modeled
with the double ellipsoid model when no particular mention was
made.
The Influence of Fluid Flow in the Weld Pool. The fluid flow in the weld
pool greatly affects the temperature distributions in or near the fusion
and heat-affected zones as it speeds the heat conductivity.
Fig. 19 shows the simulated temperature distributions. It can be seen that
the weld pool is shorter when the influence of the fluid flow is considered,
compared to the weld pool obtained without consideration of this
influence. However, this influence gets weaker in the areas away from the
weld pool.
Simulating and Predicting Weld Solidification Cracks 205
Fig. 19. The influence of the fluid flow on the temperature distributions
The Effects of Latent Heat of Fusion. The latent heat effects are more
important for improving the accuracy of the simulated results. Fig. 20 illustrates
the influence of different treatments of latent heat of fusion on
temperature fields.
When latent heat is released linearly, the weld pool is longer than that
simulated assuming that uniform latent heat is released from the solidus to
liquidus. Near the weld pool and at lower temperatures, the isotherms are
shorter than those for uniform latent heat.
Fig. 20. The effect of different treatments of latent heat
on the temperature distributions
206 Modeling and Simulation
The Influence of Welding Parameters. In this study, two different sets of
welding parameters are adopted, the first: I = 200 A, U = 13.5 V,
􀈣 = 2 mm/s and the second: I = 325 A, U = 16.5 V, 􀈣 = 4 mm/s. Fig. 21
provides two kinds of temperature fields. Although with almost equal arc
energy, the shapes and sizes of isotherms for the different welding conditions
vary greatly. Larger welding current together with higher travel speed
results in narrower and longer isotherms than at the other conditions.
Fig. 21. The influence of the welding parameters on the temperature distributions
Fig. 22. Temperature change with the time-step
Simulating and Predicting Weld Solidification Cracks 207
The Effects of Adaptive Mesh Technique. The temperature distributions
in two different analyses with adaptive mesh and without adaptive
mesh at different time-steps are shown in Fig. 22.
It can be observed that the temperature distributions of the two analyses
are almost the same within the high temperature range, but there is big difference
when the temperature decreases. The adaptive technique makes the
FEM mesh much denser, so that the temperature distributions agree well
with the experimental results.
Fig. 23. The comparison between the simulated temperatures and the measured
temperatures
Verification of the Simulated Temperature. To testify the computational
results, the temperatures at different positions perpendicular to the weld
centerline were measured using thermocouples.
The examined material is SUS310 stainless steel plate with 10 mm
thickness. The welding current is 200 A. The welding speed is 2 mm/s.
The fusion width of the weld is analyzed after welding. Fig. 23 shows the
measured temperatures and simulated ones considering two kinds of heat
sources. It can be observed that the temperature values obtained with the
double ellipsoid model coincide well with those measured especially at
high temperature.
The fusion width and shape of the weld are shown in Fig. 24. The results
from the comparison between simulations and measurements are
summarized in Table 4.
208 Modeling and Simulation
Table 4. The results of simulations and experiments
Double ellipsoid
model
[mm]
Gaussian model
[mm]
Experimental
[mm]
Fusion Width 12.5 9.5 13.0
As shown in Fig. 24, the macrograph of weld pool geometry is similar
to the calculated weld pool shape of double ellipsoid heat source.
From Table 4 it is apparent that the fusion width from the double ellipsoid
model calculation shows excellent agreement with the experimental
results.
Fig. 24. The fusion width and shape of the weld
Simulating and Predicting Weld Solidification Cracks 209
Numerical Simulation of Stress/Strain Distributions
Basic Consideration during Development of Stress/Strain
Models
The Deformation in the Weld Pool. It is known that the liquid metal of
the weld pool has two characteristics. First, the liquid metal can only withstand
minimum force which can also cause its flow and deformation. Second,
the previous strain and deformation of resolidified weld metal will be
“annealed” at the liquid-solid interface upon recrystallization because the
weld metal ahead of arc melts and solidifies after the arc has moved forward.
Therefore, except for the effect of the fluid flow in the weld pool on the
shape of the weld pool, which could be adequately dealt with in the heat
transfer analysis, the fluid flow behavior in the weld pool can be omitted in
finite element modeling of welding-specific stresses and distortions. A
good way to deal with the weld pool in such analysis is to simply exclude
it from the solution domain. Noting that the position of the weld pool continuously
changes in order to keep up with the traveling electrode, the exclusion
of the weld pool from the solution domain has to be done during
dynamic and continuous fusion over the entire welding process.
The Elasto-Viscoplasticity in the Solid-Liquid Region. At present, there
is no agreement on whether the elasto-viscoplasticity in the area of solidliquid
coexistence should be considered or not. In the late 1970’s, researchers
in the former USSR pointed out that the mechanical behavior of
weld metal in solid-liquid state had characteristics of elasto-viscoplasticity
and that its stress/strain was related not only to temperatures but also to the
history and time of the load.
However, some researchers hold the viewpoints that the elastoviscoplasticity
of liquidus metal need not be included in the course of
simulating stress/strain distributions. Among them, some believe that
elasto-viscoplastic deformation is smaller than plastic deformation in the
solid-liquid region and can be ignored. The others think that it is very difficult
to measure elasto-viscoplastic properties at elevated temperatures
and that it is also quite complicated to conduct such kinds of computations
which will make inaccurate results.
To study the elasto-viscoplasticity in solid-liquid regions, a series of
rheological tests of stainless steels SUS310 and SUS316 were conducted
within the scope of this study with the elasto-viscoplasticity measuring device
of the Beijing Institute of Science and Technology. Some of the
measured results are shown in Fig. 25.
210 Modeling and Simulation
Fig. 25. The response curves of stainless steel SUS316 during loading and unloading
Fig. 25 demonstrates that when the temperature is below the coherent
temperature as shown in (a) and (b), its behavior is mainly of the elastic
deformation type, and that when the temperature of the metal is higher
than the coherent temperature as shown in (c) and (d) its behavior is of the
elastic, viscoelastic and viscoplastic deformation type.
Elasto-viscoplasticity is not obvious in welding owing to the high cooling
rate and to the short exiting time of the solid-liquid region. When the
temperature is higher than the coherent temperature the metal in the solidliquid
region exhibits liquid characteristics of molten metal because there
is more liquid metal with interdentritic flow.
As a result, the region of solid-liquid coexistence can be divided into
two areas: liquid/solid area and solid/liquid area. In the former case, the
state of the metal is close to liquid metal of the weld pool and its mechanical
behavior can be treated with the element rebirth method, while the state
of the metal in the later case approximates solid metal and its stress/strain
constitutional relation is in accord with the elasto-plastic heat theory.
In this study, the region of the SUS310 weld with temperatures from
1375 °C to 1400 °C was treated as liquid metal and the other region with
temperatures from 1250 °C to 1375 °C was treated as solid metal.
Simulating and Predicting Weld Solidification Cracks 211
The Solidification Shrinkage. It is well known that most metals and alloys
contract during solidification and that solidification shrinkage has a
significant impact on the stress and strain fields of a weld. As far as the
thermal stress modeling of an alloy is concerned, the effect of solidification
shrinkage is phenomenologically equivalent to thermal expansion/
contraction due to temperature change. It causes mechanical deformations
in the liquid region as well as in the other parts of the weld, owing to
the compatibility requirement of the continuum solid mechanism.
In this study, the solidification shrinkage is thus assumed to be linearly
distributed in the solidification temperature range. Therefore, it can be effectively
treated as an additional thermal expansion/contraction term
caused by temperature changes.
Based on the above discussion, thermal elastic-plastic theory was used
to simulate the thermal stress/strain of interest. The computation was made
on the following assumptions or conditions:
- Materials are isotopic and homogeneous;
- Mechanical properties are temperature-dependent;
- Material yields conform to Mises rules;
- Viscosity and creep of the materials are neglected;
- The characteristics in the plastic area are represented by flow rules
of the plasticity and strain-hardening behavior.
Finite Element Implementation
As stress/strain analysis is based on the heat transfer analysis, the plate was
discretized in the same way as in heat conduction analysis as shown in
Fig. 26.
As discussed above, the best way to deal with the weld pool in such
analysis is to exclude it from the solution domain. Therefore, the solution
domain is composed only of continually changing solid parts of the welds.
The approach is the element rebirth technique by which the elements
whose temperature is above the coherent temperature (so they are inside
the traveling weld pool) are simply removed from the finite element model
and added back into the model after their temperature drop to below the
coherent temperature. When the elements are added back, they are assumed
to be annealed and have zero deformation. Since the weld pool follows
the moving welding arc, such element removal/inclusion is dynamic,
namely, it follows the location of the weld pool.
In this study, a program is written to complete the above element rebirth
scheme.
212 Modeling and Simulation
Fig. 26. The mesh used in calculating stress/strain distributions
Discussion of Numerical Analytical Results of the Stress/Strain
Distributions
The Effects of Deformation in the Weld Pool. The effects of deformation
in the weld pool are shown in Fig. 27 which indicates that larger compression
existed when the deformation in the weld pool was not taken into
account while the strain was completely annihilated when the element rebirth
technique was used.
The Effects of Solidification Shrinkage. Fig. 28 shows different strains
developed in the weld center with or without consideration of solidification
shrinkage. As indicated in Fig. 28, the strains are much higher when
solidification shrinkage is taken into account than those obtained disregarding
solidification shrinkage.
Fig. 27. The effects of deformation on stress shrinkage distributions
Simulating and Predicting Weld Solidification Cracks 213
Fig. 28. The influences of molten metal on strain distributions
The Influences of Welding Conditions. As the welding conditions
greatly affect the thermal distributions of stainless steels SUS310 and
SUS316, especially when the thermal fields hold elevated temperatures,
they will certainly influence the stress/strain development of the weld. On
the condition of penetration, two kinds of welding conditions are used to
simulate the stress/strain evolutions that are illustrated in Fig. 29.
From Fig. 29, we can find that the strains at the trail of the weld increase
much earlier at lower welding speed than at higher welding speed which
indicates that the driving force behind solidification cracking is stronger at
higher welding speed. If the welding parameters did not affect the cracking
resistance it could have been concluded that the steels are less susceptible
to solidification cracking at higher welding speed.
Fig. 29. The effect of welding conditions on strain distributions
214 Modeling and Simulation
In fact, this conclusion contradicts some phenomena in practical production
because welding conditions do have effects on the cracking resistance
of materials. For instance, dendrites will grow perpendicularly to the fusion
line and meet at the center of weld when the welding speed is high,
which results in the formation of the eutectic there and in the decrease of
the weld metal cracking resistance.
The Influences of Restraint. To study the influences of restrained weld
edge on the crack driving force, the strain development for two kinds of
restraint was simulated and compared in Fig. 30.
From Fig. 30 we can find that the strain development curve for the free
edge is lower than that of the restrained edge, which agrees well with the
general concept of crack theory.
Fig. 30. The effects of restraint on strain distributions
Development of a Simulation and Prediction System for
Solidification Cracks
From the work described above we can see that it is very complex to perform
a complete simulation of solidification cracks with a common commercial
software package, as various specific factors should be taken into
account. To solve the problem, a simulation and prediction system for
weld metal solidification cracks has been developed with which a welding
engineer can easily carry out the task on the basis of two-dimensional
simulation of weld cracks with Adina&T 8.4.
Simulating and Predicting Weld Solidification Cracks 215
Generation of Input Cards
When Adina&T is used to execute simulation, there are two kinds of input
data cards needed which require a strict data format, one is for simulating
thermal distributions and the other is for computing stress/strain fields. To
simplify the use of the software package, the system provides user-friendly
interfaces to input relative parameters, performs corresponding calculation
and data treatment, and generates input data files automatically. The interface
for generating cards is shown in Fig. 31 from which we can find that
there are three kinds of parameters for users to input, they are basic data
about grid generation, material properties and welding conditions.
The system provides two ways to discretize the workpiece as shown in
Fig. 31. One way asks users to enter the coordinate data and the other asks
users to input the proportion of discretization. Four-node elements are
used, and smaller elements are advised near the weld to account for the
possible severe temperature gradient in those regions. After the data input,
the workpiece is discretized, of which the results can be previewed and the
corresponding data are then stored in input cards.
The thermophysical properties of alloys, including elastic modulus,
Poisson's ratio, expansion coefficient, thermal conductivity and yield
stress, are input easily and the curves of the thermophysical material properties
versus temperatures can be viewed, too.
Fig. 31. Interface for generating input cards
216 Modeling and Simulation
Fig. 32. Interface for inputting weld conditions
The welding conditions are input as shown in Fig. 32. After obtaining
the welding parameters, the system calculates the heat generation according
to the Gaussian distribution of the arc heat input.
The heat generation for each element is computed and treated as a part
of internal heat during simulation and written in the input card. After the
input card is completed, the simulating work can be completed with
Adina&T which has been combined into the system.
Viewing Simulated Results
In order to view simulated results conveniently, the system adopts a program
to organize the input cards and corresponding results. Users cannot
only browse input cards, but also view corresponding thermal distributions
and stress/strain distributions in different ways.
To display the simulation results in 3-D meshes and contours, the system
makes use of a software package, Matlab5.3, as a server as well as of
Visual Basic 6.0 as a custom language to display the results. Fig. 33 shows
a 3-D mesh of thermal distributions and Fig. 34 shows the contours of temperatures.
The system uses MatrixVB to display thermal/stress/strain cycles, their
distributions in transverse or vertical sections.
Simulating and Predicting Weld Solidification Cracks 217
Fig. 33. 3-D mesh of thermal distributions
Fig. 34. Contours of temperatures
218 Modeling and Simulation
Predicting Solidification Cracks
As mentioned above, weld metal solidification cracking occurs when the
driving force to crack exceeds the resistance of a material.
The fundamental parameter representing the material resistance is its
ductility curve in the brittle temperature range (BTR) in the course of weld
metal solidification, while the mechanical strain accumulated in the BTR is
considered as the mechanical driving force. To obtain the ductility curves
in the BTR, measurements were carried out in the Trans-Varestraint-Test
under different conditions.
Fig. 35. Strain cycle of a point in the centerline
As for the driving force, a program module has been developed to deal
with it in the following three steps:
1. Obtain mechanical strain
In the simulated results, the total strain is composed of thermal
strain, elastic strain and plastic strain. The mechanical strain consisting
of elastic strain and plastic strain is obtained by subtracting
the thermal strain from the total strain.
2. Get thermal cycle
Thermal cycle, that is, temperature history, can be gotten from the
simulated results through data treatment.
Simulating and Predicting Weld Solidification Cracks 219
3. Conversion of the strain-time curve into the strain-temperature
curve
The mechanical stress obtained from the simulation is a function
of time as shown in Fig. 35, while the driving force to solidification
cracking is the function of transverse mechanical strain
against temperature. By means of a line with a slope of -1 and the
thermal cycle, the simulated strain can be converted into the driving
force. As a result, weld solidification cracking can be predicted
with the system as shown in Fig. 36.
According to the relative position of two curves, weld solidification
cracking can be predicted. In Fig. 36, the driving force is lower than the
material resistance, so there will be no solidification cracking.
Fig. 36. Predicting weld metal solidification cracking
Conclusions
1. The strain concentration exists at the trail of the weld pool during Trans-
Varestraint-Testing, so that the local strain at the trail of weld pool was
measured and the ductility curve was modified by in-situ observation
and recorded with CCD camera during Trans-Varestraint-Testing.
220 Modeling and Simulation
2. For three-dimensional simulation of thermal distributions, a double ellipsoid
heat source model is more practical compared with Gaussian surface
thermal models.
3. The latent heat of fusion, fluid flow in the weld pool and the welding
parameters have influences on the thermal distributions in the vicinity
of the weld pool, which indicates that it is necessary to consider these
factors when weld metal solidification cracking is simulated.
4. The problem of deformation of the weld pool can be solved with the
element rebirth scheme, and solidification shrinkage is taken into account
by amplifying the thermal expansion/contraction coefficient.
5. The simulation and prediction system provides an easy way for welding
engineers to perform simulation of weld solidification cracks, as the system
needs minimal inputs that are related to each other and conducts
other tasks automatically, including previous and post data treatment,
carrying out calculation, display of the results in different ways and performing
prediction of solidification cracks.
Acknowledgement
Financial support by the National Natural Science Foundation of China
under contract No. 50175040 is gratefully acknowledged.
References
1. Savage WF, Lundin CD (1965) The Varestraint Test. Welding Journal, ,
44(10): 433s–442s
2. Savage WF, Lundin CD (1966) Application of the Varestraint Technique to
the Study of Weldability. Welding Journal, 45(11): 497s–503s
3. Zujue D, Yongming P, Renpei L, Wilken K (1994) A Comparative Study on
the Criteria of Varestraint and Transvarestraint-Test of Solidification Cracking
Sensitivity. IIW-Doc IX-1764-94
4. Sun Z (1992) Study of Solidification Crack Susceptibility Using the Solidification
Cycle Hot-Tension Test. Materials Science & Engineering A: Structural
Materials: Properties, Microstructure and Processing, A154(1): 85–92
5. Takeshi S, Katsuei H, Ryouichi Y (1990) Development of Solidification
Cracking Test for MAG Narrow Gap Welding. Effect of Boron Contents on
Solidification Cracking. Quarterly Journal of the Japan Welding Society, 8(1):
21–25
6. Brooks JA (1993) On Modeling Weld Solidification Cracking. In: Proceedings
of Internal Conference on Modeling and Control of Joining Processes,
Orlando, Fla., pp 174–185
Simulating and Predicting Weld Solidification Cracks 221
7. Wu CS (1990) Numerical Analysis of Welding Heat Process. Publishing
House of Harbin Institute of Technology
8. Yang YP (1995) A Study of Preventing Welding Hot Cracking of Thin High
Strength Aluminum Alloy Plate with the Inverse Strain Method. Ph.D. Dissertation
of Harbin Institute of Technology
9. Cao ZN (1993) Numerical Analysis of Fluid Flow and Thermal Distributions
in Weld Pool for TIG/MIG Penetrations. Ph.D. Dissertation of Harbin Institute
of Technology
10. Zheng W (1996) Numerical Simulation for Transient Behavior of Fluid Flow
and Temperature Field in Pulsed Current TIG Weld Pool. Ph.D. Dissertation
of Harbin Institute of Technology
11. Feng ZL (1994) Computational Analysis of Thermal and Mechanical Conditions
for Weld Metal Solidification Cracking. Welding in the World, 33(5):
340–347
12. Feng ZL (1993) A Methodology for Quantifying the Thermal and Mechanical
Conditions for Weld Metal Solidification Cracking. Ph.D. Dissertation of the
Ohio State University, pp 15–24, 58–90, 280–300
13. Liu WP (1995) Computational Analysis and Prediction of Weld-Solidification
Cracking. Computational Materials Science, 43(9): 211–219
14. Guo SQ (1995) A Study of Controlling Welding Hot Cracking and Distortion
of Thin Aluminum Alloy Plate with High Susceptibility to Hot Cracking.
Ph.D. Dissertation of Harbin Institute of Technology
15. Dye D, Hunziker O, Reed RC (2001) Numerical Analysis of the Weldabality
of Superalloys. Acta mate, 49: 683–697
16. Shi JX (1997) The Numerical Simulation of the Driving Force of SUS310 Solidification
Cracking. Ms. Thesis of Harbin Institute of Technology
17. Zhang HZ (1998) Numerical Simulation of Driving Force for Weld Metal Solidification
Cracking of Stainless Steels. Master Thesis of Harbin Institute of
Technology
18. Wei YH et al. (2000) Stress/strain Distributions for Weld Metal Solidification
Crack in Stainless Steels. China Welding, 9(1): 1–6
19. Zhang YC (2000) A Simulation and Prediction System for Welding Solidification
Crack. Master Thesis of Harbin institute of technology, 7
20. Matsuda, Nakagawa, Matsubara (1983) Quantitative evaluation of solidification
brittleness of weld metal during solidification by in-situ observation and
measurement (Report II) – Solidification ductility curves for steels with the
MISO technique. Trans of JWRI, 12(1): 73–80
21. Shah AK et al. (1995) Weld heat-affected zone in Ti-6Al-4V Alloy, part I –
Computer simulation of weld variables on the thermal cycles in the HAZ.
Welding Journal, 76(9): 293s–305s
22. Goldak J (1984) A new finite element model for welding heat sources. Metallurgical
Transactions B, 15B: 587–600
23. Varol I, Baeslack WA, Lippold JC (1989) Characterization of Weld Solidification
Cracking in a Duplex Stainless Steel. Materials Characterization, 25(2-
5): 555–573
222 Modeling and Simulation
24. Imoto I, Kim YC (1991) Mechanical Study on Quasi-solidification Cracks
Under Pulsating Loads. Welding on Structures in Service Condition (report
II). Quarterly Journal of the Japan Welding Society, 9 (1): 43–47
25. Ichikawa K, Bhadeshia HKDH, MacKay DJC (1996) Model for Solidification
Cracking in Low-Alloy Steel Weld Metals. Science and Technology of Welding
and Joining, 1(1): 43–50
26. Matsuda F, Nakagawa H, Tomita S (1989) Between Strain-Rate Dependence
of Critical Strain Required for Solidification Crack Initiation and Grain
Boundary Sliding - Investigation on Weld Solidification Cracking by MISO
Technique (report IV). Quarterly Journal of the Japan Welding Society, 7(1),
105–110
27. Chan YW (1987) Computer Simulation of Heat Flow in Pulsed Current Arc
Welding. Metal Construction, (10): 599–606
28. Zheng W (1996) Numerical Simulation for Transient Behavior of Fluid Flow
and Temperature Field in Pulsed Current TIG Weld Pool. Ph.D. Dissertation
of Harbin Institute of Technology
29. Liu W (1987) Numerical Simulation of Al-Cu Alloy Solid Mechanical Behavior,
Stress-strain During Solidifying and Solidification Crack. Ph.D. Dissertation
of Harbin Institute of Technology
Integrated Mechanical-Metallurgical Approach to
Modeling of Solidification Cracking in Welds
V. Ploshikhin, A. Prikhodovsky, M. Makhutin, A. Ilin, H.-W. Zoch
Neue Materialien Bayreuth GmbH, Germany
Abstract
The presented approach is based on experimental observations which
clearly show that the hot cracking phenomenon is a result of accumulation
of macroscopic tensile strains in a microscopic intergranular film of liquid
enriched with segregating elements at the final stage of solidification. The
modeling takes into account the effects of strain accumulation as well as
the influence of the microstructure and the thermo-dynamical properties of
the welded material. The integrated approach provides a clear phenomenological
interrelation between parameters of the welding process, properties
of the welded material and the cracking susceptibility. It is successfully
used for the development of technological means, such as a multibeam
welding, for eliminating solidification cracking.
Hot Cracking Problematic: Existing Approaches
The pioneering experimental research on solidification cracking in Alalloy
welds was carried out by Singer et al. [1–3] and later by Pumphrey et
al. [4] at the end of the 1940ies. They have clearly shown the strong dependence
of the cracking susceptibility upon the alloy composition.
At the beginning of the 1950ies, Palling introduced a clear explanation
of the solidification cracking mechanism [5]. Based on the number of tensile
tests of solidifying samples, Pellini suggested his “strain theory of hot
cracking”. According to this theory, the tensile strain developed during
solidification is localized in the mechanically weak intergranular film
formed at the end of solidification due to the segregation. This film limits
the strain transmission and therefore it accumulates the major part of the
whole strain (Fig. 1a).
224 Modeling and Simulation
The accumulated tensile strain exceeding some critical value leads to the
tearing of the liquid film, i.e. to the formation of the hot crack. Pellini
pointed out explicitly that, because of the strain localization, even relatively
slight loading of the two-phase region leads to hot tearing. Providing
a physically justified mechanical-metallurgical explanation of the mechanism
of crack initiation, the theory of Pellini does not offer any quantitative
criterion for predicting the cracking susceptibility.
The first quantitative description of solidification cracking was proposed
by Prokhorov in the early 1950ies [6–8]. Prokhorov also considered
the mechanical tensile strain as a reason for crack formation. However, the
effect of the strain localization was not taken into account (Fig. 1b).
Prokhorov calculates the deformation of the liquid-solid region integrally
and compares it with the so-called “deformability”, a term introduced
by Prokhorov in order to characterize the ability of a material to
accumulate deformations without rupture at higher temperatures within the
solidification range. It was considered as a material property, of which the
value can be obtained experimentally. When the actual deformation of the
two-phase region exceeds the deformability of the material, hot cracking
occurs.
Fig. 1. Schematic illustration of the difference between the basic theories of solidification
cracking: (a) Pellini [5] (considering the strain localization) and (b) Prokhorov
[6] (without regard to the strain localization)
The indisputable advantage of Prokhorov’s theory is the separation of
the mechanical factors and the attempt to quantify them. However, the integral
consideration without regard to the strain localization hinders the
application of this approach to the solution of practical problems.
Integrated Mechanical-Metallurgical Approach to Modeling 225
The methods developed for the experimental measurement of such an
integral characteristics as the deformability are too sensitive to the investigational
conditions. In these experiments, the controlled external loads of
the welded specimen are usually used. In reality, the local deformation
values of the two-phase region are not uniformly distributed over the
welded joint. The larger this inhomogeneity, the bigger the experimental
error. For example, experimentally measured ductility values reported in
literature by Trans-Varestraint tests [9–11] and by MISO techniques [12,
13] differ by almost one order of magnitude.
The second critical point of Prokhorov’s theory is the neglect of the
metallurgical conditions in the mushy zone, i.e. the microstructure of the
two-phase region formed during solidification is not taken into account.
The experiments for determining the deformability are usually conducted
only for certain process parameters.
However, the alteration of the welding conditions leads to changes in
the microstructure and, correspondingly, to a different response to the external
load. In turn, this results in different values of the deformability.
Thus, the experimental determination of the ductility should be carried out
for each possible set of process parameters. Because this procedure is very
complicated, measurements are usually taken only for a certain set of process
parameters, and then the obtained strength index is considered to be a
material property independent of the welding conditions. This conclusion
can be valid for some cases. For example, if a specific welding process for
plates with a certain thickness is considered. However, the direct transfer
of the measured deformability of one welding process to another (e.g. of
MIG to laser welding), or to different plate thicknesses is always connected
with some predefined errors arising from the changes in the state of
the two-phase region.
Thus, while providing a possibility for a relative characterization of different
materials with respect to their cracking susceptibility, the theory of
Prokhorov does not offer a correct quantitative criterion for predicting the
cracking susceptibility.
The majority of work on modeling of solidification cracking in welds
[14–23] is based on Prokhorov’s approach, i.e. neglecting the strain accumulation
and the microstructure effects. Several existing criteria represent
some mechanical conditions which are usually a critical value of strain or
strain rate. A precise analysis has shown that none of the existing criteria
can be considered as universal, which indicates the necessity for revision
of existing approaches [16]. According to Feng [19], who made a considerable
contribution to precise modeling of hot tearing, it is still impossible
to predict the behavior of welded joints under real “field” conditions based
only upon laboratory experiments.
226 Modeling and Simulation
There are only two modeling approaches appearing recently which are
based upon the localization aspect of crack formation. The first one is proposed
by Shibahara [24, 25] who uses the special numerical technique of
the so-called “interface elements” for the simulation of crack initiation and
crack propagation. The assumptions concerning the physical mechanism of
crack formation are taken from the solid state fracture mechanics. The
metallurgical conditions in the two-phase region are disregarded. The
value of the stress in the liquid-solid region is taken as a criterion for the
crack initiation. This stress is compared with the critical value predetermined
through the correlation with the surface energy of the melt. The surface
energy is considered to be a known quantity which has a unique value
for the given temperature.
Despite the consideration of the localization aspect of crack formation,
the approach of Shibahara has some critical points. First, the experimental
determination of the surface energy at high temperatures is a very complicated
task. The surface energy of the melt is very sensitive to any variation
of the chemical composition. The addition of a negligibly small amount of
a surface active element into the melt can lead to an excessive alteration of
the surface energy over an order of magnitude. Second, the calculated
value of stress in the two-phase region is very sensitive to the mechanical
properties at high temperatures. The experimental measurements of such
properties like the yield strength show large systematic errors. It can lead
to significant errors during the calculation afterwards.
Another local modeling approach developed by Hilbinger [26–28] is
based on the theory of Pellini. The simulation of hot cracking is carried out
as in the previous approach using the finite element method. The localization
of the tensile stress in the liquid film of the rest melt is taken into account
through the introduction of the “liquid” elements in the middle of the
weld seam. These elements have very low yield strengths in the liquidussolidus
temperature range. The criterion of crack initiation is the maximal
allowable deformation of the “liquid” elements in the two-phase region.
The critical deformation is the experimentally fitted parameter. Like the
method of Shibahara, the method of Hilbinger also allows a visual presentation
of crack initiation and propagation.
Experiments
The aim of the experiments was to determine the transition state to hot
cracking in order to understand the mechanism of crack initiation. The
strip-shaped samples of binary AlSi-alloys with different Si-concentrations
Integrated Mechanical-Metallurgical Approach to Modeling 227
were restrained on the edge on one side remaining free on the opposite side
as shown in Fig. 2. The laser beam welds were produced at different distances
from the free side. This experimental setup is based on the effect of
the crack formation during welding close to the free edge of the plate. This
effect was used by Hilbinger [26, 28] for similar investigations.
Fig. 2. Experiments on determination of the critical thermo-mechanical conditions
needed for the crack initiation: (a) crack initiation by welding close to the free
edge and (b) no crack initiation by welding at a distance bigger than the critical
one
The parameters of welding made with Nd:YAG laser were constant for
all samples. The laser power was 1.8 kW and the welding speed was
2.8 m/min. Due to variation of the distance between the weld and the free
edge different thermo-mechanical conditions can be obtained. At small
distances to the free edge, a centerline solidification crack is usually generated
in these experiments [27, 28]. At a certain critical distance, the state
of the threshold to solidification cracking can be achieved.
Fig. 3a shows typical top view photographs of the weld in the state of
threshold to solidification cracking under different magnifications. That
“crack” at lower magnifications (Figs. 3a–b) reveals a sequence of the
holes separated by “bridges” of the eutectic, which are clearly seen at
higher magnifications (Figs. 3c–d).
228 Modeling and Simulation
Fig. 3. Initiation of the solidification crack (uncompleted crack) under different
magnifications (top-view-sections, binary alloy Al-3 % Si)
Such a threshold state is obviously a result of strain accumulation in the
intergranular film which, because of segregation, remains liquid until the
final stage of solidification is reached. An important result is that the value
of the liquid film deformation needed for the crack initiation is relatively
small. This value can be directly obtained from the metallographic view
(Fig. 3d, Fig. 4c) and is approximately 10 μm. Considering the initial
thickness of the liquid film, which is between 1 and 2 μm, the strain value
can be estimated at approx. 1000 %. The accumulated tensile strain was
high enough for crack initiation, but it was insufficient for complete tearing
of the liquid film.
A very interesting transition state is shown in Fig. 5. The metallographic
observation at low magnifications shows the existence of a centerline
crack. However, precise examination at higher magnification reveals this
seeming crack to be a thin strip of binary eutectic formed at last stages of
solidification. Apparently, this weld seam was very close to form a solidification
crack, but tearing of the liquid film did not happen. The value of
the liquid film extension obtained from the metallography is approx. 3–
5 μm. The tensile strain of this liquid film at a value of up to 500 % was
relatively large. However, the accumulated strain was not high enough for
crack initiation.
Integrated Mechanical-Metallurgical Approach to Modeling 229
Fig. 4. Initiation of the solidification crack (uncompleted crack)
under different magnifications (cross sections, binary alloy Al-3 % Si)
Fig. 5. Fictive solidification crack under different magnifications
(top-view-sections, binary alloy Al-3 % Si)
230 Modeling and Simulation
These results support Pellini’s theory and clearly show the importance
of the local consideration of the deformation in the two-phase region. In
accordance with conclusions of Pellini, the metallographic observations
indicate that even minor deformation of the two-phase region can be sufficient
for crack initiation. For the investigated alloys and welding parameters,
the critical deformation had a value of approx. 10 μm.
From the photographs it can be seen that the value of the critical deformation
is of the same order of magnitude as the dendrite arm spacing. This
shows the importance of the microstructure state for hot tearing. For this
reason, the neglect of the influence of the microstructure as practiced in
many modeling approaches of hot cracking is inadequate. Evidently, the
ability of the microstructure to accumulate the tensile strain without crack
initiation depends on both the geometric parameters (primary and secondary
dendrite arm spacing) and the rest amount of melt at the end of solidification.
Modeling
A mechanism of the solidification cracking, which can be derived from the
experiments described in the previous section, is shown schematically in
Fig. 6. The intergranular liquid film of the segregating elements at the
weld centerline accumulates the tensile strain transmitted by the solidified
dendrites from the completely solid regions of the base material. The displacement
exceeding some critical value δcr
acc results in tearing of the liquid
film of segregates, i.e. in the initiation of the solidification crack.
The strain accumulation strongly depends on the length of the mushy
zone in the direction of welding. This length is defined by two isotherms:
the real liquidus and the real solidus one denoted as TL* and TS*, respectively.
These isotherms represent the so called real solidification range, i.e.
the temperature interval of the simultaneous coexistence of the solid and
liquid phases (the mushy state).
Neglecting the undercooling of the dendritic tips, the real liquidus can
be assumed to be represented by the liquidus temperature from the equilibrium
phase diagram. The real solidus represents the lowest temperature for
the existence of the mushy state. For a certain alloy, this temperature must
be calculated taking into account the effects of micro-segregation.
For example, for Al-Si binary alloys, the residual liquid usually solidifies
as a binary eutectic. For this reason, the real solidus in this case can be
reproduced by the temperature of the binary eutectic.
Integrated Mechanical-Metallurgical Approach to Modeling 231
Correct modeling of the solidification cracking implies the solution of
the following constituent problems:
• thermo-mechanical problem (evaluation of the maximal displacement of
the liquid film of segregating elements taking into account the effect of
strain localization);
• metallurgical problem (evaluation of the real solidification range taking
into account the effects of segregation; calculation of the phase distribution
within the mushy zone; evaluation of the geometrical microstructure
parameters).
Fig. 6. Schematic representation of the solidification cracking mechanism
The general form of the criterion for crack initiation can be represented
as follows:
232 Modeling and Simulation
δacc
max(DS, WC, PBM, PFM) ≥ δacc
cr(PMS), (1)
where δacc
max is the maximal displacement accumulated in the intergranular
liquid film of the segregating elements, δacc
cr is the critical displacement
which characterizes the ability of the liquid film to accumulate
tensile strain without tearing, DS are the particularities of the design of the
welded joint, WC are the welding conditions (for instance, velocity), PMB
and PFM are the relevant properties of the base material and of the filler
material, respectively; PMS are the parameters of the microstructure.
The phenomenological interrelation between the parameters included in
the cracking criterion and the crack initiation mechanism is represented
schematically in Fig. 7.
Fig. 7. Phenomenological interrelation between parameters included in the cracking
criterion and the process of the crack initiation
A more detailed list of parameters relevant to the phenomenon of solidification
cracking is given below:
• Nature of Design (DS): complete geometry of the welded joint (including
the whole structure) and the position of the weld in the structure;
• Welding Conditions (WC): heat source (power, power distribution, velocity),
shielding gas, filler material, clamping (thermal and mechanical
influence), preheating (location and temperature), additional heating
and/or cooling during welding (location, heating and/or cooling intensity);
Integrated Mechanical-Metallurgical Approach to Modeling 233
• Properties of the base material (PBM): temperature-dependent thermal
properties (thermal conductivity), thermo-dynamical properties (heat
capacity, real solidification range), temperature-dependent thermomechanical
properties (thermal expansion coefficient, yield strength,
elasticity modulus);
• Properties of the filler material (PFM): the same properties as for the
base material, especially the thermo-dynamic ones;
• Parameters of the microstructure (PMS): distribution of the liquid and
the solid phase within the mushy zone, residual liquid phase fraction,
primary and the secondary dendrite arm spacing.
The left part of the cracking criterion relates to the thermo-mechanical
aspect of the total problem. It represents the mechanical loading of the liquid
film of segregates. The value of δacc
max can be calculated only numerically
because of the nonlinearity in material properties and the complex
boundary conditions.
It should be emphasized that the solution of the thermo-mechanical
problem directly depends on the solution of the metallurgical one. For this
reason, special attention must be given to the correct evaluation of the real
solidification range, since the real solidification range defines the length of
the mushy zone and, therefore, the length of the liquid film. A proper preliminary
study must be carried out also for the correct evaluation of the
solid phases distribution in the mushy zone, since the last one governs the
release of the latent heat and, therefore, the distributions of temperature,
stress and accumulated strain.
The right part of the cracking criterion relates to the metallurgical aspect
of the total problem. It represents the ability of the liquid film of segregates
to accumulate tensile strain without tearing. The value of the δacc
cr
depends on the thickness and on the morphology of the liquid film. The
thickness is defined by the fraction of the residual liquid phase in the microstructure.
The morphology is defined by the geometrical microstructure
parameter, i.e. by the primary and secondary dendrite arm spacing. This
dependence of the critical strain on the parameters of the microstructure
can be expressed as follows:
δacc
cr = δacc
cr(fL, λ1, λ2), (2)
where fL is the fraction of the residual liquid, λ1 and λ2 are the primary
and secondary dendrite arm spacing, respectively.
On the basis of simple considerations of the accumulation mechanism it
can be expected that the fraction of the residual liquid plays a primary role
for the ability of the liquid film to accumulate strain without tearing. A
234 Modeling and Simulation
higher liquid fraction in the microstructure leads to better strain accumulation
and, therefore, to better resistance to solidification cracking.
Concerning the role of the geometric parameters of the microstructure it
can be supposed that under the same distribution of the liquid fraction the
coarser microstructures have a better ability of strain accumulation, i.e. a
lower cracking susceptibility. More exact definition of the functional dependence
of the critical displacement on the microstructure parameters is
the subject of future development.
Direct calculation of the critical displacement will be possible on the basis
of preliminary evaluation of the microstructure parameters. This can be
done for the multi-component alloys using the modern numerical models
[29]. The understanding of the interrelations between the microstructure
parameters and the cracking resistance expressed in Eq. 2 can be effectively
used for welding technology optimization by targeted development
of filler materials. Such optimization can be made on the basis of thermodynamic
calculations for multi-component systems [30], aimed at maximization
of the residual liquid in the microstructure at the end of weld metal
solidification.
Simulation
A simulation example of crack initiation and growth is shown in Fig. 8.
The simulation is carried out for the case of laser beam welding close to
the free edge described in the section “Experiments” (Fig. 2).
The thermo-mechanical problem is solved using the finite element
method. For proper simulation of strain accumulation, the mechanical
properties of the liquid film are assigned to the finite elements at the weld
centreline (Fig. 9). For these elements, the value of the accumulated displacement
is checked at each time step of the calculation. As soon as this
value becomes critical, the material properties are changed to the properties
of air (Fig. 9).
Particular attention is given to properly taking account of the latent heat
release in accordance with the fraction of the liquid phase in the mushy
zone. The real solidification range as well as the liquid phase fraction is
calculated on the basis of the Scheil-Gulliver model [31].
Two isotherms, i.e. the real liquidus and the real solidus, outline the
mushy zone. As seen, the crack appears in the inner part of the mushy
zone. Once formed, the crack creates a thermal barrier which increases the
asymmetry of the temperature field.
Integrated Mechanical-Metallurgical Approach to Modeling 235
Fig. 8. Numerical simulation of crack initiation and growth:
temperature fields at different steps of calculation
Fig. 9. Taking into account the effect of strain localization due to
proper manipulation of the material properties during the numerical simulation
236 Modeling and Simulation
The position of the crack tip within the mushy zone is not constant; it
changes in accordance with the local thermo-mechanical conditions. At the
end of the weld seam, close to the right sample edge, the crack is forced
out of the mushy zone, i.e. its growth is interrupted. The same behavior is
also observed in all experiments. The reason for the growth interruption is
the decrease of accumulated strain occurring due to overheating and the
correspondent decrease of material strength in the near-edge region of the
sample.
Concerning the Healing Effect
A controversial question in the theory of solidification cracking is the role
of the healing effect. Here, we try to analyze the possibilities for healing a
solidification crack due to the melt from the weld pool on the basis of the
reconstructed temperature field for the real weld.
Fig. 10 shows the reconstructed mushy zone for the real weld produced
on the Al-Si binary alloy containing 3 % of silicon. As can be seen, the
length of the mushy zone is approximately 2.3 mm (Fig. 10a). The photographs
of the microstructure represented in Figs. 10b–c were obtained at
the centerline of the weld. One of them represents the region with the solidification
crack, another one shows the region of the weld where crack
growth was interrupted.
Due to the similar solidification conditions, the parameters of both observed
microstructures are equal. The value of the primary dendrite arm
spacing is in the range of 5 μm. It means that more than four hundred dendrites
can be placed on the length corresponding to the distance between
the isotherms of liquidus and solidus at the weld centerline.
It should be emphasized that during solidification the tips of dendrites
growing from both sides of the weld meet each other under the temperature,
which is close to the liquidus one, i.e. in the region of the liquidus
isotherm. Such situation can be illustrated using the real microstructure as
shown in Fig. 10b. Therefore, during solidification there must be a dense
dendritic network consisting of several hundreds of dendrites in the mushy
zone at the weld centerline.
This means that if healing would be even possible in the region close to
the liquidus isotherm, it would surely be hampered by the dense dendritic
net in the interior regions of the mushy zone. In this region, far from
liquidus, tearing of the liquid film of segregates caused by the strain accumulation
can freely take place, involving solidification cracking as
illustrated by the real microstructure shown in Fig. 10c.
Integrated Mechanical-Metallurgical Approach to Modeling 237
Fig. 10. To the role of the healing effect: (a) reconstructed shape of the mushy
zone for the real weld produced in the binary alloy Al-3 % Si, (b) microstructure
in this weld observed in the region where the crack growth was interrupted, and
(c) microstructure of the same weld observed in a region with the solidification
crack. Both microstructures illustrate the density of the dendritic network in the
mushy zone (app. 200 dendrites at 1 mm length) which should hamper the healing
effect
Example of Application: Multi-Beam Laser Welding
The idea to use additional heat sources as a means to prevent solidification
cracking in welds was suggested in the 1970ies [32, 33]. On the basis of
FEM-simulations it was demonstrated [32, 34] that introducing additional
heating “in a right place and at the right time” [32] leads to “beneficial”
compressive stress (or strain) in regions which are critical with respect to
solidification cracking. Despite the apparent simplicity of the suggested
idea, additional heating has still not found the expected industrial applica238
Modeling and Simulation
tion. The determination of “the right place and the right time”, i.e. the optimal
position, the size and the power of the additional heating spots,
seems to be the most important problem hindering the practical application
of this technique.
As will be shown below, the application of the presented modeling approach
can be very helpful for industrial realization of multi-beam techniques
for crack-free welding.
Fig. 11. Experimental validation of the multi-beam technique: (a) generation of
solidification cracking due to the small distance to the free edge and (b) suppression
of solidification cracking under the same conditions due to the application of
the additional laser beam
The experimental laser welding close to the free edge described above
was chosen for validation of the numerical calculations (Fig. 11). The idea
was to define the distance from the free edge which assures the initiation
and the stable growth of the solidification crack as shown in Fig. 11a.
Then, the optimal position and the power of the additional laser beam,
that will ensure the suppression of crack initiation, were found using the
numerical simulation (Fig. 11b). The experiments were carried out on Alalloy
AA6056 with a plate thickness of 2 mm.
Integrated Mechanical-Metallurgical Approach to Modeling 239
Fig. 12. Realization of the multi-beam technique: (a) preliminary numerical simulation
of the temperature fields and (b) samples produced under the same conditions
using single-beam (above) and double-beam (below) laser welding
Fig. 12 shows the results of the preliminary numerical simulations (temperature
fields for single- and double-beam welding) together with the results
of the experimental validation. As can be seen, the additional beam,
which moves parallel to the main laser beam at a distance of 20 mm, can
240 Modeling and Simulation
effectively be used for the suppression of crack initiation. The power of the
additional beam is significantly lower than the power of the main beam. In
this special case, the positive effect is achieved at the additional beam
power which was higher than 40 % of the main beam power.
Fig. 13. Dynamics of strain accumulation: differences between the presented
modeling approach and Prokhorov’s model
Fig. 13 represents the dynamics of the accumulated displacement calculated
for single- and multi-beam laser welding on the basis of the presented
modeling approach as well as on the basis of the model of Prokhorov. The
curves of the accumulated displacement were calculated for the initial
point of the plate at the weld centerline, i.e. for the point where crack initiation
can be expected. In all cases, the strain accumulation was calculated
for the same temperature interval, from the real liquidus until the real
solidus (the last one corresponds to the eutectic temperature).
The calculated maximal values were compared with the value of the
critical displacement determined on the basis of the known critical distance
from the free edge using the calibration of the experiment and the numerical
simulation.
Integrated Mechanical-Metallurgical Approach to Modeling 241
As can be seen from Fig. 13, according to the present approach, the use
of the additional beam leads to a continuous reduction of the displacement
accumulation. The maximal value of the accumulated displacement for
multi-beam laser welding is less then the critical one.
Prokhorov’ model neglects the effect of strain localization. For this reason,
it predicts very small, physically not explainable values of the accumulated
displacement needed for solidification cracking (single-beam
welding). Although Prokhorov’s model reveals a similar tendency to the
reduction of a accumulated strain by multi-beam welding, the difference
between both curves is not significant enough to be helpful for the prediction
of the threshold from solidification cracking to crack-free welding.
Summary
Based on experimental observations, an approach to modeling of solidification
cracking in welds is presented which integrates both the thermomechanical
and the metallurgical aspects of the cracking phenomenon. The
presented approach implies the solution of the thermo-mechanical problem
taking into account the effects of the localization and the accumulation of
displacements in the intergranular liquid film of the segregating alloying
elements. The importance of proper preliminary evaluation of the real solidification
range and the phase distribution within the mushy zone is emphasized.
The high potential of the presented approach for the development
of crack-free welding techniques in industry is demonstrated on the
basis of the realization of multi-beam laser welding.
Acknowledgements
This work was carried out within the framework of the project ”Optimisation
of weldability” sponsored by the Bavarian Government, the Foundation
of Oberfranken, Allianz-Zentrum für Technik GmbH, AUDI AG,
EADS Deutschland GmbH, Linde AG und MTU Aero Engines GmbH.
The authors would like to thank all partners for their active support as well
as for their very useful discussions. Special thank is addressed to
Dr. C. Heimerdinger (EADS Deutschland GmbH) for his creative coworking
in the experimental realization of the multi-beam technique.
242 Modeling and Simulation
References
1. Singer ARE, Jennings PH (1946) Hot-Shortness of the Aluminium-Silicon Alloys
of Commercial Purity. Journal of The Institute of Metals 73: 197–212
2. Singer ARE, Jennings PH (1947) Hot-Shortness of Some Aluminium-Iron-
Silicon Alloys of High Purity. Journal of The Institute of Metals 73: 273–284
3. Jennings PH, Singer ARE, Pumphrey WI (1948) Hot-Shortness of Some
High-Purity Alloys in the Systems Aluminium-Copper-Silicon and Aluminium-
Magnesium-Silicon. Journal of The Institute of Metals 74: 227–248
4. Pumphrey WI, Lyons JV (1948) Cracking During the Casting and Welding of
the More Common Binary Aluminium Alloys. Journal of The Institute of
Metals 74: 439–455
5. Pellini WS (1952) Strain Theory of Hot Tearing Foundry, pp 125–133, 192,
194, 196, 199
6. Prokhorov NN (1952) Hot Cracking During Welding (in Russian). Mashgiz,
Moscow
7. Bochvar AA, Rykalin NN, Prokhorov NN, Novikov II, Movchan BA (1960)
The Question of “Hot” (Crystallisation) Cracks. Welding Production 10: 5–7
8. Prokhorov NN (1962) The Technological Strength of Metals While Crystallising
During Welding. Welding Production 4: 1–8
9. Matsuda F, Nakagawa H, Nakata K, Kohmoto H, Honda Y (1983) Quantitative
Evaluation of Solidification Brittleness of Weld Metal During Solidification
by Means of In-situ Observation and Measurement (Report I). Transactions
of JWRI 12 (1): 65–72
10. Matsuda F, Nakagawa H, Tomita S (1987) Quantitative Evaluation of Solidification
Brittleness of Weld Metal During Solidification by Means of In-situ
Observation and Measurement (Report IV). Transactions of JWRI 16(2): 90–
95
11. Senda T, Matsuda F, Takano G, Watanabe K, Kobayashi T, Matzusaka T
(1971) Fundamental Investigations on Solidification Crack Susceptibility for
Weld Metals with Trans-Varestraint Test. Transactions of the Japan Welding
Society 2(2): 45–66
12. Arata Y, Matsuda F, Nakata K, Sasaki I (1976) Solidification Crack Susceptibility
of Aluminium Alloy Weld Metals I. Transactions of JWRI 5(2): 53–67
13. Arata Y, Matsuda F, Nakata K, Sinozaki K (1977) Solidification Crack Susceptibility
of Aluminium Alloy Weld Metals II. Transactions of JWRI 6(1):
91–104
14. Jonsson M, Karlsson L, Lindgren LE (1984) Thermal Stresses, Plate Motion
and Hot Cracking in Butt-Welding. In: Mechanical Behaviour of Materials
IV. Pergamon Press, New York, pp 273-279
15. Dike JJ, Brooks JA, Li M (1988) Comparison of Failure Criteria in Weld Solidification
Cracking Simulations. In: Mathematical Modelling of Weld Phenomena
4 (ed Cerjak). IOM Communications Ltd, London, pp 199–222
Integrated Mechanical-Metallurgical Approach to Modeling 243
16. Feng Z (1994) A Computational Analysis of Thermal and Mechanical Conditions
for Weld Metal Solidification Cracking. Welding in the World 33(5):
340–347
17. Zacharia T (1994) Dynamic Stresses in Weld Metal Hot Cracking. Welding
Journal 73: 164-172
18. Dike JJ, Brooks JA, Krafcik JS (1996) Finite Element Modelling and Verification
of Thermal-Mechanical Behaviour in the Weld Pool Region. In: Trends
in Welding Research (ed Smartt HB et al). ASM Intern, pp 159–164
19. Feng Z, Zaharia T, David SA (1996) On the Thermomechanical Conditions
for Weld Metal Solidification Cracking. In: Mathematical Modelling of Weld
Phenomena 3 (ed Cerjak), pp 114–147
20. Feng Z, Zaharia T, David SA (1997) Thermal Stress Development in a Nickel
Based Superalloy during Weldability Test. Welding Journal 76: 470–483
21. Makhnenko VI, Velikoivanenko EA, Rozynka GF, Musiyachenko AV, Pivtorak
NI, Seiffarth P (1998) A Computer Program for Prediction the Zones
with the Risk of Formation of Hot Cracks in Welding with Deep Penetration.
The Paton Welding Journal 10(2): 57–65
22. Weise S (1998) Heißrißbildung beim Laserstrahlschweißen von Baustählen.
BIAS-Verlag, Bremen
23. Herold H, Streitenberger M, Pchennikov A (2001) Modelling of the PRV-Test
to Examine the Origin of Different Hot Cracking Types. In: Mathematical
Modelling of Weld Phenomena 5 (ed Cerjak et al). IOM Communications Ltd,
London, pp 783–792
24. Shibahara M, Serizawa H, Murakawa H (2000) Finite Element Method for
Hot Cracking Analysis under Welding Using Temperature Dependent Interface
Element. In: Modelling of Casting, Welding and Advanced Solidification
Processes IX (ed Sahm PR et al). Shaker-Verlag, Aachen, pp 844–851
25. Shibahara M, Serizawa H, Murakawa H (2001) Finite Element Method for
Hot Cracking Analysis Using Temperature Dependent Interface Element. In:
Mathematical Modelling of Weld Phenomena 5 (ed Cerjak et al). IOM Communications
Ltd, London, pp 253–267
26. Bergmann HW, Hilbinger RM (1998) Numerical Simulation of Centre Line
Hot Cracks in Laser Beam Welding of Aluminium Close to the Sheet Edge.
In: Mathematical Modelling of Weld Phenomena 4 (ed Cerjak). IOM Communications
Ltd, London, pp 658-668
27. Hilbinger RM, Bergmann HW, Köhler W, Palm F (2001) Considering of Dynamic
Mechanical Boundary Conditions in the Characterisation of a Hot
Cracking Test by Means of Numerical Simulation. In: Mathematical Modelling
of Weld Phenomena 5 (ed Cerjak et al). IOM Communications Ltd, London,
pp 847–862
28. Hilbinger RM (2000) Heißrissbildung beim Schweißen von Aluminium in
Blechrandlage. Universität Bayreuth, Bayreuth
29. Diepers HJ (2003) Simulation des Primärabstandes gerichtet erstarrter Dendriten
mit der Phasenfeldmethode. Shaker Verlag, Aachen
244 Modeling and Simulation
30. Andersson JO, Helander T, Höglund L, Shi P, Sundman B (2002) Thermo-
Calc & DICTRA. Computational Tools For Materials Science, Calphad, 26:
273–312
31. Saunders N, Miodownik AP (1998) CALPHAD (Calculation of Phase Diagrams):
A Comprehensive Guide. Pergamon Materials Series vol 1. Elsevier
Science, Oxford, New York, Tokyo
32. Akesson B, Karlsson L (1976) Prevention of Hot Cracking of Butt Welds in
Steel Panels by Controlled Additional Heating of the Panels. Welding Research
International 6 (5): 35–52
33. Shumilin VG, Karkhin VA, Rakhman MI, Gatovsky KM (1980) A Technique
of Arc Welding. Patent No 1109280, 1980, USSR
34. Herold H, Streitenberger M, Pchennikov A, Makarov E (1999) Modelling of
One Sided Welding to Describe Hot Cracking at the End of Longer Butt Weld
Seams. Welding in the World 43(2): 56–64
Influence of the Weld Pool Geometry on
Solidification Crack Formation
M. Wolf, H. Schobbert, Th. Böllinghaus
Federal Institute for Materials Research and Testing, Berlin, Germany
Abstract
Numerous externally loaded as well as self-restraint hot cracking test procedures
have been developed over the years to evaluate soldidification
crack phenomena. But the interpretation and the subsequent transfer of test
results towards component weldments has still to be regarded as a decisive
challenge. Some progress has been achieved by hypothesizing crack / no
crack criteria referring to uniform and uniaxial loading of the solidification
front during welding. But, as the present results show, solidification cracking
significantly depends on the geometry of the solidification front, which
has so far not been considered in the respective criteria. In this contribution,
the position of solidification cracks is correlated with the positiondependent
strain rates along the solidification front and with the positiondependent
curvature of the weld pool. Especially at positions of high curvature,
an increased rate of shrinkage can be predicted on the basis of a
geometrical Rate of Feeding (ROF) Rate of Shrinkage (ROS) model, corresponding
to the measured positions of the solidification cracks. The assumption
of position-dependent strain and strain rates along the solidification
front during Varestraint-Testing has been confirmed by means of
numerical simulations.
Introduction
Solidification cracks originate and grow as interdendritic or intergranular
material separations directly behind the moving weld pool and preferably
occur in fcc-solidifying alloys. At the end of the welding process, their
dimensions can reach from a few micrometers to the whole length of the
weld.
246 Modeling and Simulation
Fig. 1. Solidification crack of a Ni-base alloy
(German mat.-no.: 2.4889 – 2000 x)
Fig 1 shows a photograph of a solidification crack taken by a scanning
electron microscope. Typical of the surface morphology of solidification
cracks is the visibility of dendrite tips.
The occurrence of solidification cracks is determined by the thermomechanical
and metallurgical processes in the mushy zone surrounding the
weld pool and corresponding to a temperature range, in which molten as
well as solid material is coexistent. Solidification cracks are initiated, if the
thermomechanical strains caused by the time-dependent temperature distribution
during welding and the material-specific shrinkage during liquid/
solid phase transition cannot be compensated by the ductility of the
weld metal or by reflow of melt.
Although the most decisive critical factors such as the local strain rates,
local shrinkage rates and the critical temperature range could be identified,
the solidification cracking susceptibility of materials can still only be
ranked in relation to other materials, referring to equal welding parameters
and other boundary conditions. The reason for the difficulties in assessing
the materials solidification crack susceptibility by absolute values is its
concatenation to a complex system of feed-back processes. As for example
technological parameters like the welding speed can simultaneously influence
the thermomechanically induced material displacements in the vicinity
of the weld pool as well as metallurgical processes like the morphology
of the solidification front, solidification cracking can often not be related
to a definite reason.
Influence of the Weld Pool Geometry on Solidification Crack Formation 247
From the metallurgical point of view, solidification cracking can be ascribed
to a great variety of different reasons. For example, some iron companions
such as sulphur, phosphor or boron can lower the solidification
cracking resistance of stainless steels drastically. Moreover, some stainless
steels with a chemical composition nearby the eutectic rim can lose their
relatively high solidification cracking resistance, if a critical solidification
rate is exceeded and their primary solidification mode changes from ferritic
to austenitic [1].
Although plenty of metallurgical factors favoring or reducing the tendency
to solidification cracking are known, the hot cracking behavior can
still very often not even approximately be prognosticated on the basis of
the materials composition. For this reason and because of the non-linear
processes of solidification crack initiation, hot cracking tests have to be
performed, in which a material is exposed to welding conditions. For a
subsequent transfer of hot cracking test results to component weldments,
crack / no crack criteria are required, by which solidification cracking can
be predicted on the basis of significant input parameters. However, universal
solidification crack criteria could not have been established so far.
Two solidification cracking criteria have gained the most acceptance on
international level: The criterion of the brittleness temperature range
(BTR) related to the solidification cracking theory of Prokhorov [2] and
the so called Rate of Feeding (ROF) Rate of Shrinkage (ROS) model with
the criterion ROF < ROS [3].
The BTR criterion of Prokhorov, which has been applied and partly
overworked, for example, by Matsuda et al. [4] and Herold et al. [5], bases
on a continuum-mechanical approach. Prokhorov postulates a defined
temperature range in which the ductility of the solidifying material is significantly
low (comparable to the ductility dip of a material at much lower
temperatures, causing the ductility dip cracks). Within this temperature
range, which reaches from a temperature between liquidus and solidus to a
temperature below solidus, the material is sensitive to solidification cracking.
A material can thus be characterized according to the extension of this
temperature range in combination with the critical strain or to a critical
temperature-related strain increase dε/dT.
Solidification cracking never occurs in a completely solidified state of
the material, so that the designation “below solidus” refers to dispersed residual
melts which might be present at the grain boundaries. According to
the literature, different shapes of the ductility curve can be found referring
either to an increase or a decrease of ductility when the solidification process
reaches the solidus temperature. It can only be assumed in this connection
that if bridging between dendrites or grains is not present during solidification,
the minimum ductility is reached at the final stages of
248 Modeling and Simulation
solidification. If cooling favors the bridging between dendrites or grains,
then the ductility might finally increase at the solidus temperature. It becomes
clear in this connection that for a characterization of the solidification
cracking, not only the temperature range but also the shape and the
absolute values of the critical strain curve have to be determined.
The ROF-ROS theory refers to the competitive and dynamic process of
volume shrinkage caused by the liquid/solid phase transition and a leveling
reflow of the melt from the weld pool (Fig 2). The balance between the
material-specific volume shrinkage, which might moreover be superimposed
by thermomechanically induced strain rates (ROS), and the backfilling
by the melt (ROF) can mathematically be expressed in simplified terms
by ROF = ROS. If the maximum melt reflow is not sufficient for filling the
interdendritic voids or the interdendritic pressure becomes lower than the
cavitation pressure, then a solidification crack is initiated. The corresponding
solidification crack criterion can then be written as ROF < ROS.
A lot of theoretical work has been done for the calculation of ROF and
ROS and a good agreement between the prognosis and the experimental
data has been observed [6–8]. The ROF-ROS criterion has furthermore
been extended by the interfacial effect of Marangoni forces influencing the
ROF [9]. However it still has to be clarified, whether this criterion can be
applied to all solidification crack phenomena.
Fig. 2. Rate of Feeding (ROF) – Rate of Shrinkage (ROS) model
In a direct comparison, the BTR criterion seems to be better applicable
to intergranular solidification cracking which is especially promoted by residual
melt at the grain boundaries and a relatively low heat conductivity
of the solid. Since the shrinkage caused by the solidification of residual
Influence of the Weld Pool Geometry on Solidification Crack Formation 249
melt is very low and also a potential reflow of the melt is implausible due
to the length of the cracks, a continuum-mechanical approach like the BTR
seems to be suitable for describing intergranular solidification cracking
caused by residual melts.
The ROF-ROS criterion is more qualified for describing the solidification
cracking, which bases on permanent interdendritic volume
shrinkage during solidification directly at the solidification front. Both criteria
have in common that strain and strain rates transverse to the solidification
direction increase the tendency to solidification cracking. However,
as an advantage of the proposed ROF-ROS mechanism, the local strain
rate at the solidification front is accepted as the decisive factor whereas for
the BTR the accumulated strain is considered.
Externally Loaded Hot Cracking Tests
The hot cracking resistance of materials is usually ranked according to the
results of externally loaded hot cracking tests, in which a specimen is deformed
during welding or a simulated time-temperature cycle (Hot Tensile
Test) under defined conditions [10]. In order to minimize the materialspecific
internal (thermomechanical) loading, the geometry of the specimens
is usually simple. One reason why the externally loaded hot cracking
tests are widely spread in research is that defined strains and strain rates related
to the mechanical loading can be applied onto the specimens. Therefore
input variables comprising loading and welding parameters can be
assigned to the measurable output data like number, length and position of
solidification cracks which makes the results interpretable compared to the
self-restraint hot cracking tests.
Numerous externally loaded hot cracking tests exist, for which the
specimen loading (tensile loading, bending) as well as the interpretation of
the test results can be quite different. Usual output data are for example the
total crack length (TCL), the critical strain rate of the specimen and the
length of the longest hot crack (MCL). Due to this multitude of variations,
a comparison of the test results can only be done qualitatively and, as mentioned
before, a direct transfer of test results to real component welds is
still controversial.
Internationally, two of the most frequently applied externally loaded hot
cracking tests are the Varestraint-Test (VAriable-REstraint-Test) [11] and
the Transvarestraint-Test [12]. However, such tests do not account for a
shrinkage restraint as it can only be applied in a self-restrained test. Both
tests differ only in their bending direction.
250 Modeling and Simulation
For a Varestraint-Test (which is sometimes designated more precisely as
Longitudinal Varestraint Test), a specimen is bent longi-tudinally over a
mandrel within a short period of time during producing the test weld run.
For a Transvarestraint-Test, the specimen is bent transversely to the weld
direction. Both types of test have specific advantages: The Varestraint Test
is suitable for testing of all types of hot cracks (solidification cracks, liquation
cracks and ductility dip cracks). The results are interpreted by the total
crack length (TCL), the maximum crack length (MCL), the number of all
cracks or by the minimum strain at which hot cracks occur (εmin).
The Transvarestraint-Test is preferably applied for the determination of
the brittleness temperature range (BTR), since in most cases a centerline
crack is engendered representing the maximum crack length. If the temperature
regime in the centerline of the weld is determined by measurements
or by numerical simulations, the BTR can directly be calculated by
the length of the centerline crack. An advanced method for the assessment
of the solidification crack susceptibility is the determination of the Solidification
Cracking Temperature Range (SCTR), by which the maximum
crack distance (MCD) is especially regarded [13].
Referring to the solidification cracking criteria, the loading rate of a hot
cracking test has to be discussed in this context. In the Tranvarestraint-
Test, for example, very high bending speeds in comparison to the welding
speed are usually applied for determining the BTR. By this mechanical
shock-loading, the length of the centerline crack represents directly the
brittle material zone in the weld metal related to the momentary temperature
distribution. The strain rate above a certain level does not influence
the result anymore and can therefore be eliminated as a free parameter for
Transvarestraint testing.
But besides a few exceptions which occur, for example, from a break
open of tack welds, shock-loadings do usually not occur in real component
welds, so that the question might be raised of which loading rate is realistic
and comparable to the solidification crack formation in real component
welds. As a compromise between shock-loading and a possibly uncritical
loading rate an intermediate average strain rate of about 1–2 %/s is applied
to the MVT-Test used in this investigation [14]. Using such a loading rate
the BTR cannot be determined after the test by measurement of the longest
solidification crack, whereas the intermediate loading rates are in particular
suitable for testing the ROFROS criterion.
Fig. 3a–d shows a sequence of the solidification crack formation from a
high speed video taken during an MVT-Test in Transvarestraint mode. It
can be seen that the first fraction of the first crack (3.2 mm) forms up almost
instantaneously within a time interval of 0.4 seconds (Fig. 3b–c) by
an average growth speed of 8 mm/s.
Influence of the Weld Pool Geometry on Solidification Crack Formation 251
Fig. 3. Solidification crack formation in Alloy 602 CA during the MVT-Test;
Transvarestraint mode (4 %); shielding gas: Ar + 5 % H2; vweld = 1.4 mm/s, I =
171 A; U = 11.4 V
Obviously, a critical strain is exceeded at this instant, leading to an
abrupt crack formation compared to the welding speed of 1.4 mm/s. After
the sudden crack opening, the growing proceeds much slower. Within a
period of 1.4 seconds between Fig. 3c and Fig. 3d, the crack growth rearward
to the welding direction is very small, but the crack lengthens in forward
direction with the speed of the solidification front until straining is
over.
It can thus be concluded that the first fraction of the crack forms up depending
on the brittleness of the material zone behind the weld pool,
whereas a second fraction grows on account of insufficient melt reflow.
It has to be emphasized in this connection that this kind of crack
formation is not necessarily representative of any alloy. But as a compromise
for including all possibilities of crack formation and for a comprehensive
test result, an intermediate bending speed seems to be most appropriate.
252 Modeling and Simulation
Objective
As a contribution to the ongoing standardization discussion correlated with
the transfer of hot cracking test results towards component weldments and,
in particular, to clarification of the mechanisms and crack criteria behind
solidification crack phenomena, the influence of the weld pool geometry
on the formation of solidification cracks has been investigated. For this
purpose the Varestraint- as well as Transvarestraint-Test have been applied.
The weld pool geometry is formed by material- and process-specific
properties such as the heat conductivity of the solid, the melt flow within
the weld pool, the energy distribution of the heat source and the solidification
kinetics. It can be shown in this connection that the positiondependent
probability of solidification crack formation can be correlated to
geometrical parameters of the weld pool in combination with the global
loading direction.
Besides the theoretical fundamentals of the numerous solidification
cracking theories which mostly base only upon metallurgical effects measurements
as well as computations of the position-, time- and directiondependent
strains and strain rates in the surrounding of the weld pool are
very difficult due to the extreme temperature gradients and a complex
high-temperature material behavior. This is the reason why the results of
different test procedures might scatter broadly. But, it has to be emphasized
in this connection that the strains and strain rates, respectively, occurring
in the vicinity of the weld pool are generally the fundamental reason
for solidification crack formation and represent the most explicit test
parameter in every externally loaded hot cracking test to correlate laboratory
results and practical cracking.
Therefore, the local material displacements leading to solidification
cracks have to be quantified for a transfer of solidification cracking test results.
On this account, in particular the second part of this contribution
deals with numerical simulation of the position dependent material displacements
in the surrounding of a weld pool during MVT testing.
Experimental Procedure
The following investigations have been carried out using the Modified
Varestraint Transvarestraint-Test (MVT-Test) already mentioned above
(Fig. 4). The major feature of the MVT-Test is to change the loading direction
(transversely or longitudinally to weld direction) only by replacing the
dies (Fig. 5).
Influence of the Weld Pool Geometry on Solidification Crack Formation 253
Fig. 4. MVT-Hot Cracking Test
Fig. 5. MVT - Varestraint mode (a); MVT - Transvarestraint mode (b)
Therefore, by means of the MVT-Test it is possible to perform alternatively
Varestraint- or Transvarestraint Tests leaving relevant parameters
like the geometry of the specimens or the timing of the test procedure unaffected.
254 Modeling and Simulation
The investigations have been performed based on the observation that
under defined MVT-Test conditions, solidification cracks often form up
symmetrically (Fig. 6). A symmetry can be observed preferentially for a
small number of solidification cracks and therefore with minor crack formation
interaction. From the symmetry it can thus be concluded that the
weld pool geometry as well as the position-dependent strain or strain rates
must be considered as a critical factor for solidification cracking, since
their position correlated with the weld pool geometry is deterministic.
Fig. 6. Crack initiation site after MVT testing; TIG bead-on-plate weld; Varestraint
mode; 4 % total strain
For quantifying the influence of the weld pool geometry on the position
of solidification cracks, the welding parameters such as heat input per unit
length and welding speed have been systematically varied in MVT experiments.
The material used in these experiments was AISI 316 L. It was
found in pre-tests that the shape of the end craters of bead-on-plate welds
agree well with the solidification front at the surface during welding.
Therefore the specific end crater shape of every weld has been used for the
determination of the weld pool geometry and solidification front, respectively.
Fig. 7 shows exemplarily photographs of end craters and the resulting
solidification cracks after MVT-testing. For these kinds of experiments, a
quantification of the test results has been carried out by measuring the position
of solidification cracks in correlation with the weld pool geometry.
The results can be described as follows: For a slow welding speed, which
means that the weld pool geometry is approximately circular, the solidification
cracks preferably appeared nearby the fusion line (Fig. 7a).
Influence of the Weld Pool Geometry on Solidification Crack Formation 255
Fig. 7. Alloy AISI 316 L; Correlation of weld pool geometry with solidification
cracking; MVT Varestraint mode; total strain = 2.3 %; shielding gas: Ar + 5 % N2
256 Modeling and Simulation
When the welding speed was increased and the weld pool geometry became
tapered, then the solidification cracks concentrated more and more to
positions nearby the centerline, and even a small centerline crack occurred
(Fig. 7b). When the weld pool geometry was sharply tapered and mathematically
unsteady at the centerline, then a single centerline crack occurred
(Fig. 7c). The geometries of the end craters were measured and fitted by
mathematical functions basing on the ellipsoid equation and correction
functions. By means of end crater functions, the curvature along the solidification
front as well as the normalized strain rate function occurring
during bending were calculated and plotted in a common diagram as
shown on the right hand side in Fig. 7a to 7c.
It has to be remarked in this connection, that the position dependent
strain rates were calculated transverse to the solidification direction, which
means transverse to a potential crack direction, and base only on the
geometry of the weld pool. Thermomechanical effects occurring during a
weld process are not considered so far, but their influence will be discussed
below. Moreover, the positions and lengths of the solidification
cracks were charted in the respective diagrams. Following tentative conclusions
can be drawn: Due to testing in Varestraint mode, which means
that the specimen is bent longitudinally to the welding direction, the strain
rates transverse to the potential crack are monotonic increasing from the
centerline to the fusion line. The trend of the curvature functions highly
depends on the geometry function of the solidification front. Correlating
the positions of the solidification cracks with the position-dependent curvature
and strain rate, it can be concluded that a solidification crack occurs
preferably at positions, where a relatively high curvature and a relatively
high strain rate coincide.
A significant correlation of the weld pool geometry in form of the curvature
with the solidification cracking becomes in particular clear on the
basis of the example illustrated in Fig. 7c. Although theoretically the strain
rates at or in the direct vicinity of the centerline were very low, a single
centerline crack occurred which originated slightly next to the centerline.
The fact that straining longitudinally to the weld direction can cause centerline
cracks, even though the mechanical loading is lowest at this position,
means that the influence of the respective weld pool geometry on
solidification cracking is determinant in some cases.
Fig. 8 shows an example of an MVT-Test result from the Transvarestraint
mode, where the welding parameters correspond as far as possible
to the example in the second row of Fig. 7. It can be seen that, due to the
change of the bending direction, the strain rate function is oppositely directed
and that the positions of the solidification cracks are shifted towards
the centerline.
Influence of the Weld Pool Geometry on Solidification Crack Formation 257
Fig. 8. Alloy AISI 316 L; Correlation of weld pool geometry with solidification
cracking; MVT Transvarestraint mode; total strain = 2.3 %; shielding gas: Ar +
5 % N2
In this example, the strain rate has to be regarded as the determinant factor
for solidification cracking, since highest strain rates occurred mathematically
at the crack positions. It can thus be concluded, that both the position-
dependent strain rate and the position dependent curvature influence
solidification cracking. The effect of the curvature on solidification crack
formation is discussed in the next paragraph.
Rate of Area Shrinkage Correlated with the Weld Pool
Geometry
The influence of the weld pool geometry on the position of solidification
cracks has been theoretically analyzed by a geometric model. In this twodimensional
model the rate of shrinkage by phase transition at the solidification
front is related to the position-dependent growth direction and speed
of the solidification front.
As a first approach it bases on strain rate free solidification. The function
of the growth path (Fig. 9), which represents the idealized trace of an
arbitrary point at the solidification front shifting from the fusion line to the
centerline, can be calculated from the end crater function representing the
solidification front at an arbitrary point in time, as described below. It is
assumed that cells with a length directly corresponding to the local extension
of the solidification interval, protrude into the melt.
258 Modeling and Simulation
Fig. 9. Calculation of the time-dependent solidification front evolution; welding
direction -y
The root point, for example, of such a cell is differentially shifted by:
ds = vs⋅cos(ϕ(x))⋅dt (1)
perpendicular to the solidification front, whereby ϕ(x) is the position
dependent angle between the welding direction and the normal of the solidification
front and vs is the welding speed.
From Eq. 1 follows:
vs⋅dt = 2⋅dx / sin(2⋅ϕ(x)) (2)
By integration of Eq. 2 the time-dependent shifting function of the relevant
root point as well as of the tip of a cell can be calculated.
For the determination of the area change per time correlated with the
proceeding of the solidification front, the change of the area of liquid between
two adjacent cells caused by the phase transition involving a change
of the specific density ρl => ρs has to be calculated (Fig. 10).
As a first approximation this area shrinkage is set as a measure ROSγ for
the ROS. Two arbitrary adjacent cells start growing with an offset of λ in
x-direction.
Influence of the Weld Pool Geometry on Solidification Crack Formation 259
Fig. 10. Change of area by the proceeding of a bent solidification front
The functions of the liquidus and solidus isotherms y = y(x) are set as
equal but they are shifted by the solidification interval moving with the
speed of welding in y-direction.
After the solidification of dFS within the period dt, there exists a residual
liquid portion FL(t) - γdFS, whereby γ = ρs/ρl. With the shrinkage of dFS in
correlation with the corresponding liquid portion, it follows for the change
of area:
ROSγ dt = (γ - 1)⋅dFS(t) (3)
and
ROSγ = (γ - 1) ⋅ dFS/dt (4)
For a verification of the geometric ROF-ROS model, the end crater
function of Fig. 7b has been used which exhibits a significant curvature at
a distance of about 3 mm from the centerline.
The position-dependent ROS function, for which a value for the cell distance
λ was estimated, is shown in Fig. 11.
260 Modeling and Simulation
Fig. 11. Position-dependent area shrinkage per time ~ ROS referring to Fig. 5b
The ROS function shows a peak at the position where the solidification
front exhibits a high curvature and, moreover, this position corresponds to
a section of increased solidification cracking. Referring to the ROF-ROS
model it can thus be concluded that in particular at positions of high curvature
the probability of ROF < ROS is increased. As a first approximation
of the ROF, the position-dependent function of the opening length of the
cells representing the geometric obstacle for the melt flow was calculated.
This function increases monotonically from the centerline to the fusion
line, but a quantitatively critical intersection to the ROS function can so far
not be deduced.
From these calculations it can be concluded that at highly bent solidification
fronts an increased rate of shrinkage occurs requiring an increased
ROF. Since the ROF can be assumed to be only slightly linearly increasing
along the solidification front, the probability of solidification cracking
must be increased at positions of high curvature.
Numerical Simulation of the MVT-Tests
Solidification cracks generated in an externally loaded hot cracking test
usually result from a superimposition of local thermomechanical and
global mechanical strains and strain rates, respectively. For a correlation of
Influence of the Weld Pool Geometry on Solidification Crack Formation 261
the position-, time- and direction-dependent material displacements leading
to solidification crack formation, numerical simulations are most appropriate.
With the help of computer technology, non-linear material properties,
heat source characteristics and experimental boundary conditions,
for example, can be considered. The software chosen for the numerical
analysis of the MVT-Test was Ansys 6.0.
Fig. 12. MVT model of the relevant components: specimen, radius and die
Fig. 12 shows the discretized model for the numerical simulation of the
MVT-Test in Varestraint mode. In order to calibrate and verify the
material properties, the numerical model, the experimental boundary conditions
and the modeled heat transfer between arc and specimen the temperature
regimes at defined positions were measured by means of thermocouples
for different conditions and compared to the results of the
numerical simulations.
Fig. 13 represents such a comparison, in which the temperature regimes
at four positions at the surface of a MVT specimen (Alloy 602 CA) were
simultaneously measured during a TIG weld run without any mechanical
loading. By an iterative variation of the energy efficiency used in the numerical
simulations the computed curves could be adapted to the measured
curves by which an average energy efficiency of η = 0.66 was determined
for the TIG-weld process of the MVT test.
262 Modeling and Simulation
Fig. 13. Temperature regimes at the specimen surface compared to the numerical
simulation by means of Ansys; Alloy 602 CA; I = 205 A; U = 11.5 V; vweld =
3 mm/s; η = 0.66
It should be noted in this connection that the energy efficiency refers directly
to the welding process and that the energy loss of the welding apparatus
is not considered.
Fig. 14 depicts another comparison between experimental and simulated
temperatures. The thermocouples were placed in a row 5 mm below the
surface in the symmetry line parallel to the welding direction. The temperature
regimes were recorded during a real MVT-Test in order to take
account of the increased heat dissipation taking place via the dies during
bending. For quantifying the influence of the heat flow via the dies, numerical
simulations were performed with and without consideration of heat
dissipation via the contact area.
The results show that for Alloy 602 CA and a specimens thickness of 10
mm significant deviation of the simulated temperature regimes occurs only
at a temperature below 700 °C. Since solidification cracking occurs at
much higher temperatures, the heat dissipation via the dies can be neglected
for the numerical simulation and has no direct influence on the
MVT-Test results from standard size specimens.
The simulation of the time-dependent temperature distribution represents
the basis for subsequent thermomechanical simulations in which the
strains and strain rates in the vicinity of the weld pool can be computed.
Influence of the Weld Pool Geometry on Solidification Crack Formation 263
Fig. 14. Temperature regimes 5 mm below the surface; with / without heat dissipation
via the dies; Alloy 602 CA; I = 194 A; U = 10.8 V; vweld = 3 mm/s;
η = 0.66
For the numerical simulation of the MVT-Test, the test procedure is divided
into three steps: In the first step, the TIG bead-on-pate welding is
simulated from the edge of the specimen to the middle without any external
loading. In the second step, for which the load stepping must be well
adjusted in the simulation, the specimen is bent while the TIG welding
continues.
Bending is finished within 1.5 to 3.5 seconds, depending on the respective
radius. In a third step, the heat source proceeds until the final position
is reached without any further external loading. In order to reduce the
computation time, the third step, which is irrelevant to solidification cracking,
is omitted in the numerical simulations.
Fig. 15a–b shows the strain distribution in x- and z-direction at the
specimen surface before bending. The weld pool is thermomechanically
represented by a moving void generated by dynamical “kill” and “rebirth”
of elements. The illustrations show that the weld pool is flanked by an area
of compressive strains caused by the heating of the respective material
zones. The weld metal exhibits compressive strains, too, which correlated
with the appearance of the TIG-weld after an MVT-Test, which was
slightly overcut.
Fig. 15c–d shows the strain distribution in the vicinity of the weld pool
directly after the bending process in x- and z-direction.
264 Modeling and Simulation
Fig. 15. Simulated transient strain distribution on the specimens surface; material
properties according to Alloy 602 CA; I = 225 A; U = 12.5 V; vweld = 1.8 mm/s; η
= 0.66; radius 125 mm (total strain 4 % ); Varestraint mode
The strain distribution becomes highly inhomogeneous by the mechanical
loading compared to that before. For solidification cracking close to the
fusion line, the strain distribution in z-direction is relevant representing the
strain direction transverse to a potential solidification crack.
The strain distribution in z-direction reveals that the maximal strain of
7.2 % occurs not directly at the solidification front but in a distance of
2.8 mm. Obviously, the global tensile strain by external loading is superimposed
by compressive strains present in the direct vicinity of the weld
pool leading to a slight decrease of the total strain. According to the numerical
simulation, the area close to the centerline directly behind the weld
pool experiences compressive strains in z-direction but compressions in zdirection
are not relevant for solidification crack formation at this position
Influence of the Weld Pool Geometry on Solidification Crack Formation 265
and might be correlated with the transverse contraction of the specimen by
bending in longitudinal direction.
Fig. 16. Position-dependent strain increase during bending at 3 positions correlating
to the weld pool geometry and a reference position at the specimens edge
Fig. 16 shows time-dependent strain curves and the corresponding temperature
regimes at different positions on the specimen surface located at
the crack initiation site which is the middle zone of the MVT specimen.
The positions were chosen according to the weld pool geometry based on
the respective thermal simulation, so that the weld pool runs over position
1 and position 2, whereby position 3 is only touched.
266 Modeling and Simulation
The strain functions refer to a strain direction transverse to a potential
crack, which means, that the function of position 1 represents the strain in
x-direction, positions 2 is a vectorial sum of the respective strains in z and
x-direction and position 3 represents solely the strain in z-direction. When
the weld pool approaches towards these positions, the material is firstly
under compressive strains. Then the weld pool runs over position 1–2, the
material becomes liquid, and no strain or stresses emerge. About two seconds
before the material becomes solid again at position 1 and position 2,
the mechanical bending process begins which can be seen by the strain increase
at position 3 and position 4. The latter one is located nearby the
upper edge of the specimen representing the function of the mere mechanical
strain increase since nearly no thermal influences occur at this position.
From Fig. 16 following conclusion can be drawn: At position 1 the
strain rate (0.015 /s) as well as the total strain after the bending process
(0.009 m/m) is lower compared to position 2 at which a strain rate of
0.021 /s and a total strain of 0.015 m/m occurs. At position 3 the total
strain accumulates most (0.068 m/m) and also the strain rate is highest
(0.038 /s) since the material is not melted and the relevant strain direction
is also the mechanical loading direction. The total strain at position 3 and
the strain rate at position 2 and 3 are higher than the total mechanical strain
(0.041 m/m) and the strain rate (0.018 /s) at position 4 since the material is
exposed to high temperatures in the vicinity of the weld pool leading to a
locally decrease of the mechanical strength.With respect to the Varestraint
mode of the MVT-Test and under the condition of a steady weld pool geometry
as shown in Fig. 15 it follows that strain and strain rate transverse
to a potential crack are lowest but not zero at the centerline. At the fusion
line, strain and strain rates transverse to a potential crack are absolutely
highest, which has been verified by further calculation of strains and strain
rates transverse to a potential crack between the fusion line and the centerline
along the solidification front. The assumption, that the normalized position-
dependent strain rate function can approximately be calculated by
the geometry of the weld pool and by the loading direction, as shown
above, can thus be confirmed by the results of the numerical simulations,
although the strain rate is not zero at the centerline.
Regarding the transfer of experimental results to component welds, absolute
values of critical strain and strain rate in the vicinity of a weld pool
are required, which can most precisely be obtained by numerical simulations.
Such values are given depending on the position along the weld pool
geometry. Since the influence of the melt flow on the weld pool shape has
not been considered in the numerical simulation, so far and the hightemperature
material properties can only be assumed, a non-restricted
transfer of the numerical results is still difficult. However, the weld pool
Influence of the Weld Pool Geometry on Solidification Crack Formation 267
shape of the presented numerical simulation is best comparable to the weld
pool shape shown in Fig.7a. Since solidification cracks were preferably
found at position 2 and position 3, the critical strain rate can be approximated
between 0.021 /s and 0.038 /s.
Conclusions and Perspectives
The geometric analysis of MVT-results confirmed that there has to be expected
significant dependency of the solidification crack formation on the
geometry of the weld pool. The solidification crack position has been correlated
with the local curvature of the solidification front and with the local
strain rates.
By means of the end crater functions, which represent the solidification
front during welding at the surface, theoretical solidification paths can be
calculated which are the basis for a geometric ROFROS-model. It can be
demonstrated that high curvatures of the solidification front cause a high
rate of shrinkage. This means that a high probability of crack formation
can be predicted at these positions, verified by MVT-results. By means of
FEM, the strains in the vicinity of the weld pool during MVT-Testing in
Varestraint mode have been analyzed. The results confirm the assumption
of a monotonically increased strain and strain rate distribution around the
weld pool during bending, but highest strains related to the MVT specimen
do not occur directly at the solidification front.
The correlation of solidification cracking and weld pool geometry can
especially be used for hybrid welding technologies, where by the specific
geometric arrangement of at least two heat sources the weld pool shape as
well as the thermomechanical strain and strain rates around the weld pool
can be varied within broad limits. Such optimization tasks can be performed
most conveniently by numerical simulations for which an appropriate
solidification cracking criterion is a prerequisite. Therefore, further
investigations are being pursued with a view to determining a solidification
crack criterion especially matched to the requirements of numerical
simulations.
References
1. Schobbert H, Böllinghaus Th, Wolf M (2003) Hot cracking resistance of laser
and hybrid welded austenitic stainless steels. Trends in Welding Research 6:
76–81
268 Modeling and Simulation
2. Prokhorov N (1962) The technological strength of metals while crystallizing
during welding. Welding Production 9(4): 1–8
3. Fuerer U (1977) In: Proc Int Sym Eng Alloys, Delft, The Netherlands, p 131
4. Matsuda F, Hashimoto T, Senda (1969) Fundamental investigation on solidification
structure in weld metal. Trans Nat Res Inst Metals 11: 43–58
5. Herold H, Streitenberger M, Pechnnikov A (2001) Prevention of centreline
solidification cracking during one-side welding. IIW-Doc IX-2000-01
6. Rappaz M, Drezet J M, Gremaud M (1999) A New Hot-Tearing Criterion.
Met Mater Trans 30A (2): 449–455
7. Rindler W et. al (2002) A Modified Hot Tearing Criterion for Steels. In:
Mathematical Modelling of Weld Phenomena 6, pp 819–835
8. Kannengiesser Th, McInerney T, Florian W, Böllinghaus Th, Cross C E
(2002) The Influence of Local Weld Deformation on Hot Cracking Susceptibility.
In: Mathematical Modelling of Weld Phenomena 6: 803–818
9. Olson DL, Edwards R (1998) Interfacially driven mass transport in joining
and coating technologies. In: Marangoni and Interfacial Phenomena in Materials
Processing (Book 692), The Royal Society, Cambridge, pp 113–126
10. prCEN ISO/TR 17641-3 Destructive tests on welds in metallic materials – Hot
cracking tests for weldments – Part 3 – Externally loaded tests11. Savage W,
Lundin C (1965) The Varestraint Test. Weld Res Suppl 10: 433–442
12. Mc Keown D (1970) Versatile Weld Metal Cracking Test. Met Constr 8: 351–
352
13. Lippold J.C, Lin, W (1996) Weldability of commercial Al-Cu-Li alloys. In:
Proc of ICAA5, Transtec Publications, pp 1685–1690
14. Wilken K, Kleistner H (1982) The MVT-Test – A novel procedure for testing
of the hot cracking susceptibility during welding (in German). Material und
Technik 1: 3–10
IV Testing and Standardization
Recent Developments in Weldability Testing
J.C. Lippold
Welding Engineering Program,
The Ohio State University, USA
Abstract
The term “weldability” can be used to describe a wide variety of material
characteristics when a material is subjected to welding. These may include
the physical and mechanical properties of the material, the ease with which
welding can be accomplished from a practitioner’s standpoint, or the ability
of the material to avoid metallurgical degradation, usually assessed by
its susceptibility to cracking during welding or subsequent heat treatment.
A number of weldability tests have been developed over the years to
evaluate and quantify material weldability. Many of these test techniques
have focused on the phenomenon known as “hot cracking”. This paper will
review the basic concepts associated with hot cracking and other forms of
elevated temperature cracking associated with welds and describe some recent
advances in the use of testing approaches to quantify susceptibility to
these forms of cracking. This description will include the use of the Varestraint
test, the cast pin tear test, and the Gleeble1 thermo-mechanical simulator
for quantifying cracking susceptibility and providing comparative
measures of weldability among alloys.
Introduction
The cracking of welded construction during fabrication has been a problem
since the first welding processes were first widely adopted in the early 20th
century. Cracking occurs for two reasons, 1) the presence of tensile stress,
and 2) a susceptible microstructure in the weld metal or heat-affected zone.
1 Gleeble is a registered trademark of DSI. Inc, Troy NY, USA.
272 Testing and Standardization
Since the elimination or control of stresses during welding is usually quite
difficult, a better approach is often control of the weldment microstructure.
The various forms of cracking are generally grouped by the temperature
range over which they occur. “Hot cracking” is associated with the presence
of liquid films along grain boundaries or elsewhere in the structure
and includes weld solidification cracking, HAZ liquation cracking, and
weld metal liquation cracking. “Warm cracking” occurs in the solid state at
temperatures between the solidus and approximately half the melting temperature
of the material and may occur either during fabrication or subsequent
postweld heat treatment. Various forms of warm cracking include
ductility-dip cracking, reheat cracking, strain-age cracking, and lamellar
cracking. Finally, “cold cracking” occurs at or near room temperature and
is usually associated with the presence of hydrogen and hydrogen-assisted
cracking mechanisms.
Over the past 50 years, numerous studies of weld cracking have been
published and various theories proposed to describe the cracking mechanisms.
In addition, a number of test techniques (well over 200) have been
developed to study and quantify cracking susceptibility.
In this paper, four types of weld cracking are reviewed and test procedures
described to quantify susceptibility, namely, 1) weld solidification
cracking, 2) HAZ liquation cracking, 3) ductility-dip cracking, and 4)
strain-age cracking. All three of these cracking phenomena have been observed
in austenitic stainless steels and Ni-base alloys, and often plague
the fabrication of these materials.
Weld Solidification Cracking
Weld solidification cracking occurs during the final stages of solidification
when tensile shrinkage stress accumulates and liquid films still persist
along solidification grain boundaries in the structure. If the imposed
shrinkage strain exceeds the inherent ductility of the solidifying weld
metal, cracking will occur.
The temperature range over which this occurs has been defined by Prokhorov
[1], Matsuda [2], and others as the Brittle Temperature Range
(BTR), where the ductility of the weld metal shows a minimum value
where weld solidification cracking is possible.
As a general rule, the wider the BTR, the more susceptible the material
is to solidification cracking. Thus, the ability to measure the BTR during
welding should provide some approximation of a material’s susceptibility
to weld solidification cracking.
Recent Developments in Weldability Testing 273
The Transverse Varestraint Test
The Varestraint test, in a variety of forms, has been used since the 1960’s
to quantify susceptibility to “hot cracking”, i.e. weld solidification and liquation
cracking. A variety of methods have been used for quantification,
including total crack length, maximum crack length, total number of
cracks, threshold strain for cracking, and others. Recently, Lin et al. [3]
developed a new methodology for evaluating cracking susceptibility using
the transverse Varestraint test that provides a measure of the temperature
range over which cracking occurs. This has been termed the solidification
cracking temperature range (SCTR) and is a subset of the BTR. Samples
are tested over a range of augmented strain and the maximum crack distance
(MCD) in the fusion zone is measured. A schematic of the transverse
Varestraint test is shown in Fig. 1.
Fig. 1. Schematic of the transverse Varestraint test for evaluating weld solidification
cracking susceptibility
Above a critical strain level, designated the saturated strain, the MCD
does not increase with increasing strain. This indicates that the solidification
crack has propagated the full length of the crack susceptible region.
Representative cracking as it appears at the trailing edge of the weld pool
is shown in Fig. 2 for a Type 310 sample tested at 5 % strain. By testing
over a range of augmented strain, an MCD versus strain plot such as that
shown in Fig. 3 can be generated.
274 Testing and Standardization
Fig. 2. Weld solidification cracking in a Type 310 stainless steel transverse Varestraint
specimen
Fig. 3. Maximum crack distance in the fusion zone versus applied strain during
transverse Varestraint testing
In this manner, the threshold strain for cracking to occur and the saturated
strain above which the MCD does not increase can be identified.
Typical strain ranges over which samples are tested are 0–7 %. Most fully
austenitic weld metals stainless steels and Ni-base alloys exhibit saturated
strain levels between 5 and 7 %. Threshold strain levels are generally in
the range from 0.5 to 2.0 %.
Recent Developments in Weldability Testing 275
Although, the threshold strain may, in fact, be an important criterion for
judging susceptibility to weld solidification cracking, the MCD at or above
saturated strain is much easier to determine and provides a measure of the
SCTR.
In order to determine SCTR, the cooling rate through the solidification
temperature range is determined by plunging a thermocouple into the weld
pool. The time over which cracking occurs is approximated by the MCD
above saturated strain divided by the solidification velocity. Using this approach,
SCTR can be calculated using the following relationship, where V
represents the welding velocity.
SCTR = [Cooling Rate] × [MCD/V]
The concept for determining SCTR using this approach is shown in
Fig. 4. By using a temperature rather than a crack length as a measure of
cracking susceptibility, the influence of welding variables (heat input,
travel speed, etc.) can be eliminated. SCTR then represents a metallurgically
significant, material-specific measure of weld solidification cracking
susceptibility.
Fig. 4. Method for determining the solidification cracking temperature range
(SCTR) using the cooling rate through the solidification temperature range and
MCD at saturated strain
276 Testing and Standardization
A bar chart showing the SCTR values for a number of austenitic and
duplex stainless steels is shown in Fig. 5. Alloys that solidify as primary
ferrite (duplex stainless steels 2205 and 2507, and Types 304 and 316L)
have low SCTR values, typically less than 50 °C. Alloys that solidify as
austenite exhibit SCTR values above 100 °C. Alloy A-286, which is notoriously
susceptible to solidification and liquation cracking, has a very high
SCTR value.
The SCTR data allows a straightforward comparison of cracking susceptibility.
These values may also allow alloy selection based on restraint
conditions. For example, in high restraint situations, SCTR values below
50 °C may be required to prevent cracking, while for low restraint weldments
150 °C may be sufficient.
Fig. 5. SCTR values for a number of austenitic and duplex stainless steels
Recently, Finton and Lippold [4] used a statistical approach to evaluate
the variables associated with transverse Varestraint testing. This study
used both austenitic stainless steels (Type 304 and 310) and Ni-base alloys
(Alloys 625 and 690) to determine the statistical importance of different
variables and to establish variable ranges in which testing should be conducted
to give reproducible results. Based on this study, they recommended
the variable ranges in Table 1 for use with stainless steel and Nibase
alloys.
Recent Developments in Weldability Testing 277
These ranges are not necessarily appropriate for other alloy systems.
More research is required to relate the SCTR to the local restraint conditions
required for cracking. As a minimum, the overall transverse Varestraint
approach described here appears to provide a good relative measure
of solidification cracking susceptibility and has been found to work well in
predicting the behavior in other systems, including structural steels and
aluminum alloys.
Table 1. Variables and variable ranges for transverse Varestraint testing of
stainless steels and Ni-base alloys
Arc Length Range: 0.05–0.15 in.
Maximum Voltage Changes: ± 1–1.5 volts
Minimum Specimen Length: 3.5 in.
Minimum Specimen Width
(parallel to welding direction): 3.0 in.
Current Range: 160–190 amps
Travel Speed Range: 4–6 in./min
Arc Length Range: 0.05–0.15 in.
Maximum Voltage Changes: ± 1–1.5 volts
The Cast Pin Tear Test
Although the Varestraint test has great utility for assessing the solidification
cracking susceptibility of most structural alloys, the test may be too
severe for evaluating some of the highly-alloyed Ni-and Co-base alloys
used for repair of turbine engine components, since these alloys may crack
at very low strain levels. For these materials, a modified version of the cast
pin tear (CPT) test, originally introduced by Hull [5], can be used. Using
this test, small charges of the material of interest are melted in a copper
crucible using a gas tungsten arc welding (GTAW) torch under argon
shielding. This charge is then dropped through the bottom of the crucible
into a copper mold. A range of mold diameters and lengths are used to
control the restraint in the solidifying pin. This procedure and apparatus
design is described in detail elsewhere [6].
Using the CPT test, a plot of percent cracking versus mold size is developed,
as shown in Fig. 6. 100 % cracking represents the situation where
cracking occurs completely around the diameter of the pin or there is complete
separation of the pin. Note that for Alloys 3 and 5 only small increases
in mold length result in large changes in cracking susceptibility,
while Alloy 1 and Alloy 625 are resistant to cracking until relatively long
mold lengths are used.
278 Testing and Standardization
Fig. 6. Cast pin tear test data for Ni- and Co-base alloys. Alloy 1 is Co-20Cr-
10Ni-W-4Al, Alloy 3 is Ni-10Co-8Cr-10W-5Al, and Alloy 5 is Ni-12Co-7Cr-5W-
6Al
Other distinct advantages of this test are that virtually no sample preparation
is required and very little material is used. An entire curve, such as
those shown in Fig. 6 can be generated with about 200 grams of material.
Additionally, testing is not time intensive. For a given material, testing and
analysis can be completed in just a few hours. The cooling rates achieved,
based on evaluation of solidification substructure size, are equivalent to
those in arc welds.
The test is under further development and refinement at Ohio State University.
An improved molten metal delivery system is being tested that will
greatly facilitate mold filling and allow a wider range of mold geometries
to be used. This should allow alloys with only moderate solidification
cracking susceptibility to be tested.
Heat-affected Zone Liquation Cracking
Heat-affected zone liquation cracking is associated with the partiallymelted
zone (PMZ) that forms in the HAZ just adjacent to the fusion
boundary. This form of cracking occurs along grain boundaries that liquate
at elevated temperature due to a variety of mechanisms.
Recent Developments in Weldability Testing 279
In many alloys, this liquation occurs simply due to the segregation of
impurity and/or other melting point depressant elements to the grain
boundary during the HAZ thermal cycle. If the concentration of these elements
and the time above the boundary melting temperature are sufficient,
local boundary melting will occur.
Another mechanism for PMZ formation is known as constitutional liquation
[7]. Constitutional liquation occurs when a “constituent” particle,
such as a carbide, reacts with the surrounding matrix resulting in local
melting at the particle/matrix interface. The particle itself does not melt,
rather the reaction zone at the interface undergoes eutectic melting. If a
grain boundary then intersects this liquid region, the boundary can be wet
by the liquid and becomes susceptible to liquation cracking. These mechanisms
have been described in some detail elsewhere [7]. Susceptibility to
HAZ liquation cracking can be quantified using both the Varestraint test
and the hot ductility test, as described by Lin et al. [8].
The Spot Varestraint Test
The Varestraint test technique for HAZ liquation cracking differs from that
described previously for weld solidification cracking, since it uses a stationary
spot weld to generate a stable HAZ thermal gradient and microstructure.
Fig. 7. Schematic illustration of the spot Varestraint test [5]
280 Testing and Standardization
The technique developed by Lin [9] used both an “on-heating” and “oncooling”
approach to quantify HAZ liquation cracking. A schematic of the
test is shown in Fig. 7.
The on-heating test is conducted by initiating the GTA spot weld, ramping
to the desired current level, and then maintaining this current level until
the weld pool size stabilizes and the desired temperature gradient in the
HAZ is achieved. For the Type 310 and A-286 stainless steels evaluated by
Lin et al. a spot weld surface diameter of approximately 12 mm was
achieved after a weld time of 35 seconds. The arc is then extinguished and
the load immediately applied to force the sample to conform to the die
block. By using no delay time between extinction of the arc and application
of load, HAZ liquation cracks initiate at the fusion boundary and
propagate back into the HAZ along liquated grain boundaries (Fig. 8).
Fig. 8. Spot Varestraint test sample (plan view) of A-286 stainless steel tested at
5 %strain
By plotting maximum crack length (MCL) versus strain, a “saturated
strain” can be determined which defines the strain above which the maximum
crack length does not change. Examples of on-heating MCL versus
strain plots for Type 310 and A-286 stainless steels are shown in Fig. 9A.
Note that a threshold strain for cracking cannot be identified for A-286 and
that both materials achieve saturated strain at 3 %.
For the on-cooling test, the same procedure as described above is used,
but after the arc is extinguished there is a delay before the sample is bent.
Recent Developments in Weldability Testing 281
Fig. 9. On-heating (A) and on-cooling (B) HAZ liquation cracking
as measured by the spot Varestraint test [5]
By controlling the delay, or cooling, time, the weld is allowed to solidify
and the temperature in the PMZ drops until eventually the liquid films
along grain boundaries are completely solidified. By plotting MCL versus
cooling time, the time required for liquid films in the PMZ to solidify can
be determined. This is shown in Fig. 9B.
282 Testing and Standardization
Note that over 4 seconds is required before cracking disappears in A-
286, indicating that the grain boundary liquid films persist to quite low
temperatures. By measuring the temperature gradient in the HAZ using
implanted thermocouples, it is possible to determine the thermal crack susceptible
region (CSR) surrounding the weld within which HAZ liquation
cracking is possible. As described by Lin [9] this is done by converting the
MCL at saturated strain for both on-heating and on-cooling spot Varestraint
tests to a temperature by multiplying the MCL by the temperature
gradient in the HAZ. Using this approach, it is possible to describe the region
around a moving weld pool that is susceptible to HAZ liquation
cracking.
Plots of the thermal CSR for A-286 and Type 310 are shown in Fig. 10.
Note that the width of the CSR at the periphery of the weld is 222 °C for
A-286 and only 61 ºC for Type 310. It is also interesting that, based on the
on-cooling data, the PMZ in A-286 does not fully solidify until the temperature
reaches 1035 °C while for Type 310 it is 1296 °C. This technique
provides a quantifiable method to determine the precise temperature ranges
within which cracking occurs and to allows differences in HAZ liquation
cracking susceptibility among materials to be readily measured.
Fig. 10. The thermal crack susceptible region (CSR) as determined
from the spot Varestraint test for (A) A-286, and (B) Type 310 [5]
Recent Developments in Weldability Testing 283
Ductility-Dip Cracking
Ductility-dip cracking (DDC) refers to elevated temperature, solid-state
cracking that results from a sharp drop in ductility at temperatures above
approximately half the melting temperature of the material [10, 11]. It can
occur in wrought alloys, castings, and in the HAZ and fusion zone of
highly restrained weldments. Characteristically, it is associated with single
phase austenitic alloys with large grain size, and is intergranular in nature.
Considerable work has been conducted over the last few years to understand
the nature of DDC in welded austenitic stainless steels and Ni-base
alloys [12, 13, 14, 15]. This work has included investigation of austenitic
stainless steels (Types 304 and 310, and AL6XN), Ni-base alloy 690, and
Ni-base filler metals 82 and 52.
A typical ductility-dip crack in a Ni-base weld metal is shown in
Fig. 11.
Fig. 11. Ductility-dip cracking along migrated grain boundaries in fully austenitic
weld metal
In weld metals, DDC always occurs along migrated grain boundaries
(MGBs). These are crystallographic, high-angle boundaries that have migrated
away from their parent solidification grain boundaries during cooling
below the solidification temperature range and/or during reheating in
multipass welds. A detailed description of these boundaries can be found
elsewhere [16].
Weld metal DDC in stainless steels and Ni-base alloys has been found
to be a strong function of grain size, grain boundary character, and preci284
Testing and Standardization
pitation behavior. Weld metals exhibiting large grains with straight MGBs
and few grain boundary precipitates tend to be the most susceptible. An increase
in grain boundary tortuosity resulting from local pinning by precipitates
or second phases that form at elevated temperature will decrease susceptibility
to DDC.
This occurs by a grain boundary locking effect that resists grain boundary
sliding. While not directly linked to DDC, impurity segregation to the
MGBs tends to further increase susceptibility to this form of cracking. In
Ni-base filler metals, the addition of hydrogen to the shielding gas increases
susceptibility to DDC [14].
In order to quantify susceptibility to DDC, the strain-to-fracture (STF)
test was recently developed by Nissley at Ohio State University [12]. The
STF test employs a “dogbone” tensile sample with a GTA spot weld applied
in the center of the gage section. The spot weld is made under controlled
solidification conditions using current downslope control.
This results in an essentially radial array of migrated grain boundaries
within the spot weld. Samples are then tested in a Gleeble™ thermomechanical
simulator at different temperatures and strains. Temperature
and strain ranges are typically 650–1200 °C and 0–20 %, respectively. After
testing at a specific temperature-strain combination, the sample is examined
under a binocular microscope at 50X to determine if cracking has
occurred. The number of cracks present on the surface is counted.
Fig. 12. Strain-to-fracture test results for three austenitic stainless steels [9]
Recent Developments in Weldability Testing 285
Using this data, a temperature vs. strain envelope is developed that defines
the regime within which DDC may occur. Both a threshold strain for
cracking (εmin) and ductility-dip temperature range (DTR) can be extracted
from these curves. Temperature-strain curves are shown in Fig. 12 for
Type 310, Type 304, and the super-austenitic alloy AL6XN. Based on
these curves, Type 310 would be expected to have the highest susceptibility
to DDC since the DTR at 15 % strain is 400 °C and εmin is approximately
5 %.
This test has been shown to be remarkably sensitive to the onset of grain
boundary cracking in the DDC range and should prove to be a valuable
tool for studying elevated temperature embrittlement in the weld metal and
HAZ. Work is ongoing at Ohio State University to further optimize the
test. This test is also being using the study the composition and metallurgical
variables that affect susceptibility to ductility-dip cracking.
Strain-age Cracking
Strain-age cracking (SAC) is a form of postweld heat treatment cracking
that is associated with Ni-base superalloys. The term “strain-age” is derived
from the fact that cracking occurs in the temperature range were extrinsic
and intrinsic strain accumulation overlaps the onset of precipitation
hardening, or aging. In most Ni-base superalloys, this form of cracking is
closely related to the precipitation of gamma-prime, Ni3(Ti,Al). Alloys
with higher Ti + Al contents tend to be more susceptible to SAC. [17].
Grain size and impurity content also influence susceptibility. Materials
with much finer HAZ grain size or those with low levels of sulfur, phosphorus,
and boron are more resistant to SAC. A relationship between SAC
susceptibility and grain boundary liquation has also been reported [18].
SAC is usually associated with the HAZ of either wrought alloys or castings
and occurs along grain boundaries in close proximity to the fusion
boundary. A strain-age crack in Waspaloy and the corresponding fracture
surface are shown in Fig. 13. From a mechanistic standpoint, cracking occurs
along the grain boundary due to intragranular strengthening by precipitation
and the corresponding formation of a precipitate-free zone at or
near the grain boundary. Upon the application of sufficient strain, cracking
occurs through this weakened region. There is an ongoing debate about the
validity of this mechanism and further research is required to resolve this
issue.
As with other forms of cracking, numerous tests have been developed to
determine susceptibility to SAC. To date however, there is no standardized
test for quantifying the susceptibility of an alloy to SAC.
286 Testing and Standardization
Fig. 13. Strain-age crack in Waspaloy, (A) optical micrograph,
(B) SEM fractograph
Many of the test techniques that have been developed use a Gleeble™
thermo-mechanical simulator. The problem with these tests is that they do
not accurately simulate the thermo-mechanical history of a weld. Most
tests do not adequately simulate the development of residual stresses in a
weldment as it cools after the weld metal is deposited. Those tests that impose
stresses on cooling from the peak temperature to simulate weld residual
stress do not allow relaxation of the stresses in subsequent PWHT
simulations.
The approach used by Norton [19] at Ohio State University attempted to
more closely simulate the actual conditions experienced by the HAZ during
welding. The Gleeble was used to impose a simulated thermal cycle on
Recent Developments in Weldability Testing 287
the specimen. After reaching the peak temperature, the sample was restrained
in the Gleeble jaws and allowed to cool to room temperature. This
resulted in the buildup of considerable stress in the sample due to thermal
contraction. The sample was then heated to an appropriate PWHT temperature
and held at temperature for a predetermined time (0 to 4 hours).
Upon reheating to the PWHT temperature, considerable stress relaxation
occurs and then stress begins to build in the sample as aging occurs, as
shown in Fig. 14. The starting stress for both alloys has been subtracted so
that that the stress buildup relative to each other can be shown. Note that
the increase in stress is more rapid in Waspaloy due to the more rapid aging
response associate with gamma-prime precipitates relative to gamma
double prime in Alloy 718. After the prescribed hold time the sample is
pulled to failure at the PWHT temperature and the ductility measured. The
hot ductility following PWHT was used to develop a multivariate polynomial
for calculating the ductility as a function of PWHT temperature and
time.
Fig. 14. Stress increase during hold at constant displacement during post-weld
heat treatment cracking test. The initial stress for both alloys has been normalized
to zero to allow for a better comparison of their hardening response
The collected data appeared to have a parabolic curve, so the model was
chosen to be a second order polynomial. A spreadsheet for both alloys was
created with factors of time, temperature, and the interactions between the
two in a second order equation. A line was fit to the ductility (reduction in
area) measurements. The resulting output gave the intercept and coefficient
288 Testing and Standardization
for each of the five variables as well as the coefficient of determination for
the fit.
An example of ductility versus temperature curves for Waspaloy and
Alloy 718 determined using this method is shown in Fig. 15. The coefficients
of determination (R2) for the Waspaloy and Alloy 718 surface plot
polynomials are 0.92 and 0.91, respectively. The regression models show
good fit to the measured data over the range of tested times and temperatures
both by their high coefficients of determination and the ability to
predict the ductility of samples. The curves in Fig. 7 are consistent with actual
experience in welding these alloys. Alloy 718 is generally quite resistant
to SAC, while Waspaloy is considered moderately susceptible.
Additional details of this test and its potential importance for determining
susceptibility to PWHT cracking can be found elsewhere [19]. It
should be noted that this test is not limited to the evaluation of Ni-base superalloys,
but can also be applied to other materials that are susceptible to
PWHT cracking, such as Cr-Mo-V steels and stainless steels (Type 347).
Fig. 15. Comparison of elevated temperature ductility of Waspaloy and Alloy 718
regression models for 3 hours of PWHT [19]
Summary
The weldability of engineering materials continues to be a topic of considerable
interest and relevance in the manufacturing community. Despite the
many studies that have been conducted and the hundreds of weldability
tests that have been proposed, the quest to quantify weldability with the
goal of avoiding (or at least predicting) susceptibility to weld cracking remains
elusive. This elusiveness arises from the fact that most of the
Recent Developments in Weldability Testing 289
cracking mechanisms are still not well understood and there are relatively
few standardized tests to measure weldability.
This paper has described tests for weld solidification cracking, HAZ liquation
cracking, ductility-dip cracking, and postweld heat treatment
cracking that have carefully considered the metallurgical variables that influence
cracking in the development of the test techniques. These tests are
useful not only for quantifying susceptibility, but for studying the fundamental
mechanisms that promote weld cracking in engineering materials.
While none of these tests are yet “officially” standardized, they each provide
sufficient detail regarding sample preparation, testing procedure, and
analysis that laboratory-to-laboratory variability in performing the tests
should be minimized.
The goal to develop standardized weldability tests has been an elusive
one and will probably not be realized until international cooperation and
collaboration is achieved. It is hoped that this conference will be a first
step in realizing that cooperation.
Acknowledgements
This paper represents the work of many students and colleagues over the
last 15 years. I would particularly like to thank Dr. Wangen Lin for his insight
and diligence in the development of the transverse and spot Varestraint
test techniques, Mr. Daniel Ryan for developing the modified cast
pin tear test, Mr. Nathan Nissley and Mr. Matthew Collins for the development
and testing associated with the strain-to-fracture test, and Mr. Seth
Norton for the evolution of the strain-age cracking test. It is my privilege
to have had these individuals work in my research laboratory and I wish to
acknowledge their “blood, sweat, and tears” that made these advancements
possible. I would also like to recognize my research sponsors, including
Edison Welding Institute, the American Welding Society, and BWXT, Inc.
for providing the resources to support students and conduct the research
presented here.
References
1. Prokhorov NN (1962) The technological strength of metals while crystallizing
during welding. Welding Production 9 (4): 1–8
2. Matsuda F, Nakagawa H, Tomita S (1986) Quantitative evaluation of solidification
brittleness of weld metal during solidification by in-situ observation
and measurement (Rpt. 3). Transactions JWRI 15 (2)
290 Testing and Standardization
3. Lippold JC, Lin W (1996) Weldability of commercial Al-Cu-Li alloys. In:
Proc. of ICAA5, Aluminum Alloys – Their Physical and Mechanical Properties,
eds.. Driver JH et al., Transtec Publications, pp 1685–1690
4. Finton T, Lippold JC (2004) Standardization of the Transvarestraint test, EWI
Summary Report No 04–XX
5. Hull FC (1959) Cast-pin tear test for susceptibility to hot cracking. Welding
Journal 38 (4): 176s–181s
6. Ryan D (2003) Development of a Modified Cast Pin Tear Test for Ni-base Alloys.
Master’s Thesis, The Ohio State University
7. Pepe JJ, and Savage WF (1967) Effects of Constitutional Liquation in 18 Ni
Maraging Steel Weldments. Welding Journal, Vol. 46: 411s–422s
8. Lippold JC, Varol I, Baeslack WA (1992) An investigation of heat-affected
zone liquation cracking in austenitic and duplex stainless steels. Welding
Journal 71(1): 1s–14s
9. Lin W, Lippold JC, Baeslack WA (1993) An investigation of heat-affected
zone liquation cracking, part 1 – a methodology for quantification. Welding
Journal 71(4): 135s–153s
10. Hemsworth B, Boniszewski T, Eaton NF (1969) Classification and Definition
of High Temperature Welding Cracks in Alloys. Metal Construction & British
Welding Journal, February: 5–16
11. Rhines FN, Wray PJ (1961) Investigation of the Intermediate Temperature
Ductility Minimum in Metals. Transactions of the ASM, Volume 54: 117–128
12. Nissley NE, Lippold JC (2003) Ductility-dip cracking susceptibility of austenitic
alloys. 6th International Trends in Welding Research, ASM International,
pp 64–69
13. Nissley NE, Guaytima G, Lippold JC (2002) Development of the strain-tofracture
test for evaluating ductility-dip cracking in austenitic alloys. IIW Doc
#IX-2037-02
14. Collins MG, Lippold JC, Kikel JM (2003) Quantifying ductility-dip cracking
susceptibility in nickel-base weld metals using the strain-to-fracture test. 6th
International Trends in Welding Research, ASM International, pp 586–590
15. Kikel JM, Parker DM (1999) Ductility dip cracking susceptibility of Filler
Metal 52 and Alloy 690. 5th International Trends in Welding Research, ASM
International, pp 757–762
16. Lippold JC, Clark WAT, Tumuluru M (1992) An investigation of weld metal
interfaces. In: The Metal Science of Joining. The Metals, Minerals and Materials
Society, Warrendale, PA, pp 141–146
17. Prager M, Shira S (1968) Welding of precipitation-hardening nickel-base alloys.
WRC Bulletin No 128, February
18. Duval DS, Owczarski WA (1969) Studies of postweld heat treatment cracking
in nickel-base alloys. Welding Journal 48 (1): 10s–22s
19. Norton SJ, Lippold JC (2003) Development of a Gleeble-based test for postweld
heat treatment cracking susceptibility. 6th International Trends in Welding
Research, ASM International, pp 609–614
Hot Cracking Tests –
The Route to International Standardization
J.C.M. Farrar
Introduction
Hot cracking in weld metals and parent material heat affected zones (HAZ)
has been a problem for as long as steels and alloys have been fusion
welded. In most codes and standards, detectable cracks of any type are not
permitted and so efforts have been directed to the development of weld
metals and parent alloys which are not sensitive to hot cracking and to the
design of welding procedures which reduce or eliminate the risks of hot
cracking.
A natural consequence of this has been the need for tests, which can be
used to assess the sensitivity of weld metals and parent alloys, compare the
sensitivity of different compositions, and predict possible occurrence in
real fabrications.
Many tens of different tests have been developed over the years, some
of which are specific to individual groups of materials, some to specific
welding processes and some to individual pieces of test equipment at one
or more laboratories. The complexities of the actual cracking mechanisms
combined with the diversity of individual tests means that a form of standardization
is difficult, and true international standardization is particularly
challenging.
Nevertheless, considerable progress has been made in the last 20 years
and this paper attempts to identify those issues, which need to be resolved,
if an international standard is to be developed in the coming years. It concentrates
on two specific standards. The first of these is the Russian Standard
namely GOST 26389 – 84 [1], which was developed twenty years ago
and has been widely used by countries of the former Soviet Union and by a
number of countries in Eastern Europe which have strong technical links.
The second is the draft International and European Union standard
prEN/ISO 17641 [2], which has been developed within CEN TC121
Working Group 13, with considerable support and assistance from IIW sub
commissions II-C and IX-H. A major review of the scientific basis of possible
harmonization of Russian Federation and European Union Weldability
Tests has been carried out at the Bauman University Moscow [3].
292 Testing and Standardization
This paper also recognizes the major project carried out at BAM, Berlin
and reported within Commissions II and IX of IIW. The final report, presented
in 1999, not only described the results of a large worldwide ‘Round
Robin’ test programme carried out by members of IIW but also highlighted
the very real problems of test consistency and reproducibility between different
laboratories [4].
Fundamentals of Hot Cracking and Definitions
Hot cracks arise in weld metals and in the HAZ of parent materials when
the strains developed during cooling of welded joints, or imposed externally,
exceed the ductility of a particular part of the joint. They range in
size from very small liquation cracks (< 1 mm in length) in HAZ’s or multipass
welds up to large solidification cracks, which may extend along the
complete length of a welded joint.
Hot cracks are not confined to any particular alloy system and can occur
in steels, stainless steels, nickel-base alloys, copper and aluminium based
alloys. The detailed reasons for the incidence of hot cracks are many and
complex and fall outside the scope of this paper. In general terms, they occur
when localized ductility is insufficient to support any imposed strains.
The lack of ductility can result from the presence of liquid metal, micro
structural features, orientation (relative to the strain) and in some cases
upon the presence of brittle impurities and low melting point (or liquated)
films. In this respect some alloy systems are particularly sensitive to the
presence of impurity elements such as sulphur, phosphorus and lead etc. It
is widely recognized that fully austenitic single-phase microstructures, particularly
weld metals, are susceptible to hot cracking of one form or another.
Impurity levels strongly influence the incidence of cracking in such
microstructures.
Simple definitions, taken from prEN ISO 17641-1 are given below:
− Hot Cracks are material separations occurring at high temperatures
along the grain boundaries (dendrite boundaries), when the level of
strain and the strain rate exceed a certain level. Small cracks, which are
visible only at magnifications of more than about 50x, are often described
as micro fissures.
The general term is further sub-divided into three specific types:
− Solidification Cracks are formed during solidification from the liquid
phase of weld metals. They can sometimes be visualized as dendrites,
Hot Cracking Tests – The Route to International Standardization 293
which are growing more slowly than the strains being imposed, so that
they fail to fuse on the weld center line and leave a cavity or centerline
crack. They usually extend up to the surface of the weld metal, but
sometimes remain subsurface.
− Liquation Cracks are formed in areas of liquation in the HAZ of a parent
material or in multipass welds where weld metal is reheated during
the deposition of subsequent weld beads. They are often small (micro
fissures) and often subsurface in multi pass welds.
− Ductility Dip Cracks are formed in regions of reduced hot ductility in either
parent HAZ or in multipass welds. They are associated with specific
temperature ranges and may result from diffusion related compositional
changes.
Types of Test
From the above brief commentary and definitions it can be seen that the
successful design of a single all-embracing test is difficult.
A universal test would have to take in account all of the following possible
variables:
1. Many different alloy types, steels, aluminium alloys etc.;
2. The three different cracking types, as described above;
3. Weld metal cracking in single and multipass welds;
4. Parent material HAZ cracking in single and multipass welds;
5. Visible and subsurface cracks;
6. Effects of both variable strain and variable strain rate.
It is therefore perfectly reasonable that a whole variety of tests should
have evolved over the last fifty years or so. Some tests have been developed
to assess a specific type of cracking in a specific material type. Some
have been concerned with the relative susceptibility of similar alloys
whereas others have been more concerned with the prediction of whether
or not cracking would occur in a real situation.
There are tests, which are primarily used for the optimization of alloy
compositions, and others are used to develop and optimize weld procedures.
Some tests are highly quantitative and are designed to give a
numerical indication of susceptibility, others tend be go/no-go or crack/nocrack
types of test. However it is possible to broadly classify the various
tests into two major groups, with a simple subdivision of one of the
groups.
294 Testing and Standardization
Self Restrained Tests
These are tests where the specimen loading is produced by stresses developed
during welding of a restrained weldment. There are no externally applied
loads and in the Russian standard these are described as ‘technological’
tests.
A number of specimen designs have been developed with varying degrees
of restraint and a range of weld geometries, but there appears to be
general agreement on the use of restrained fillet weld tests. The Russian
standard also includes a circular patch type test with various design options
to vary the level of restraint.
A major difference between the two standards is the EN/ISO use of
weld procedure butt welds where longitudinal tensile or bend test specimens
are extracted and subjected to additional external straining. It must
be emphasized that this additional ambient temperature straining does not
induce any new hot cracking but only serves to help detect cracks already
existing in the weld. In this respect the tests are probably most useful in locating
small cracks and micro fissures in weld metal.
Externally Loaded Tests
These are tests that require the use of equipment, which is capable of imposing
a strain on the test specimen whilst welding or some form of thermal
cycle is in progress. In the Russian standard these are described as
‘machine’ tests. Attempts have been made to standardize both test equipment
and test methods and two Russian standards, namely GOST 15533-
80 and GOST 7855-84 are available.
However many different laboratories world wide, have either built their
own testing equipment or have developed specimen designs and procedures
to suit existing test equipment. This was the main reason for the lack
of between-laboratory reproducibility in the hot cracking Round Robin.
It is also the reason that part 3 of the EN/ISO standard is to be published
first as technical report and not as a full standard. This document is intended
to provide information and guidance on the use of externally loaded
tests in the hope that future developments can lead to reproducibility of
test methods and the procedures considered essential for an international
standard.
Hot Cracking Tests – The Route to International Standardization 295
Description of Tests
Restrained Weldments
Whatever the detailed design, (restrained fillet, patch test etc) these tests
are intended to assess the solidification cracking susceptibility of weld
metal in a single pass restrained fillet weld. Various types of test are available,
with increasing restraint being provided by increasing thickness
and/or stiffness of the plates used in the test assembly. The tests were
originally designed for use with arc welding processes and filler metals, although
they might prove useful with some of the newer processes (such as
hybrid laser/MAG welding). A typical test assembly is shown in Fig. 1.
Fig. 1. One example of a self restraint T-joint. (From [2], Part 2.)
Assessment is based upon the length and position of cracks (if any) in
the test weld. Assessment is essentially qualitative since no direct measurement
of strain is available and it is not possible to directly link the incidence
of cracking in the test with the risk of cracking in a real full size fabrication.
However the test can be useful in ranking weld metals and in
investigating variables such as heat input or welding speed.
296 Testing and Standardization
Weld Metal Tensile Test
This test is designed to assess the susceptibility of weld metals to liquation
and ductility dip cracking. Applying a load to rupture a cylindrical all-weld
metal specimen taken from a butt weld opens cracks initiated during the
welding operation. The surface regions adjacent to the fracture surface are
examined and any cracking can be detected and measured, as shown in
Fig. 2. This test can be combined with normal butt weld procedure tests, to
qualify a particular consumable and/or procedure. Alternatively it can be
used to investigate the effect of welding process variables, such as heat input,
and even the influence of weld restraint if thicker test plates are used
and welding is carried out in a restraining jig.
Fig. 2. Fractured all-weld metal tensile specimen, showing region examined for
hot cracking. (From[2], Part 2)
Fig. 3. Arrangement and dimensions of the Longitudinal bend Test (LBT) prior to
bending, showing the area to be examined - 4 (From [2], Part 2)
Hot Cracking Tests – The Route to International Standardization 297
Longitudinal Bend Test
This test is designed to assess the sensitivity to solidification, liquation and
ductility dip cracking in all-weld metal butt welds. Bending of a test
specimen taken longitudinally from a butt weld opens up any cracks initiated
during welding as shown in Fig. 3.
Possible applications and variables are similar to the weld metal tensile
test. In principle the positioning of the specimen can be such as to include
parent metal HAZ, so that liquation cracking in this region can also be investigated.
Externally Loaded Tests – Background
These tests are designed to produce quantitative rather than qualitative results
and to closely simulate, under controlled conditions, the important
variables, which relate either directly or indirectly to hot cracking in real
welds.
It is generally recognized [5] that hot cracking is a result of the exhaustion
of plasticity/ductility at high temperatures and is controlled by the simultaneous
influence of different key factors for example the size of the
brittle temperature range (BTR), the ductility of the material within the
BTR and the magnitude and rate of straining arising from weld metal
shrinkage and overall structural restraint.
Many of the externally loaded tests attempt to combine the effects of the
main parameters by:
1. Starting straining at the liquid state and finishing at full solidification,
thus covering the complete BTR.
2. The strain rate in the BTR is similar to that encountered in real welds,
thus allowing diffusion, stress relaxation during solidification and self
healing of cracks by liquid metal, to take place during the course of the
test.
3. Reproducing both local shrinkage strains and external strains up to the
point of crack formation.
The four tests described in the draft EN/ISO technical report prCEN TR
17641-3 together with some of the tests from the Russian standard will be
briefly reviewed in relation to their ability to meet the above criteria.
298 Testing and Standardization
Hot Tensile Test
Hot cracking susceptibility is determined by carrying out a tensile test
whilst at the same time heating the specimen to simulate a welding thermal
cycle. The specimen can be extracted from parent material or from allweld
metal. In principle, imposing an appropriate prior thermal cycle on
the parent material can be used to test HAZ material. These tests are usually
carried out in specialized equipment such a Gleeble® machine which
enables precise control of thermal cycles and imposed strains. Different
procedures and specimen dimensions are used to assess susceptibility to
solidification cracking from those used to assess susceptibility to liquation
cracking:
1. To simulate solidification cracking the specimen is heated to the melting
temperature and the equipment jaws holding the specimen are then held
fixed so that the shrinkage strains are able to induce cracking.
2. To simulate liquation cracking, specimens are heated to a range of temperatures
just below the solidus. A number of specimens are strained to
failure and susceptibility is based on the production of a hot ductility
curve.
Assessment is based on establishing the BTR as a fundamental characteristic
of the material. In this respect, the test is useful for material selection,
ranking and approval but as has been explained above the test does
not necessarily cover all the key variables, which affect hot cracking, particularly
procedural variables in weld metals and welded joints.
Weldments with Externally Imposed Strain
The basis of these tests is to impose a bending or tensile strain on a weld,
whilst the weld is being deposited. Bending offers the potential advantage
of inducing hot cracking in the surface regions of the test specimen which
enables a more reliable assessment to be carried out. The weld may be an
autogenous TIG melt run on parent material or it may be an arc welding
process with filler metal. It is usual to employ a specially designed test
machine and different design philosophies have been adopted in different
countries. In Russia the emphasis has been on progressively increasing external
strain, using arrangements similar to those shown in Fig. 4 and included
in the GOST standard.
In the USA and parts of Europe machines based on the Varestraint or
Transvarestraint Test have been developed, but not fully standardized
(Fig. 5).
Hot Cracking Tests – The Route to International Standardization 299
Fig. 4. Diagrams taken from the GOST standard [1] showing general arrangements
for tensile and bend straining of externally loaded specimens
In these tests, the load to induce the strain is imposed at a predetermined
point in the production of the weld. Yakushin [3] has expressed severe reservations
about the use of so called ‘shock’ loading in some of these tests
and claims that it does not reproduce the conditions experienced by real
welds when strains develop progressively throughout the complete BTR
during solidification and cooling.
Assessment of such tests is usually based on some measurement of
cracks once the test is complete. The current consensus appears to favor
using the total visible crack length, although some cracks may be subsurface
particularly in the root region. It still needs to be confirmed that total
crack length is a true and rapid measure of the hot cracking sensitivity.
These tests are widely used for both parent material and weld metal selection
and approval. They are also useful for examining the effect of
welding procedural variables. They are certainly used successfully within a
given laboratory where sound protocols have been established and where a
database correlating test results with service experience exists.
300 Testing and Standardization
Fig. 5. General arrangement of the Varestraint and Transvarestraint Tests [4]
However, inter-laboratory trials have shown relatively poor reproducibility
and this has been attributed largely to differences in equipment and
test procedures [4].
The Flat Tensile Test
This test, sometimes known as the Controlled Deformation Cracking Test
or PVR Test, was developed to give a relatively simple test, using more
widely available and less specialized equipment. It is capable of quantifying
the hot cracking susceptibility of base materials, weld metals and a
range of conventional arc welding processes.
Hot Cracking Tests – The Route to International Standardization 301
Fig. 6. General arrangement of the Flat Tensile Test or PVR-Test showing position
of first hot cracking [6]
This test is carried out using a single flat tensile specimen, which is
strained in a horizontal tensile testing machine capable of being programmed
with a linearly increasing strain rate. The test differs fundamentally
from the Varestraint or Transvarestraint tests in that a linearly increasing
tension speed, rather than instantaneous loading imposes the
strain. The test is normally carried out with the strain being imposed in the
same direction as the weld run. However the test can be carried out with
the weld being made transverse to the applied strain. The test can be performed
both with and without filler materials, and in the latter case, base
metals can be assessed using a TIG melt run and standardized welding parameters.
Filler materials can be assessed using welding conditions recommended
by the consumable supplier, or actual welding conditions can be applied to
reproduce a practical situation. Welding procedures can be optimized by
variation of welding conditions e.g. electrical parameters, travel speed,
shielding gas mixtures, flux-wire combinations etc. in order to minimize
the risk of hot cracking.
All three types of hot cracking (solidification, liquation, and ductility
dip) can be reproduced in a single test, and in principal, only a single test is
302 Testing and Standardization
needed to characterize a material, provided that some reference data are
available for comparison.
It is claimed that the test is reproducible, with low scatter and is capable
of good discrimination between the three main types of hot cracking [6].
The test procedure in its standard form is shown in Fig. 6. Specimen
dimensions can be modified to suit the capacity of available testing equipment.
Welding is carried using a constant welding speed and the tension
speed is programmed to increase from zero at the start of the test to
70 mm/min at the end of the test.
On completion of the test, the test specimen is examined at about 25x
magnification to identify the first hot cracks. The tension speed corresponding
to the appearance of the first crack (more than one type may be
present) is defined as the critical tension speed vcr, and is a measure of hot
cracking susceptibility. Streitenberger [6] has carried out a comprehensive
comparison of two externally loaded tests namely the Modified Varestraint
Test (MVT) and the PVR test.
Discussion
The main advantage of self-restrained tests is that they are relatively simple
to carry out and require no specialized testing equipment. The major
disadvantage is that the results are qualitative, or possibly semi quantitative
in the case of the longitudinal bend test.
To assess the effect of welding variables a large number of tests are required
and these can be both time-consuming and costly. However, they
may have a place in a workshop environment where some reassurance of
avoidance of hot cracking is required. The tests are not really suitable for
the investigation of critical variables or material assessments in the laboratory.
The main general disadvantage of the externally loaded tests is the requirement
for specialized equipment and procedures. Previous work has
shown that reproducibility between laboratories is poor and that it is
probably unreasonable to expect good agreement between a number of different
types of test.
If progress is to be made towards an a truly international standard for
externally loaded hot cracking tests, then one test should be chosen for further
development, establishment of procedures and widespread validation
and verification.
Such a test should meet the following requirements:
Hot Cracking Tests – The Route to International Standardization 303
1. Be geometrically simple with controlled straining. Relatively complex
bending and ‘shock’ loading should be avoided;
2. Utilize readily available, relatively inexpensive test equipment;
3. Be capable of reproducing all forms of hot cracking;
4. Be suitable for testing parent materials and weld metals;
5. Be suitable for investigating welding procedural variables;
6. Be quantitative;
7. Be reproducible both within and between laboratories.
Of the tests described in both the prEN/ISO and GOST standards , the
one, which most clearly meets many of the above criteria, is the Flat Tensile
Test or PVR Test although it should be noted that the actual strain in
the molten is not measured. In addition, this test, because of its relative
geometrical simplicity is the one most amenable to mathematical modeling.
However the thermo-mechanical interactions are highly complex and
much further work is required. Modeling could prove useful in further development
of the test and in future validation [7]. It is therefore suggested
that consideration should be given to carrying out the work necessary for
this test to be developed as an international standard test.
It is believed that the following will be required:
i. A very clear and precise procedure for performing the test should be
developed;
ii. If possible, the dimensions of the test piece should be standardized and
fixed;
iii. The minimum specification for the testing machine, consistent with ii)
should be defined;
iv. Evidence of within laboratory reproducibility should be published;
v. Evidence of between laboratory reproducibility should be published
and if this is not available, then consideration should be given to carrying
out an appropriate Round Robin exercise once items i) to iii) have
been completed.
Conclusions
A comprehensive and critical review of the European International
prEN/ISO and Russian GOST standards on hot cracking tests has been carried
out. It is concluded that:
1. There is good agreement between the two standards on self-restrained
tests.
304 Testing and Standardization
2. It should be relatively easy to produce a common test (or tests), but it is
recognized that such tests have limited application and are at best semiquantitative
in their output.
3. There is a need for a relatively simple, quantitative, externally loaded
and programmable ‘machine’ test.
4. The test best suited, at the present time, to meet the needs of an International
standard is the Flat Tensile or PVR Test.
5. More work is needed on testing procedures and demonstrations of reproducibility
before the test can be standardized.
References
1. Nikolayev G, Yakushin B, Saharov Y, Misurov A, Deev A. Welded Joints –
Test methods on resistance to shrinkage crack formation under fusion welding.
State Standard of Soviet Union, GOST 26389-84
2. Destructive tests on welds in metallic materials – Hot cracking tests for
weldments – Arc welding processes. Draft European International Standard.
PrEN ISO 17641
Part 1: General
Part 2: Self restraint tests
Part 3: Externally loaded tests (technical report)
3. Yakushin BF (2003) The Scientific Base for Russian Federation and European
Weldability Tests Standards Harmonisation. IIW doc IX-H-573-03
4. Wilken K (1999) Investigation to Compare Hot Cracking Tests – Externally
Loaded Specimens. IIW doc IX-1945-99
5. Yakushin BF (2003) Analysis of GOST Standard 26389-84.
IIW doc IX-H-582-03 (translated by Peter Bernasovsky)
6. Streitenberger M (2002) Comparison of hot cracking test methods: PVR-test
and MVT- test. IIW doc IX-H-527-02
7. Herold H, Streitenberger M, Pchennikov A (2000) Modelling of the PVR-test
to Examine the Origin of Different Hot Cracking Types.
IIW doc IX-H-474-00
Value of Different Hot Cracking Tests for the
Manufacturer of Filler Metals
H. Heuser
Böhler Thyssen Schweisstechnik Deutschland GmbH, Germany
Introduction
For material evaluation and quality assurance an efficient, flexible and profitable
test concept for estimation of welding qualification with regard to
hot cracking susceptibility is needed. It has to provide the possibility for
evaluating base materials, filler metals and welded joints. There is a need
of both test methods for permanent evaluation during production of filler
metals and for special tests in critical cases as for instance hot cracking
problems the user is faced with. Here, comparative analyses are necessary
to some extent. Adequate hot cracking tests have to be available which can
be used to determine the effects of batch and welding conditions. Manufacturers
of filler metals therefore need different test methods.
For quality control, normally self-restraint test methods (tensile test,
double fillet weld test, cylinder test and ring segment test) are used [1]. For
material optimization and for the elimination of hot cracking problems, externally
loaded tests such as the Modified Varestraint Test (MVT) or the
Hot Deformation Rate Test (HDR test) [1] are necessary.
In the following report, fields of application and the benefit of different
test methods are described.
Self-Restraint Tests
Tensile Tests
A test method that is used every day by manufacturers of filler metals is
the tensile test – a tensile specimen of the all-weld-metal. Fig. 1 shows a
sample for the tensile test of austenitic weld metal of a stick electrode. The
weld metal shows a strong deformation (high elongation) and strong Re306
Testing and Standardization
duction of Area (RoA). There is no visible evidence of hot cracks. Fig. 2
shows in contrast to Fig. 1a distinctly different appearance. Weld metal alloy
230 shows a poor deformability (elongation 25 %, Reduction of Area
(RoA) 24 %) and many small cracks near the fracture zone. These are hot
cracks. Hot cracks occur during the welding process, when a critical temperature
range is passed, and are therefore part of the all-weld metal
sample.
Cracks near the surface of a tensile test sample – cut out along the
welded seam – will open towards the surface during the tensile test. Cracks
are often L-shaped and therefore called "hook cracks".
Fig. 1. Hot cracking test with the tensile specimen (all-weld metal Thermanit
19/15 W; E Z 20 16 3 Mn N L R 1 2; no hot cracks)
Fig. 2. SMAW all-weld metal alloy 230 (22 Cr, 2 Mo, 14 W, Co, Al, La, remain
Ni) open cracks are hot cracks; El. = 25 %, RoA = 24 %
Value of Different Hot Cracking Tests 307
Prior to fracture is always such a crack. The RoA is considerably
smaller than with crack-free samples. But not every surface crack of a tensile
test sample points to a hot crack.
Fig. 3 shows a crack caused by a pore near the surface. The sample
shows an excellent ability of elongation at a RoA of 78 %. Some experience
is required to derive assured findings regarding the hot cracking susceptibility
of weld metal from a tensile test. A weldment that shows a RoA
of a minimum of 40 % is, despite small isolated hot cracks, unobjectionable.
Fig. 4 shows the dependency of RoA on the number and size of hot
cracks at a CrNiMo weld [2].
Fig. 3. GTAW all-weld metal ER 80 S-Ni 2 (3,5 % Ni) open cracks caused by a
pore, no hot crack; RoA = 78 %
Fig. 4. Relationship between number of hot cracks (>1 mm) and RoA at the tensile
specimen for austenitic weld metal
308 Testing and Standardization
But RoA is not the overall indicator for all alloys. Filler metals with application
temperatures > 800 °C for petrochemistry with 0.4 % C, 25 % Cr,
1 % Si, 1.5 % Nb and 35 % Ni produce elongation and RoA values < 10 %
at room temperature (RT). With this little deformability of the weld hot
cracks covered under the surface will not be opened in every case during
tensile testing. Fig. 5 shows a sample for the tensile test of such a weld
bead.
Fig. 5. GTAW all weld metal; W Z 25 35
T-Joint Test
The T-joint test is often required in specifications of Classification Societies
and also included in the draft version of prEN 17641, part 2 [1]. This
test method is very often used for tests of filler metals for welding dissimilar
base materials. Here, the base plate should be made of low-alloyed materials.
Fig. 6 shows such a sample. The informational value of that test depends
first and foremost on the deformation resistance of the base metal plate. It
has to be thick enough or stiffened to avoid shrinking. This is described in
prEN ISO 17641, part 2 [1].
Unless otherwise agreed, filler materials of the following diameters are
used:
Stick electrode: 4 mm
Filler rods (TIG): 3 mm
Wire electrodes: 1.2 mm
Samples should be welded down hand. The first fillet weld should be
welded at a consistent thickness (A-measure) of about 5 mm.
Value of Different Hot Cracking Tests 309
Fig. 6. Standardized hot cracking test required in various specifications,
T-joint
Not later than 20 s after welding the first weld, the test weld (sealing
run) in reverse direction (each without interruption) is to be produced. The
test weld should be about 20 % thinner. After cooling down, a dye penetration
test is carried out. If cracks are detected in the first weld, the hot crack
test is not valid. The test can be carried out quickly and easily and is particularly
useful for an optimal choice of a filler metal for single-pass fillet
welds on dissimilar metals.
Cylinder Sample Test
Beside the T-joint test, there is another sample test standardized in the German
Standard DIN 50129 [3] – the cylinder sample test – which is also
suitable for performing hot cracking testing quickly and easily. Two cylinder
samples with diameters of 50 mm and of 60 mm lengths, respectively,
are necessary. Both are firmly clamped and a double-pass weld with the
filler metal to be tested is produced. Afterwards, the actual test weld will
be carried out (Fig. 7). Due to the existing prevention from shrinkage filler
metals susceptible to hot cracking will show throat cracks which will be
visible after cooling down in a surface fracture test. The result of that test
depends very much on the welding parameters and the welding procedure.
A reliable result can only be expected after many preliminary tests in
which the welding parameters and the hot cracking susceptibility of the
filler metal are synchronized. A disadvantage of this test is the fact that
similar base materials are normally not available as bar stock.
310 Testing and Standardization
Fig. 7. Hot cracking test on cylindrical specimen (length: 60 mm)
Ring Segment Test
This test is – in compliance with the German Standards – mandatory for
high-alloyed filler metals and filler metals on a Ni-base which are used in
nuclear fields of application.
The test sample (Fig. 8) consists of four quadratic segments of the same
size into which, after double-sided tack welding, a single-sided groove is
cut. Base material with a thickness of 25 mm matching the filler metal to
be tested should be used. This often cannot be realized. Therefore, compromises
have to be found with the certifying organization and the customer.
Fig. 8. Standardized hot cracking test required in various specifications,
ring segment
Value of Different Hot Cracking Tests 311
This test applies to filler metals for GTAW, GMAW and SMAW. Type
of current, polarity and inert gas have to meet the welding conditions under
which the test filler metal should not be susceptible to hot cracking.
A weld has to be carried out without oscillation or interruption starting
at about 30° before a butt joint and ending at about ¾ of circumference.
After cooling down to room temperature and cleaning of the weld surface
and the groove (brushing), the remaining length (¼) has to be welded in
the same direction. After cooling down of the sample, the cleaned ring
weld has to be analyzed for cracks using the dye-penetrant test.
The informational value of that test is very small. Improper preparation
of the sample and welding performance can cause hot cracks that cannot be
attributed to the filler metal used. In the future European Standard prEN
14532 dealing with qualifying examinations of filler metals, this test is not
considered. It should be carried out only if mandatory (production in compliance
with Standards 1408 of KTA [Nuclear Safety Standards Commission]).
Cracking Test According to Stamicarbon
In many cases the processing fabricator does not rely on the manufacturer's
standard tests but requests also a filler metal check according to his own
specifications. Stamicarbon (licensor of urea synthesis facilities) developed
a particular specification to verify the crack resistance of filler metals
with large surface cladding. Here, large surface double-layer band claddings
(SAW or RES) are welded with austenitic strips to an unalloyed
support material. Cladding joints are welded to specific gaps and laterally
with a fully austenitic stick electrode (E Z 25 22 2 L B 2 2, 25 % Cr,
22 % Ni, 2 % Mo).
Fig. 9 shows the set-up and the samples to be taken. Their surfaces will
be leveled by mechanically grinding. Afterwards, the surface will be pickled
and checked for defects under a stereomicroscope with 20-fold magnification.
Defects have to be identified as pores, metal inclusions or hot
cracks. This procedure will be repeated three times after 1 mm of the surface
has been ground off each time. After each grinding, only a certain
number of defects is accepted. This test is very demanding in regard to the
hot cracking susceptibility of filler metal. Particularly the weld metal of
the stick electrode reacts very sensitively to this test.
Fig. 10 shows an overview of an analyzed ground surface, a metal inclusion
(tolerable) and a hot crack. Both failure types are marked with
different symbols. This test shows the highest sensitivity of the hot
cracking tests presented so far.
312 Testing and Standardization
Fig. 9. Stamicarbon Test, test plate layout (urea synthetic plant)
Fig. 10. Stamicarbon Test (54035), filler metal: E Z 25 22 2 L B 2 2
Comparative studies on all-weld metals between tensile test, side bend
and longitudinal bend test with the Stamicarbon test (STAC 54035)
showed for the three different electrodes no cracks in the tensile test and
side bend test, isolated cracks in the longitudinal bending test and an unacceptable
number of cracks in the Stamicarbon test (Table 1 and Table 2).
Table 1. Chemical composition (all-weld metal); El-∅: 4.0 mm
No. C Si Mn P S Cr Mo Ni N
A 0.034 0.27 4.68 0.015 0.007 24.69 2.28 21.60 0.15
B 0.033 0.30 5.31 0.011 0.007 25.04 2.26 22.08 0.13
C 0.042 0.66 5.64 0.012 0.006 24.57 1.97 20.55 0.14
Value of Different Hot Cracking Tests 313
Table 2. Results of different hot cracking tests; SMAW: E Z 25 22 2 L B 22 (all
weld metal)
No. Tensile
Specimen
Side Bend
Test (180 °C)
Longitudinal
Bend Test
STAC 54035
A No cracks;
RoA 60 %
No defects 4 cracks Not accepted
B No cracks;
RoA 58 %
No defects 1 crack Not accepted
C No cracks;
RoA 61 %
No defects 4 cracks Not accepted
Externally Loaded Tests
Hot cracking tests during which the sample is stressed with controlled external
load are used for basic considerations in view of optimizing the hot
cracking behavior and analyzing hot cracking problems the processing
fabricator is faced with.
The most important tests in the perspective of a filler metal manufacturer
will be outlined below and their relevance will be described considering
performed analyses [1].
Programmable Deformation Cracking Test (PVR Test)
For a programmable deformation cracking test, a tensile test with a defined
rising elongation rate is carried out during welding. After exceeding the
critical value for hot crack formation, all three types of hot cracks (solidification
cracks, liquation cracks and ductility dip cracks (DDC)) can occur.
Fig. 11 shows the principle of approach. The achieved deformation rate
and deformation of the sample at the point in time at which the first forced
hot crack appears is a measure for evaluating the hot cracking susceptibility.
Fig. 12 shows the effect of the 􀄯-ferrite content on the critical elongation
rate. In the examined alloy (19 % Cr, 9 % Ni), up to about 3 FN primary
ferritic solidification is present, which causes a high deformation rate
without crack formation. Below 3 FN, the critical elongation rate decreases
strongly [4] with diminishing 􀄯-ferrite.
314 Testing and Standardization
Fig. 11. PVR Test: Hot cracking test procedure of the
controlled deformation cracking test (externally loaded test)
Fig. 12. Controlled deformation cracking test of weld metal CrNi 19 9 L;
influence of the ferrite content and the solidification mode
on the critical tension speed [4]
Value of Different Hot Cracking Tests 315
Fig. 13. Influence of the arc length on the hot cracking sensitivity of austenitic
weld metal (PVR test) [4]
Fig. 13 shows what effect a welder can exert on the hot cracking behavior.
On the left part of the tensile test sample the welder assured a short
electric arc (low content of N). This resulted in a ferrite content of about
3 %. On the right part of the tensile test sample the welder adjusted a long
electric arc that leads to a N pick-up (0.11 %). As a result, the ferrite content
decreased to less than 1 %. That necessarily leads to a higher hot
cracking susceptibility.
Hot Deformation Rate Test (HDR Test)
During welding in the HDR test, a pre-defined transverse deformation rate
will be applied which varies from test to test until the first or last hot crack
is forced (Fig. 14) [5]. This test is used to determine the susceptibility to
solidification cracking. It is particularly useful for welding parameter optimization
and appropriate selection of filler metals regarding hot crack
prevention, because the effects of the base material and of the weld geometry
as well as further variables can also be analyzed.
Fig. 15 shows the dependency of the hot cracking susceptibility of a
fully austenitic SA weld (19 % Cr, 15 % Ni, 3 % Mo, 0.15 % N) on the
welding parameters, the welding amperage, the welding voltage and the
welding speed. The motor revolutions per minute are plotted versus the
welding parameters which can be used, on the basis of the geometric data
of the test facility, to determine the critical transverse deformation rate.
316 Testing and Standardization
These analyses are carried out to prevent hot cracks during SA welding
of a liquid petrol gas container. The problem could be minimized by reducing
welding amperage and welding speed. By additional analyses using the
Modified Varestraint Test, the hot cracking susceptibility could be totally
prevented (see next section).
Fig. 14. Hot Deformation Rate Test
Fig. 15. Influence of welding amperage, voltage and welding speed on the hot
cracking susceptibility (solidification cracking); as higher the spindle drive during
the HDR-test without forcing cracks, as higher is the hot cracking resistance of the
proved material
Value of Different Hot Cracking Tests 317
Modified Varestraint Transvarestraint Test (MVT Test)
In the MVT Test, a defined bending deformation is applied to the test
sample during welding (Fig. 16) [1]. After exceeding the critical value for
hot cracking, all three hot crack types (solidification cracks, liquation
cracks and DDC) can occur. Base materials, filler metals, welding conditions,
effects of welding auxiliary materials (flux, gas) and welding procedures
can be clearly distinguished and evaluated in their effects on hot
cracking susceptibility. The test weld bead will be analyzed under the stereomicroscope
(25-fold), i.e. the individual crack lengths will be recorded,
added up and inserted into a chart showing the applied elongation by variation
of mandrel radius (Fig. 16).
Fig. 16. MVT Test
This test is used to optimize filler metals, meet customer specifications
and detect reasons for hot cracking problems. Some examples will be
given further on.
Liquid Petrol Gas Container Made of Material 1.4406
Hot cracks occurred during SA welding of a liquid gas tank (Fig. 17) made
of the material 1.4406. The wall thickness was 15–22 mm. Welding was
carried out with a fully austenitic filler metal (1.4455) and a highly basic
flux (degree of basicity according to Boniszewski B = 2.7).
Fig. 18 shows a weld cross-section of a joint. A typical hot crack of that
joint is shown in Fig. 19.
Partially such hot cracks originated from Mn-silicate lines in the base
material (Fig. 20).
318 Testing and Standardization
Fig. 17. Base metal: 1.4406;
SAW-filler: S 20 16 3 Mn N L acc. EN 12072 (ER316L mod.)
Fig. 18. SA-weld of LG-Tank; Base metal: 1.4406 (X2CrNiMoN17-11-2);
filler metal: S 20 16 3 Mn N L acc. EN 12072
MVT Tests showed that a flux with a lower index of basicity (B = 1.3)
resulted in a much better hot cracking behavior. The problem could be
solved by using this flux and by adjusting welding parameters (see section
about the HDR test). Fig. 21 is divided into three segments:
Test results located in Section I (weldable) indicate that hot cracks will
hardly occur in practice. Section II (restricted weldable) means that special
measures may be necessary to prevent hot cracks (welding parameters).
Section III (not weldable) is an area that does not permit crack-free weldValue
of Different Hot Cracking Tests 319
ing. This classification relies on a huge field of experiences of the Bundesanstalt
für Materialforschung und –prüfung, Berlin (BAM) that carried out
this test in regard to many thousand research results with this test and practical
knowledge.
Fig. 19. Hot crack in SA-weld of S 20 16 3 Mn N L (ER316L mod.)
Fig. 20. Formation of Mn-silicates lines in the base metal 1.4406
(X2CrNiMoN17-11-2)
320 Testing and Standardization
Fig. 21. SA-weldment; BM: 1.4406 (X2CrNiMoN17-11-2) ;
filler metal : S 20 16 3 Mn N L (ER 316L mod.)
Hot Cracks during SA Welding of Heavy-Walled Components of
Duplex Steel 1.4462
Duplex steel and matching filler metals are considered as unsusceptible to
hot cracks, because they solidify as ferrite. But in practice, hot cracks are
sometimes problematic when using this material during SA welding of
heavy-walled components (wall thickness > 20 mm).
In the present case, circumferential welds were welded on pipes of
42 mm wall thickness and 300 mm of diameter. Welding procedure and
welding parameters can be found in Fig. 22.
Flux with an index of basicity B = 1.8 was used for welding.
Fig. 23 shows a hot crack along the original ferritic grain boundary.
Even by varying the welding parameters, no crack-free welds could be
achieved. Cracks were revealed in side bend tests. The MTV test helped to
determine optimal wire / flux combinations. In this case, a highly basic
flux B = 2.7 was successfully used.
Value of Different Hot Cracking Tests 321
Fig. 22. Welding sequence and parameters of circumferential welding on DSS
1.4462 (UNS S 31803); matching filler metal
Fig. 23. Hot crack in the DSS weld (SAW; 1.4462, UNS S 31803), crack proceeding
along the former ferrite grain boundaries
Fig. 24 shows the differing hot cracking susceptibility of different SAW
fluxes for this application. A reduction of the amperage to 300 A or even
250 A also generated advantageous effects (Fig. 24). In addition to this,
MVTs proved that the hot cracking susceptibility diminishes with an increasing
ferrite content in all-weld metals (Fig. 25) [6].
322 Testing and Standardization
Fig. 24. SA-weldment; BM: DSS 1.4462 (UNS S 31803);
filler metal: S 2209 (ER2209)
Fig. 25. SA-weldment; BM: DSS 1.4462 (UNS S 31803)
matching filler metals; from different heats with different
ferrite contents
Value of Different Hot Cracking Tests 323
SA Welding of Alloy 617
As a rule, nickel base filler metals are more susceptible to hot cracking
than stainless steel ones. Therefore, a smaller wire diameter is used for SA
welding. During SA welding with a wire matching to alloy 617, increasing
formation of hot cracks occurred. A wire with a diameter of 1.6 mm and a
highly basic flux (Marathon 444) was used for welding.
Tests with different current densities but constant heat input led to a
prevention of hot cracks. Instead of the wire with a diameter of 1.6 mm, a
wire with a greater diameter was used, whereas the amperage of the 1.6
mm wire was kept, that means current density was reduced because of the
greater wire diameter. This leads to colder droplets, and hence to lower
heat input. Fig. 26 shows this dependency in the MVT.
Fig. 26. Influence of wire-∅ in the SAW-process on the hot cracking susceptibility
of alloy 617
Hot Cracks during Welding of Alloy 690
The alloy 690 is characterized in the references as very sensitive to hot
cracking. This material is said to be particularly susceptibile to ductility
dip cracking (DDC). This crack appearance is not characteristic in comparison
to other hot crack appearances, because it occurs in alloys with a
324 Testing and Standardization
low degree of contamination. Hot cracks in alloy 690 welds are caused by
grain boundary precipitation as illustrated in Fig. 27.
Fig. 28 illustrates that different batches of wire show a different hot
cracking behavior.
Fig. 27. GTA-weld of alloy 690; interdendritic cracks (hot cracks); grain size segregation
Fig. 28. Hot cracking behavior of different wire heats of alloy 690
Analyses underlying those MVT’s are shown in Table 3. A comparison
of the different analyses, however, gives no hint of different hot cracking
susceptibilities.
Value of Different Hot Cracking Tests 325
Table 3. Different alloy 690-heats; wire analysis (in wt.-%); heats with different
hot cracking behaviors in the MVT Test
heat C Si Mn P S Cr Mo Ni Fe Al Ti
A 0.029 0.25 0.33 0.006 0.004 29.4 0.01 rem. 9.11 0.60 0.30
B 0.017 0.23 0.29 0.003 0.003 29.4 0.02 rem. 9.37 0.74 0.30
C 0.032 0.25 0.33 0.008 0.003 29.3 0.09 rem. 9.85 0.77 0.30
D 0.030 0.22 0.27 0.003 0.002 29.4 0.01 rem. 8.70 0.51 0.39
E 0.030 0.20 0.25 0.003 0.002 29.4 0.01 rem. 9.14 0.68 0.31
It seems that different effects play a role, e.g. the Ti carbon nitrides that
can be recognized in Fig. 29 and that occur in a linear shape in the hot
rolled wire (starting product for filler metals). Those wires led to an increased
hot cracking susceptibility.
Fig. 29. Alloy 690; Ti-carbonitrides in the wire
The potential hot cracking susceptibility of these alloys can be seen in
Fig. 30. Beside the weld in the HAZ, several transversal cracks can be distinguished
which are classified as DDC. An evaluation of MVTs is shown
in Fig. 31. The high susceptibility to DCC is clearly shown. But also hot
cracks within the GTAW-remelted bead have a great length (section II: restrictedly
weldable).
The hot cracking susceptibility of matching filler metals of this alloy
can be reduced to a higher extent by adjustment of welding parameters or
also of the welding process than by a restriction of the chemical composition
of the hot rolled wire. The production process of the hot rolled wire
seems to have an effect, too.
326 Testing and Standardization
Fig. 30. Solidification cracks in the weld of alloy 690; Preheat Solidification
Cracks and Ductility Dip Cracks in the HAZ; heat input: 7.5 kJ/cm;
elongation 2 %
Fig. 31. Solidification Cracks (SC) and Ductility Dip Cracks (DDC)
in SMA-weld of alloy 690
Value of Different Hot Cracking Tests 327
Conclusions
Manufacturers of filler metals use several hot cracking tests, particularly
for filler metals being susceptible to hot cracking. Self-restraint tests are
not always informative. Therefore, in critical cases and mainly for the disclosure
of reasons for hot cracking at a component, externally loaded hot
cracking tests are necessary.
In addition to metallurgical examinations, externally loaded tests such as
the PVR or MVT tests should also be used to determine the hot cracking
behavior and to be able to take appropriate measures. Here, the base material
should be taken into consideration. If a base material is susceptible to
hot cracking, one cannot expect that a matching filler metal shows a considerably
lower hot cracking behavior. For that reason, welding parameters
or welding procedures have to be adjusted to the hot cracking susceptibility
of the material. To do this, externally loaded test are the appropriate
tools.
References
1. prEN ISO 17641: Destruction tests on welds in metallic materials – Hot
cracking tests for weldments – Arc welding processes
Part 1: General
Part 2: Self-restraint tests
Part 3: Externally loaded tests
2. Pohle C (1985) Möglichkeiten für die Beurteilung der Heißrissanfälligkeit von
austenitischem Schweißgut. Schweißen und Schneiden 37, Heft 2: 498–502
3. DIN 50129: Prüfung metallischer Werkstoffe; Prüfung der Rissanfälligkeit
von Schweißzusatzwerkstoffen. Beuth Verlag
4. Tösch J, Schabereiter H, Perteneder E, Rabensteiner G (1997): Bedeutung und
praktische Beeinflussbarkeit des Ferritgehaltes bei der Schweißung austenitischer
Stähle. Schweiß- und Prüftechnik Wien, Heft 2: 18–26
5. Wilken K, Kleistner H (1985) Remarks on the classification and evaluation of
hot cracking tests. IIW-Doc IX-1379-85
6. Heuser H, Groß V, Ladwein T (1997) Hot cracking problems with submerged
arc welding of heavy wall components from 22 % chromium duplex stainless
steels. Duplex Stainless Steels 97 – 5th World Conference 1997, Maastricht,
The Netherlands; KCI Publishing BV; D97–034
Influence of the Deformation Rate
of Different Tests on Hot Cracking Formation
H. Herold, A. Pchennikov, M. Streitenberger
Institut für Füge- und Strahltechnik,
Otto-von-Guericke-Universität Magdeburg, Germany
Abstract
Referring to the ISO standardization of hot cracking test procedures with
externally loaded specimens, three different and fundamental test procedures
are assessed with the help of experiments and finite element analyses
to find out the influence of different deformation rates on the test results of
three well known stainless steels.
A comparison is made between the hot tensile test carried out with help
of the Gleeble system using the Moscow procedure of the Standard of
Russian Federation, the American instructions for the Hot Ductility Curve
as well as the PVR-test (controlled deformation crack test) created in
Austria. These main test procedures have generally been used for more
then 30 years.
While within the American Hot Ductility Curve Test a round tensile
specimens is torn after heating by thermal resistance to a test temperature
some degrees below liquation at a tension speed of about 50 mm/s, the
Russian method heats all round tensile specimens to the same maximum
temperature just above the melting point and loads each specimen at another
tension speed between 0.01 and 0.1 mm/s for approximation of the
critical tension speed.
For tensile test comparison, the PVR test is carried out on round specimens
with a diameter of 10 mm. A TIG melt run is simultaneous welded
longitudinally at a superposed tension speed linearly increasing from zero
to 1 mm/s in welding direction. The length position of the first crack appearing
at the specimen defines the critical tension speed.
Three different stainless steels have been subjected to these test procedures.
The test results of the different procedures precisely correlate within
the materials ranking, but differ in the amount of evaluated test criteria.
Influence of the Deformation Rate of Different Tests 329
Introduction
Hot cracking tests are conducted with the following different procedures
using self restraint specimens or externally loaded specimens. Most of
them relate to the theories of Prokhorov [1]and Matsuda [2]. They are
described in Drafts of ISO standards [3–5].
While the theoretical hot cracking research is focused on local thermomechanical
kinetics and metallurgical dynamics during solidification, the
practically applied hot cracking test procedures are based on global concepts
with measurable constitutional, mechanical, thermal and timeaffected
test criteria. Their development began with descriptions of hot
cracking events by yes/no criteria (test procedure with self loaded specimens
[4]) to quantify the hot cracking sensitivity by thermo-physical simulations,
and finally to tests direct in welds during the welding process [5].
For example, the hot tensile test (Gleeble test) quantifies by the hot ductility
curve the Brittleness Temperature Range (BTR), where the ultimate
tensile strength and the reduction in area are plotted dependent on both the
heating and the cooling process as a function of the test temperature. Both
types, the Varestraint Test and the Transvarestraint Test, assess the hot
cracking sensitivity by a diagram of total crack length versus applied surface
bending strain longitudinally or transversally to the welding direction
of a TIG-bead-on-plate weld. The PVR test (controlled deformation crack
test) defines the hot crack initiation by a critical tension speed for the first
hot crack of each hot cracking type.
Former comparisons of the results of different hot cracking tests with
externally loaded specimen showed by [6, 7] that there is a good consistency
in the cracking sensitivity of the selected four materials tested, although
the test procedures and test criteria are very different.
The hot cracking theories by Prokhorov and Matsuda are based on a
thermomechanical concept using the deformation rate, evaluating the equilibrium
between initiation of and resistance against solidification cracking.
The deformation rate is the mathematical derivation of the internal deformation
per temperature within the Brittleness Temperature Range (BTR).
Its dimension [%/°C] is the percentage of the changed deformation per
temperature. Hot cracking appears when this deformation rate dε/dT
during solidification within the BTR is higher than the critical deformation
rate.
The preference of test procedures with external load is their possibility
of quantifying the hot cracking sensitivity of high-alloyed materials and of
deposited filler metal as well as the reactions of applied welding procedures
to the test materials.
330 Testing and Standardization
But opposite to the theory all test procedures work with different application
of the tension speed. Thus, different test conditions are used to superpose
thermal cycles by simulation or during welding and the tension for
external deformation.
It still has to be clarified by which factors the test procedures are
varying and how this is influencing the different test results.
Test Materials
The materials used for the comparison of test procedures were two austenitic
stainless steels and one ferritic free-machining stainless steel, characterized
by their Cr-equivalents and Ni-equivalents given in Table 1 and
their chemical compositions given in Table 2.
Table 1. Tested stainless steels and their Cr-Ni-equivalents
AISI Material Mat. No Cr-Equi
WRC 92
Ni-Equi
WRC 92
Cr-Equi
1948
Ni-Equi
1948
304 X5CrNi18-10 1.4301 19.13 13.48 19.00 10.78
316L X2CrNiMo17-12-2 1.4404 20.32 13.90 19.35 11.50
430F X14CrMoS17 1.4104 18.28 11.02 18.15 5.02
Table 2. Chemical composition of tested stainless steels
Mat.
No
C
[%]
Si
[%]
Mn
[%]
P
[%]
S
[%]
Cr
[%]
Mo
[%]
Ni
[%]
Nb
[%]
Ti
[%]
N
[%]
1.4301 0.05 0.37 0.96 0.034 0.048 18.2 0.29 8.68 0.018 0.006 0.09
1.4404 0.02 0.40 1.87 0.045 0.053 16.3 2.39 10.00 0.076 0.007 0.08
1.4104 0.12 0.42 2.64 0.015 >0.096 17.2 0.29 0.22 0.005 0.007 0.20
In Fig. 1 the test materials are characterized with their Cr-Ni-equivalents
within the Schaeffler Diagram. This Schaeffler Diagram was also classified
by Bystram [8] to show possible sensitivities and failure during welding
of stainless alloys.
While both austenitic stainless steels would tend to hot cracking because
of the small portion of δ-ferrite, the ferritic free-machining stainless steel
could be sensitive to hardening brittleness because of its higher C-content.
The contents of 0.6 % sulphur and 0.2 % nitrogen are interesting in view
of hot cracking resistance.
Influence of the Deformation Rate of Different Tests 331
Fig. 1. Characterization of test materials within the Schaeffler-Diagram referring
to Bystram [7]
Test Procedures and Results
Evaluation of Solidification Crack Initiation During Welding
Investigations of centerline solidification cracking in ship building industries
have stimulated the calculation and measurement of cross displacements
during tandem-submerged arc welding of larger plates. The measurements
were taken for solidification crack-free welding design. They
resulted in the assessment of the critical value for solidification-crack initiation,
too [9].
A welding process-associated hot-cracking test method was created
(Fig. 2) to measure the value and the speed of cross displacement which
initiates the local solidification cracking. Such displacements are produced
by plate movements due to assembly effects. The measuring device works
with two cross gauges versus the welding direction.
332 Testing and Standardization
Fig. 2. Measuring device with two special cross gauges measuring the cross displacement
uy (t) versus time during one side welding of large components
The total displacements between the gauges across the weld seam were
measured during the whole welding process over the seam length of about
10 to 18 m during welding fabrication in ship building industries. The
speed of cross displacement vq was calculated by its temporal change for
this special welding process. The occurrence of hot cracking or the crackfree
welding was examined by non-destructive weld inspection of a large
number of fabricated large components.
The assessment of centerline solidification crack initiation during welding
processing is based on the balance between the values of cross displacement
and its changes versus time, distinguishing between welding assembly
which initiates hot cracking and such which yields the movement
of the gap without any hot cracking during welding [10].
The condition for hot crack initiation is given when the speed of cross
displacement between the welding components is higher than the materialassociated
speed of cross displacement of the applied welding procedure.
Influence of the Deformation Rate of Different Tests 333
Fig. 3. Model and determination of the critical speed of cross displacement (tension
speed) by the threshold curve within the diagram of cross displacement versus
time during welding processing
The left hand side in Fig. 3 shows the model for the thermal tension vertical
to the weld. The welding process on larger plates moves the components
apart at the instant at which the observed point is passed by the welding
torch. Above on the right, Fig. 3 shows the thermal cycle of the applied
process.
The assessment diagram on the right shows the estimation of the critical
speed of cross displacement (tension speed) during real welding. In this
diagram the curves of measured speeds of cross displacement vq are plotted
as the cooling time of the weld metal versus the temporal change of the
cross displacement Δuy(t) in [mm] and valuated according to the appearance
of hot cracking.
A threshold curve separates the conditions for hot crack initiation (grey
line) and hot cracking resistance (bold line), as a result of low amounts of
cross displacement and different speeds of cross displacement.
For example, the critical tension speed for the applied tandemsubmerged
arc welding of 10 mm thick shipbuilding plates was found to
be 0.04 mm/s, the critical amount of cross displacement was about
0.05 mm within the gap.
334 Testing and Standardization
Hot Cracking Test Procedures Based on Analytical Simulation
of Welding Procedures
The hot cracking test procedures which are based on analytical thermomechanical
simulation use round tensile specimens with diameters of about
10 mm in contrast to the real weldment. The round specimens are arranged
in accordance with the model of centerline solidification cracking in Fig. 3
and with the model of the Moscow MIS-test procedure in Fig. 4 created by
Prokhorov from his theories. The resistant heated central zone of the round
tensile specimen is in accordance with the tension direction vertical to the
weld. The thermo-physical simulation with the Gleeble®-3500 System superposes
the thermal cycle on the deformation using two different controlled
test procedures.
Fig. 4. Moscow hot cracking test procedure MIS, standardized by the Russian
Federation
Determination of Hot Cracking Sensitivity by the Simulated
Moscow Test Procedure
The simulation of the Moscow test procedure MIS needs several round test
specimens which are heated up to the maximum temperature just above the
liquation temperature. The same thermal cycle is applied to all tests. Each
specimen is strengthened at another tension speed between 0.001 and
0.1 mm/s. The heating requires varied heating rates deviating from those of
the real welding thermal cycle due to the specimen geometry. The thermomechanical
superposition includes several steps (Fig. 5).
At first the heating rate is controlled by free expansion (Force = 0) up to
1000 °C, measuring the thermal extension. The slower heating rate up to
the maximum temperature of about 1450 °C starts after setting the extension
to zero.
Influence of the Deformation Rate of Different Tests 335
Fig. 5. Simulated Moscow test procedure for determination of critical tension
speed for solidification cracking
The deformation at varied tension speeds of the other specimens starts
after a short holding time at the maximum temperature under free cooling.
The critical tension speed for solidification cracking results from the empirical
approximation of crack-free yielding or hot cracking.
The effect of the simulated Moscow test procedure on the test materials
is shown in Fig. 6–8 with the simulation records plotting only two varied
tensions speeds near the critical one.
AISI 304 in Fig. 6 yields a tension speed of about 0.010 mm/s with increasing
force without any hot cracking. The same material failures after
loading with 0.012 mm/s tension speed.
The crack initiation is marked in the test record by the loss of force , determining
a temperature for solidification crack initiation of about
1320 °C. Subsequently, the critical tension speed for crack initiation of the
solidifying material is given by the threshold curve at vcr= 0.011 mm/s. It
also defines the lowest temperature of the BTR (brittleness temperature
range).
336 Testing and Standardization
Fig. 6. Test record of the simulated Moscow test procedure of AISI 304
Fig. 7. Test record of the simulated Moscow test procedure of AISI 316 L
Influence of the Deformation Rate of Different Tests 337
Fig. 8. Test record of the simulated Moscow test procedure of AISI 430 F
Fig. 9. Comparison of results obtained from the simulated Moscow test procedure
Various tension speeds induce two different crack mechanisms at the
molten stainless steel AISI 316 L (Fig. 7). Solidification cracking was
determined at a critical tension speed of about 0.015 mm/s.
338 Testing and Standardization
Lower tension speeds in Fig. 7 result in failure at 1100 °C characterized
by the type of ductility dip cracking (DDC). The critical tension speed for
DDC was assessed to be 0.01 mm/s.
The hot cracking behavior of the ferritic stainless steel (Fig. 8) is not
surprising considering the very low critical tension speed of about
0.006 mm/s, initiating solidification cracking already at 1420 °C. The
graph in Fig. 9 compares the quantified hot cracking sensitivities determined
in the simulated Moscow test for the three stainless steels.
Determination of Hot Cracking Sensitivity by the Standard
Procedure of the Hot Tensile Test
The hot tensile test procedure is based on the analytical simulation of weld
thermal cycle, distinguishing between controlled base metal heating up to
the maximum temperature and cooling after exceeding the maximum temperature
for the component section under examination.
The analyzed thermal cycle is interrupted at different chosen temperatures
for each specimen by always the same high stroke rate (tension speed
of about 50 mm/s) both during heating and during cooling (Fig. 10).
Fig. 10. Simulation programs for hot tensile tests during heating and during
cooling (above),
analytical modeling of a thermal cycle for a welding process (below)
Influence of the Deformation Rate of Different Tests 339
Fig. 11. Results of the hot tensile tests obtained during heating and cooling of the
stainless steels
The test procedure during heating was carried out in correlation with the
heating process of the simulated Moscow test by varied steps and onset of
stroke after a holding time of about 1 s at test temperature.
The question mark in Fig. 10 notes that heating can be done by free expansion
of with clamped specimen. This hot tensile test was carried out
with clamped specimen.
The test results obtained during heating are plotted as the reduction of
area versus the test temperature in Fig. 11. The nil ductility temperature
(NDT) for all test metals was found to be near 1360 °C. The diagram is
completed in principle for the cooling process by a curve of a modified
AISI 316 from Lundin 1988 [11].
Added test results from cooling AISI 316 L down to two temperatures
indicate a decreasing reduction of area within the ductility dip temperature
range in comparison to the tests during heating. At 1100 °C, there is a
27 %-decrease in reduction of area during heating and cooling.
The ductility dip at this temperature correlates with grain growth and
grain boundary effects (Fig. 12) during passing the maximum temperature
of the thermal cycle and holding at a temperature exceeding 1100 °C. This
results in significant grain boundary strengthening of AISI 316 L.
340 Testing and Standardization
Fig. 12. Comparison of AISI 316 L microstructure sections for testing
during heating and cooling, respectively
Fig. 13. Microstructure after testing during heating at maximum temperature
Influence of the Deformation Rate of Different Tests 341
The applied maximum temperatures of the thermal cycles during heating
result in characteristic microstructures of the pulled zones (Fig. 13).
While AISI 304 solidifies from 1390 °C as austenite with low embrittlement,
AISI 316 L solidifies from 1360 °C as ferrite without any cracking.
The remaining ferrite content at room temperature is about 6 %. The
ferritic free-machining stainless steel embrittles at 1360 °C by partially
liquated MnS particles.
PVR-Test (Controlled Deformation Crack Test, Longitudinal
Tension Crack Test or Controlled Flat Tension Test)
The PVR test represents a test procedure for evaluating the effects of the
welding procedure with regard to hot cracking [12]. Only one single test
specimen is required to determine the hot cracking sensitivity of a base
metal and/or a weld metal during welding.
Fig. 14. Initiation of solidification cracking, liquation cracking and
ductility dip cracking on one specimen (above) and effective superposition
of real thermal cycles in the heated zones during the applied welding procedure
with the deformation rate ε’ induced by the linearly increasing tension speed
during welding (below)
342 Testing and Standardization
The ordinary PVR test procedure in Fig. 14 uses flat specimens with the
dimensions of 40 x 10 x 300 mm (width x thickness x length), clamped
into a special tension fixturing device, lengthened in a horizontal servohydraulic
system of the test equipment. The welding process with a constant
welding speed is superposed by a linearly increased tension speed
vPVR in welding direction.
The standard PVR test procedure is carried out at tension speeds linearly
increasing from zero to 1 mm/s, using bead-on-plate TIG-welding with argon
shielding gas and applying two types of heat input per unit length, i.e.
about 7 kJ/cm (Is = 180 A, Us = 12 V, vs = 19 cm/min) and 10 kJ/cm, respectively.
The critical tension speed vcr is the test criteria for the PVR test.
vcr corresponds to the length coordinate of the first hot crack detected visually
on the specimen surface at a magnification of 25. For each of the hot
crack types it is possible to determine solidification cracking, liquation
cracking and ductility dip cracking (Fig. 14).
The possibility for to assessing all hot cracking types within one specimen
depends on the superposition of the local heated zone beside the
welding bead at the applied tension rate.
Different indices for the critical tension speed vcr are used, depending
on the size and number of visible cracks. For example: vcr1st is the critical
tension speed for the first microscopically visible hot cracking feature, vcr3
describes the first three hot cracks per 10 mm of weld bead and vcr9 determines
the first nine hot cracks per 10 mm of weld bead. Modeling of the
PVR test [13] has shown the correlation between test criterion, hot cracking
theory by Prokhorov, and its applicability.
For a comparison with the tensile test results the PVR test is performed
on round 10 mm diameter specimens. A TIG-melt run is simultaneously
welded longitudinally at a superposed linearly increasing tension speed
from zero to 1 mm/s in welding direction.
Fig. 15–17 combine the length position of cracks within the specimen
both with their critical values to quantify the hot cracking sensitivity and
with their microstructure for the classification of the hot cracking type.
Primarily, all types of hot cracks disappear vertical to the welding and tension
direction.
AISI 304 in Fig. 15 is sensitive to liquation cracking beside the fusion
line because of the critical tension speed vcr9liqu of about 15 mm/min
(0.25 mm/s). The resistance against solidification cracking is higher. The
first micro-solidification cracks are initiated by the liquation cracks at vcr 1st
SC of about 36 mm/min (0.6 mm/s) directly at the fusion line. Macrosolidification
cracks were not observed in the round specimens up to
60 mm/min (1 mm/s).
Influence of the Deformation Rate of Different Tests 343
Fig. 15. Determination of vcr of hot cracking in the PVR test of AISI 304
Fig. 16. Determination of vcr of hot cracking types in the PVR test of AISI 316 L
344 Testing and Standardization
Fig. 17. Determination of vcr of hot cracking types in the PVR test of AISI 430 F
AISI 316 L develops preferably DDC within the ductility dip temperature
range in some distance from the fusion line already at vcr DDC of about
13 mm/min (0.22 mm/s) (Fig. 16). The micro-solidification cracks are detectable
near the fusion line only above the high critical tension speeds of
about 40 mm/min (0.66 mm/s). Macro-solidification cracks were not found
in the specimens below 60 mm/min (1 mm/s).
Fig. 18. Comparison of PVR-test results from tested stainless steel
Influence of the Deformation Rate of Different Tests 345
The ferritic free-machining stainless steel AISI 430 F in Fig. 17 exhibits
classical solidification cracking. The first micro-solidification crack occurs
in comparison with the other materials at the critical tension speed of about
20 mm/min (0.33 mm/s) (Fig. 18) followed by macroscopic solidification
cracks above vcr.
Conclusion
The type of hot cracking test procedure influences the probability for quantifying
the sensitivity to different hot cracking types by its specific superposition
of thermal cycle and deformation rate.
Each test procedure programs the conditions for materials characterization
in its own way. The material characterizing results depend on the success
of superposition of the simulated welding thermal cycle or real welding
with a specific deformation, independently of the question of whether
the test was carried out using a tension speed or a deformation rate.
Each procedure has its preference:
− The simulated Moscow test procedure determines the sensitivity to
solidification cracking and DDC as illustrated Fig. 5 globally. The critical
tension speed for a specific hot crack type in the test material is estimated
as a threshold curve from a number of tests carryied out at
different tension speeds between 0 and 0.1mm/s;
− The classic hot tensile test procedure uses maximum temperatures below
the liquidus temperature rupturing several specimens at a high tension
speed of about 50 mm/s and different temperatures during heating
and cooling as shown in Fig. 10 and 11. The test results are quantified
by the deformation (reduction in area in %) and characterized by the
microstructure and fracture behavior, classifying the material-specific
crack mechanisms (liquation, solidification, grain boundary migration,
segregation, precipitation, reheating, e.g.) by means of additional testing
techniques;
− The PVR test determines the critical deformation rates for hot crack initiation
quantifying the first solidification crack initiation in the weld
metal and the different crack types within the HAZ during one single
test welding procedure globally.
The application of different hot cracking test procedures results in equal
ranking of material characteristics, independent of the deformation rate and
of the specific test criteria. The hot cracking sensitivities, quantified with
the help of different test criteria on three known stainless steels, agree with
346 Testing and Standardization
the tendency of hot cracking types detected in practice on the condition
that suitable deformation rates and characteristic thermal cycles of the
applied three test procedures are used.
References
1. Prokhorov NN (1962) The technological strength of metals while crystallisation
during welding. Welding production Vol 9: No 4 April: 1–8
2. Matsuda F, Hashimoto T, Senda T (1969) Fundamental investigation on solidification
structure in weld metal. Trans Nat Res Inst Metals 11: 43–58
3. PrEN ISO 17641-1, Nov 2003. Destructive tests on welds in metallic materials
– Hot cracking tests for weldments – Arc welding processes – Part 1: General
(ISO/FDIS 17641-1:1003)
4. PrEN ISO 17641-2, Nov. 2003. Destructive tests on welds in metallic materials
– Hot cracking tests for weldments – Arc welding processes – Part 2 : Selfrestraint
tests (ISO/FDIS 17641-2:1003)
5. prCEN ISO/TR 17641-3 Destructive tests on welds in metallic materials – Hot
cracking tests for weldments – Arc welding processes – Part 3 : Externally
loaded tests (ISO/DTR 17641-3:1003)
6. The Japan Welding Society (1986) Hot cracking susceptibility evaluation of
austenitic stainless steel weld materials. International Institute of Welding
(Doc IIW IX-1395-86)
7. Wilken K (1999) Investigation to compare hot cracking tests – Externally
loaded specimen. International Institute of Welding (Doc IIW IX-1945-99)
8. Bystram MCT (1956) Some aspects of stainless alloy metallurgy and their application
to welding problems. British Welding Journal. Febr: 41–46
9. Herold H, Streitenberger M, Pchennikov A (2001) Prevention of centreline
solidification cracking during one side welding. International Institute of
Welding (Doc IIW IX-2000-01 (and II-C-220-01)
10. Herold H, Pchennikov A, Streitenberger M (2003) Current problems in hot
cracking research described on the example of PVP test. Materials Science
Forum 426–432: 4093–4098
11. Lundin CD, Lee CH, Menon R, Osorio V (1988) Weldability evaluations of
modified 316 and 347 austenitic stainless steels: Part I – Preliminary Results.
Welding Research Supplement 67 Febr.: 35-s – 46-s
12. Folkhard E (1984) Metallurgie der Schweißung nichtrostender Stähle. Springer
Verlag, Wien New York, p 153
13. Herold H, Streitenberger M, Pchennikov A (2001) Hot Cracking Theory by
Prokhorov and Modelling of the PVR-test. Welding in the World 45, ¾: 17–
22
Testing for Susceptibility to Hot Cracking
on GleebleTM Physical Simulator
S.T. Mandziej
Advanced Materials Analysis, Enschede, The Netherlands
Abstract
Hot cracks appear when thermal shrinkage together with deformation
caused by restraint cannot be accommodated by plastic deformation. This
happens during welding to such alloys, which segregate on heating and
cooling at near-solidus temperatures, in particular when low-melting and
mechanically weak phases form and occur over a wide range of temperatures.
To check for susceptibility to the liquation cracking caused by
the low-melting, weak phases, hot tensile testing can be used in combination
with a thermal cycle resembling that of real welding. This procedure,
which can be executed on a Gleeble™ 1 thermal-mechanical simulator,
comprises tensile testing of a number of cylindrical samples at the temperatures
below solidus and determining their hot strength and ductility.
For measuring of the brittle temperature range (BTR), the nil strength temperature
(NST) is determined and used as the peak point, down from which
the ductility recovery temperature (DRT) is searched for. The ductility is
measured after the tensile test as a reduction in area at fracture. An alternative
to the hot tensile test during the simulated welding cycle is the straininduced
crack opening (SICO) test, in which a rod-like sample mounted in
“cold” copper jaws of the Gleeble is heated by electric current and then
compressed till formation of a bulge in its uniformly heated central portion
and appearance of cracks due to secondary tensile strain developed along
the maximum perimeter of this bulge. Next to the studies of liquation
cracking and ductility-dip cracking of the reheat-type, the Gleeble procedures
can be also used to determine sensitivity to solidification cracking
1 Gleeble is a registered trademark of Dynamic Systems Inc., Poestenkill, NY,
USA.
348 Testing and Standardization
and for this the samples are melted and solidified in a controlled manner
before the hot tensile testing or the SICO testing. The developed routines
of the Gleeble testing allow accurate determination of temperatures at
which the cracks occur as well as measurement of critical strains to fracture
and strain rates which are associated with the hot cracking. At an appropriate
geometry of samples and optimum setting of the experiments, the
hot cracks generated in the samples have sizes comparable to those that
occur in heat-affected zones of real welds or in weld metals.
Introduction
To date, over 150 weldability tests exist, many of which are designed to
assess the susceptibility of welds to hot cracking. In general, they can be
put into two categories as representative (self-restraint) and simulative (externally
loaded) test techniques [1]. The representative test technique
usually tells only “cracking” or “no-cracking” of a material when an actual
welding situation is represented, which cannot quantify the cracking susceptibility
of the material under different welding condition. The simulative
test can follow a thermal-mechanical history of a material during
welding, while an external strain is usually applied to produce cracks or to
record material characteristics allowing quantification of cracking susceptibility.
Hot cracking of welds appears when shrinkages of the solidifying and
thus stiffening weld metal and of the adjacent heat affected zone cannot be
compensated or accommodated by at first the back-filling (liquid flow) of
the molten material and afterwards by the visco-plastic flow of the solidified
hot material of the welded joint. The process is dynamic and to avoid
hot cracking the welded material must be able to plastically deform at
critical sites of the welded joint with the strain rate exceeding that of the
shrinkage during the relatively fast heat cycle of the welding.
Studies in the 1950's carried out in the USA by W.F. Savage at Rensselaer
Polytechnic Institute and at Duffers Associates Inc., led to the development
of the Gleeble thermal-mechanical simulator which allowed the
hot ductility of welded materials during thermal cycles of the welding
processes to be determined [2]. This study showed that the region most
susceptible to hot cracking is the heat affected zone of the parent metal, in
which contaminants entrapped at grain boundaries form liquid or low
strength solid films while the grains become stiff and strong. It was also
found that if such weak films exist over a large temperature range after solidification,
the welded materials show hot cracks in the HAZ.
Testing for Susceptibility to Hot Cracking on Gleeble™ 349
To determine the range at which the weld HAZ is prone to hot cracking,
a concept of nil strength temperature was introduced as the higher temperature
of the brittle range, and appropriate attachments were designed to
measure it. The lower temperature of the brittle range, so-called nil ductility
temperature, was then taken as that at which 5 % reduction in area on
hot tensile samples appeared [3].
The Gleeble testing procedure required a large number of samples to be
hot tensile tested with strain rates representative of various welding
methods (heat inputs), and this stimulated a study to develop another, simpler
test, compatible with the deformation rates already known from the
Gleeble testing. As a result, the Varestraint test was proposed in the early
1960's by W.F. Savage and C.D. Lundin [4], and applied to study the hot
cracking susceptibility of welded alloys.
The Varestraint test comprises bending a test plate while the weld bead
is being made on the long axis of the plate. The original Varestraint test
had some limitations, e.g. difficulty in controlling the real amount of strain
at the outer bent surface due to the position of a neutral bending axis,
which varied depending on the strength and strain partitioning between the
hot and cold parts of the sample during bending.
Fig. 1. Weld HAZ liquation crack appearing in a small distance from
the fusion line i.e. nucleating at nil strength temperature (NST) lower than
the solidus temperature Ts; marked ductility recovery temperature (DRT)
near which the crack is closing
350 Testing and Standardization
Fig. 2. Schematic presentation of the brittle temperature range (BTR) below
solidus temperature (Ts) relating the hot crack susceptibility to critical strain (CS)
and strain rate (CST)
The Gleeble testing method is dealing with the occurrence of the brittle
temperature range (BTR) at high temperature during the solidification of
the weld and the subsequent cooling. The BTR reflects the general appearance
of the cracks occurring in heat-affected zones of welds (Fig. 1), and
can be illustrated on the graph proposed by Prokhorov [5], as shown in
Fig. 2. The graph shows that for any welded alloy there is some critical
strain value, which can be accommodated at very low cooling rates by, for
example, self-diffusion controlled visco-plastic flow of the hot metal. The
crack in HAZ does not form exactly on the fusion line between weld metal
and base metal; there is always certain space in which high diffusion rate
and back-filling of the nucleating cracks or voids allow healing of the base
metal. The hot cracks in HAZ occur in a small distance from the fusion
line when the strain rate caused by shrinkage during cooling of the weld is
sufficiently high for the solidifying material to "enter" the BTR. The wider
the BTR of the alloy, the higher the hot crack susceptibility.
What is a Gleeble?
The thermal-mechanical simulator called Gleeble originated from the
welding thermal cycle simulator built in 1948 at Rensselaer Polytechnic
Testing for Susceptibility to Hot Cracking on Gleeble™ 351
Institute in Troy, NY, USA, which soon after (1951) was equipped with
pneumatic deformation system and more recently with dynamic servohydraulic
system (1979) and computer controls (1980). Its developer, Dr.
W.F. Savage, wrote in one of his review articles [6]: “In 1946 Nippes and I
attempted to evaluate the notch toughness of various regions of the heataffected
zone of ship steel weldments by careful placement of the notches
in Charpy V-Notch specimens machined from weldments. However, the results
were biased by the sloping nature of the heat-affected zone, which
caused the fracture to traverse more than one portion of the zone. Consequently,
we decided to develop a thermal simulator and utilize temperature
measurement data accumulated during the earlier cooling-rate studies at
Rensselaer.”
By its design, Gleeble has been dedicated to reproduce weld thermal
cycles characteristic of the heat-affected zones with consequences related
to thermal-mechanical effects caused by thermal gradients, restraints and
shrinkage deformations. For this the up-to-date Gleeble uses “bulk”
samples of 10 mm or more diameter and in appropriate parts of the thermal
cycle can exert tensile or compressive deformations with strain rates
adequate to those of the real welding process. The amounts of deformation
and strain rates resulting from cooling rates and thermal gradients can be
calculated using contemporary computer modeling techniques. The last
should take into account apparent strengthening caused by thermal
gradients and strain rates [7].
The thermal-mechanical situation in HAZ during welding comprises in
the thin element of the base metal parallel to the fusion plane and longitudinal
to the welding direction twice crosswise and lengthwise compressions
on heating and on cooling and once on cooling a tension in the direction
normal to the fusion plane, which is also the main direction of the heat
flow.
For the electric arc welding it also involves the presence of electric current
and related electro-thermal effects. Gleeble accounts for these by using
AC electric resistance heating uniform through the cross section of the
specimen.
This situation is schematically presented in Fig. 3. To simulate the HAZ
in Gleeble, rod-like specimens mounted in water-cooled copper grips are
heated at a rate resembling the weld thermal cycle and a short span
between the grips assures that the heated zone is about twice the width of a
real weld HAZ (Fig. 4).
When necessary, controlled tensile or compressive strains can be added
in any point of the thermal cycle. The thermal cycle in the Gleeble is controlled
by a thermocouple, percussion welded to the surface usually in the
middle of the specimen.
352 Testing and Standardization
Fig. 3. Schematic presentation of the heat affected zone formed
during electric arc welding with heat flow from the weld metal
to parent plate causing thermal gradient
Fig. 4. Simulation of HAZ situation on a rod-like specimen
mounted in water-cooled copper jaws of the Gleeble
The ability of Gleeble to deform heated specimens can be effectively
used for hot tensile testing. The length of the uniformly heated zone in the
middle of the Gleeble’s specimen results from the balance between electric
current heating the specimen and the heat flow towards the mounting
grips.
Testing for Susceptibility to Hot Cracking on Gleeble™ 353
Fig. 5. Stainless steel “hot” jaws of Gleeble for mounting of hot tensile specimens;
note reduced contact portion between the jaws and the specimen
Depending on the span between the grips, as well as on conductivity of
the grip material and on size of the contact area between the specimen and
the grips, this length can be manipulated to a reasonable extent.
The photographs in Fig. 5 and in Fig. 6 show an example of the “hot”
Gleeble jaws made of stainless steel, and the resulting substantial length of
the uniformly heated zone on the specimen mounted in such “hot” jaws.
Fig. 6. Elongated uniformly heated zone on a plain carbon steel specimen
mounted in the “hot” stainless steel jaws of the Gleeble
354 Testing and Standardization
Hot Ductility and Hot Cracking
As the hot cracks form on-cooling when tensile strains caused by shrinkage
and assisted by restraint cannot be compensated by ductility of an alloy,
then method of studying susceptibility of an alloy to hot cracking
should involve tensile testing at the conditions simulating these of the real
welding (or casting) process. The use of thermal-mechanical simulator like
Gleeble, able to reproduce on a specimen the welding thermal cycle and
impose a strain in a controlled manner, allows achieving this goal.
Testing on a Gleeble for susceptibility to weld liquation cracking / HAZ
hot cracking, means hot tensile testing of a number of specimens onheating
and then on-cooling, and determining their hot ductility measured
as a reduction in area at the specimen’s neck portion after the test. This
procedure is schematically presented in Fig. 7 [1].
Fig. 7. Schematic of Gleeble’s procedure for hot ductility testing, including hot
tensile testing on-heating up to NST and then on-cooling after weld thermal cycle
with NST as the peak temperature; note the NST being lower than the melting
point TL
The hot ductility of a weldable alloy increases gradually with increase
of testing temperature from ambient towards melting point, however before
reaching this point it drops abruptly from certain maximum to nil. Just
above this nil ductility temperature (NDT) appears the nil strength temperature
(NST) at which the alloy looses its strength due to formation of
weak or liquid phases along grain boundaries. The real physically measurable
melting temperature of such alloy –TL, is higher than NST, as shown
in Fig. 7.
Testing for Susceptibility to Hot Cracking on Gleeble™ 355
On-cooling from the melt or from the NST, the ductility does not
recover exactly at NDT but below it at so-called ductility recovery
temperature – DRT. The temperature span from NST to DRT is considered
to be the brittle temperature range – BTR.
The extent of BTR can be used as a rough criterion of the susceptibility
to hot cracking, however more exact criterion is the measure how fast does
the ductility recover on-cooling as compared with its decrease on-heating.
As the reference point for this measurement the maximum of ductility
from the on-heating ductility curve is taken and the representative areas
below the on-heating and on-cooling curves are compared (Fig. 8) [8]. Arbitrarily
considering 5 % of reduction-in-area on-cooling as the ductility
recovery point and comparing the hot ductility curves on-heating and oncooling,
the nil ductility range (NDR = BTR), ductility recovery rate
(DRR) and ratio of ductility recovery (RDR) can be determined to exactly
characterize the susceptibility of an alloy to hot cracking.
Fig. 8. Evaluation of hot ductility curves for hot cracking susceptibility [8];
– RDR = ration of ductility recovery,
– DRR = ductility recovery rate,
– NDR = nil ductility range (or BTR),
– Point F = nil ductility temperature (NDT) on-heating,
– Point D = ductility recovery temperature on-cooling
The reference point for the ductility measurements determining the susceptibility
to liquation cracking is the nil strength temperature. This temperature
has to be used as a peak of the welding thermal cycle, on-cooling
after which the hot tensile tests should be run. To measure the NST on
Gleeble an attachment is used, schematically presented in Fig. 9.
356 Testing and Standardization
Fig. 9. Nil strength attachment of Gleeble using trapezoidal spring grip
maintaining constant small tensile load on specimen during heating
This NST attachment keeps the specimen under a constant tensile load
of about 50N while allowing specimen heating with an initial heating rate
the same as of the welding thermal cycle to be simulated up to a temperature
50 °C below the solidus temperature of the material, then change to a
heating rate of 2–5 °C/s until the NST is reached, as shown in Fig. 10.
Fig. 10. Schematic time–temperature graph of Gleeble test
for nil strength temperature measurement
Testing for Susceptibility to Hot Cracking on Gleeble™ 357
Fig. 11. Gleeble setup for controlled melting and solidification study
To avoid mixing of phenomena related to liquation cracking with those
of solidification cracking, in the weld thermal cycles simulated to measure
hot ductility for the liquation cracking the NST must not be exceeded.
To test for the susceptibility to solidification cracking, controlled melting
and solidification of a rod-like specimen is carried out on Gleeble, and
after the solidification the hot ductility is determined in the manner as described
above. Here the ability of Gleeble to melt electrically conductive
specimens is used. In this test the central portion of the specimen protected
by a crucible / quartz sleeve is brought to a temperature above solidus and
then this crucible contains the molten / semi-liquid metal, as shown in
Fig. 11.
Fig. 12. Schematic of Fig. 11, with end nuts for hot tensile testing
358 Testing and Standardization
The thermal gradient between the molten portion and mounting jaws
prevent the metal from flowing out of the crucible while controlled thermal
cycle allows conducting the solidification in a manner similar to that of
real casting or welding. Schematic of the assembly used for this purpose in
Gleeble is given in Fig. 12.
More details of the Gleeble testing for liquation and solidification cracking
can be found in a Gleeble application note [9] available upon request
from DSI in USA (info@gleeble.com). The Gleeble hot tensile testing procedures
for liquation and solidification cracking are also mentioned in the
technical report CEN ISO/TR 17641-3:2003, under the chapter “Hot Tensile
Test” [10].
Fig. 13. SICO sample tested in Gleeble
An alternative procedure to study susceptibility to solidification cracking
on Gleeble is the strain-induced crack opening test - SICO, developed
at Dynamic Systems Inc., in USA, for studying hot deformability of alloys.
In this test the central hot portion of the Gleeble specimen is compressed to
form a bulge (Fig. 13), on outer perimeter of which cracks appear at the
critical secondary tensile strain, as shown in Fig. 14.
The critical strain to fracture in a SICO test is defined as the hoop strain
at onset of cracking in the bulge zone:
εc = ln (Df / Do), (1)
where Do is the initial diameter of the specimen, while Df is the final
maximum diameter in the bulge zone.
Testing for Susceptibility to Hot Cracking on Gleeble™ 359
Fig. 14. Schematic of SICO test showing crack formed on bulge portion of sample
As during the controlled melting and solidification in the Gleeble the
dendritic crystals grow mainly in the direction of the heat flow i.e. in the
axial direction, the mid-span segregation may occur in the specimen causing
deep and partly hidden cracking of the SICO specimen along the central
plane perpendicular to the compression axis (Fig. 15). In such situation
it is advised to check for the true critical diameter on the cross-section of
SICO sample, like it is shown in Fig. 16.
As the hot tensile testing on Gleeble gives the adequate characteristics
of an alloy regarding its hot cracking susceptibility, the SICO test appears
to be more accurate for measuring of critical strains to fracture and related
critical strain rates.
Fig. 15. Central plane crack in SICO often appearing
after melting and solidification
360 Testing and Standardization
Fig. 16. Method of measuring critical diameter Dcr
when central plane crack appears
Hot Cracking of Austenitic Steels and Alloys
Austenitic stainless steels, Ni-base alloys and other metal alloys having
large thermal expansion coefficients are susceptible to hot cracking during
welding. In most of these alloys the high content of alloying elements extends
their solidification temperature range thus increasing segregation of
impurities along grain boundaries. These alloys are susceptible to solidification
cracking in weld metal and to liquation cracking in the HAZ, depending
on the relative strength of the HAZ and weld metal in the hot
range, and to avoid cracking both the HAZ and weld must be able to plastically
accommodate the shrinkage strains.
The hot strength of austenitic stainless steels and Ni-base alloys below
solidus is almost zero in a certain range of temperature till NST, and below
the NST the metal becomes rigid and strong but its ductility is still be very
low. During cooling through the low ductility range, the thermal shrinkage
may cause hot cracks, and the susceptibility of steel to this cracking can be
expressed as a difference between the NST and the ductility recovery temperature
- DRT. Below the DRT, the steel is able to plastically accommodate
the shrinkage thus avoiding cracks.
To determine the hot ductility for some austenitic alloys and their susceptibility
to hot cracking, Gleeble hot tensile tests were carried out at DSI
for a collaborative program of IIW – Commission II, of which some results
are presented here. The full report of this testing [11] is available from DSI
upon request at info@gleeble.com. Four austenitic steels and alloys of the
composition given in Table 1 were examined. Test specimens were 6mm
diameter by 100 mm long. They were heated to fracture in a nil-strength
test fixture under a load of ~8kgf. The measured NST for each alloy, as
well as the determined BTR and RDR, are listed in Table 2.
Testing for Susceptibility to Hot Cracking on Gleeble™ 361
Table 1. Chemical composition of tested alloys, Fe is balance to 100 %
chemical composition [wt%]
Alloy C Si Mn P S Cr Mo Ni Ti Al Others
800H .068 .38 .70 .011 .002 20.40 _ 30.55 .33 .28
AC66 .072 .21 .50 _ _ 27.35 _ 31.45 _ .014 Nb=.83,
Ce=.085
926 .010 .32 .82 .017 .003 20.85 6.38 24.80 _ _ Cu=.91,
N=.196
825 .006 .32 .69 .015 .003 22.25 3.16 39.15 .77 .09 Cu=1.9
Table 2. Nil-strength temperature (NST), brittle temperature range (BTR) and ratio
of ductility recovery (RDR) for each alloy
Alloy 800H AC66 Alloy 926 Alloy 825
NST 1365°C 1331°C 1315°C 1343°C
BTR (°C) 115 94 40 71
RDR (%) 29.9 21.6 71.7 28.2
The results, given in Table 2, have shown a substantial difference in the
behavior of Alloy 926 and 800H, and they also showed a good agreement
with the hot cracking susceptibility data available from the modified
Varestraint test and from practical observations [12].
Selected examples from these hot strength / hot ductility tests are given
below on the graphs of Alloy 926 and Alloy 800H (Figs. 17 and 18), and
one may see from them that the real hot ductility curves are not exactly so
“smooth” like these in Figs. 7 and 8. The overlapping and competing
effects of liquation and homogenization may result in a solid-state diffusion
healing dependent on applied time / temperature / strain rate of the
executed hot tensile test. Thus, for the applied strain rate representative to
the shrinkage rate of the weld thermal cycle ductility dips were observed at
1240 °C and 1170 °C for alloys 800H and 926 respectively, while for
slower strain rates these dips were absent.
This last observation calls for metallographic verification of the tests, in
particular for comparing the microstructures of test specimens with these
362 Testing and Standardization
of real welds containing hot cracks and for identifying the micromechanisms
of cracks formation at the conditions of the hot tensile tests.
What follows, is a part of metallographic verification of the specimens
from the mentioned IIW testing programme, which revealed various microstructural
features coinciding with the microstructures of the real welds
as well as some discrepancies further used to modify the Gleeble testing
conditions.
Fig. 17. Hot strength and hot ductility curves of alloy 800H
Fig. 18. Hot strength and hot ductility curves of alloy 926
Testing for Susceptibility to Hot Cracking on Gleeble™ 363
In Alloy 926, very well resistant to hot liquation cracking, substantial
grain growth appeared near to NST however no grain-boundary segregation
/ liquation could be noticed.
Samples tested at 1250 °C showed existence of long segregation bands
filled with weak / low-melting eutectics (Fig. 19) however when tested at
1200 °C these bands almost entirely disappeared due to diffusion healing
(Fig. 20).
Fig. 19. Longitudinal cavities containing eutectic near to fracture zone
of alloy 926 tensile specimen tested at 1250 °C
Fig. 20. Isolated longitudinal cracks near to fracture zone
of alloy 926 tensile specimen tested at 1200 °C
364 Testing and Standardization
In a more susceptible to hot cracking alloy 800H, a rapid grain growth
occurred near to the NST (Fig. 21) which resulted in formation of thin
films as well as voids along grain boundaries (Fig. 22).
Fig. 21. Abnormal grain growth near to surface of specimen
in alloy 800H at nil strength temperature (NST)
Fig. 22. Formation of grain boundary films and cavities in alloy 800H near to NST
Also chains of equiaxial voids appeared along segregation bands, however
at the highest testing temperatures these voids did not easily coalesce
to cause the failure – the weakest link of the microstructure appeared to be
the grain boundaries (Fig. 23).
At temperatures lower than NST, i.e. at 1250 °C and below, elongated
cracks appeared along the segregation bands and from them intergranular
Testing for Susceptibility to Hot Cracking on Gleeble™ 365
Fig. 23. Cavities in segregation bands close to fracture
in alloy 800H tensile tested at NST
cracks extended in transverse directions (Fig. 24). The behaviour of alloy
800H indicated that various micromechanisms could operate at different
testing temperatures and that the hot crack susceptibility is strain rate
dependent.
Fig. 24. Longitudinal cracks along segregation bands
and transverse intergranular cracks in alloy 800H tested at 1200 °C
It was possible to find on the longitudinal sections of the hot tensile
specimens the regions at which the highest density of intergranular transverse
cracks appeared and using the formula:
ε = 2 ln (D/d) (2)
366 Testing and Standardization
to determine the local critical strains as well as strain rates for the hot
cracking at particular temperatures.
An example below (Fig. 25), shows the section through neck portion of
hot tensile specimen, on which this relation is drawn and the following two
pictures (Figs. 26 and 27) present cracks, which appeared in alloy 800H at
temperature 1200 °C and strain rate 1.20 /sec while no cracks formed at
the same temperature and strain rate 0.30 /sec.
Fig. 25. Method of determining strain from the neck portion
of tensile test specimen: ε = 2 ln (D/d)
Fig. 26. Cracks revealed in tensile test specimen of alloy 800H
tested at 1200 °C, at region deformed with strain rate ~ 1.20 /sec
Testing for Susceptibility to Hot Cracking on Gleeble™ 367
Fig. 27. No cracks visible in tensile test specimen of alloy 800H tested at 1200 °C,
at region deformed with strain rate ~ 0.30 /sec
In another case, of alloy AC66, the loss of ductility due to liquation was
quite severe and the recovery of ductility on-cooling delayed. Here, the liquation
caused a permanent precipitation of a carbide eutectic, which evidently
hampered deformability of the alloy.
The precipitates appeared in segregation bands (Fig. 28), assisting in the
hot tensile test formation of transverse cracks nucleating from voids at
these precipitates.
Fig. 28. Carbide eutectic near to fracture and on the fracture surface
of hot tensile specimen of alloy AC66 tested at 1225 °C
368 Testing and Standardization
In Fig. 29 carbide eutectic is shown, which separated during the test
forming a deep transverse void. These precipitates in alloy AC66 visibly
assisted the hot cracking for more than 100 °C down from the NST, mainly
by the formation of voids around them. Fracture of the hot tensile specimens
near NST occurred mainly by joining of these voids in the transverse
direction thus indicating additional weakness of grain boundaries at this
temperature (Fig. 30).
Fig. 29. An example of a cavity formed during the hot tensile test
across the carbide eutectic in alloy AC66
Fig. 30. Extended transverse crack near to the fracture portion
and numerous voids in alloy AC66 tensile tested at NST
Testing for Susceptibility to Hot Cracking on Gleeble™ 369
The eutectics appeared elongated in the rolling direction of the tested
material (Fig. 31), suggesting its metallurgical low quality. Nevertheless,
at lower temperatures a substantial recovery of ductility appeared, due to
thermal-mechanical grain refinement (Fig. 32), most probably resulting
from a dynamic recrystallization in the regions of strain localization and
concentration between the elongated fields of the eutectics.
Fig. 31. Elongated carbide eutectics in alloy AC66 after heating
up to NST and cooling down; total strain ~ 0.05
Fig. 32. Refined grains and coarse primary grains formed
next to the precipitates of eutectics in alloy AC66 at the portion
of tensile specimen deformed with strain rate ~ 0.50 /sec at 1175 °C
370 Testing and Standardization
This last phenomenon “successfully” competed with the formation of
voids near to the eutectics, giving a high value of reduction-in-area and
simultaneously generating in the neck portion of the specimen a large
amount of voids surrounded by fine recrystallized grains (Fig. 33).
Fig. 33. Refined grains and numerous voids formed in the neck portion
of hot tensile specimen of alloy AC66 deformed with strain rate ~ 2.50 /sec
at 1175 °C
In conclusion it may be said that the metallurgical quality of an alloy is
important as regards its susceptibility to hot cracking and that sometimes
the “standard” test conditions may need to be adjusted in order to reach
correct results.
Microfissuring in Multi-Bead Welds
Microfissures are reheat-type fine cracks of length about 1 mm or less
when visible on transverse sections of welds and up to a few millimeters in
the length direction of the weld. They are often of a ductility-dip origin
and form in inter-bead heat affected zones of multi-layer, multi-bead
welds, almost exclusively in the upper layers of these welds, i.e. when underlying
portion of the weld is stiff enough to provide adequate restraint.
The microfissures are related to thermal-mechanical history of the weld
manufacturing and assisted by the primary segregation of weld metal solidification
and the secondary reheating liquation as well as by solid-state
embrittling processes resulting in ductility dips.
Testing for Susceptibility to Hot Cracking on Gleeble™ 371
They form in the multi-pass welds in zones of adjacent (underlying)
weld beads, which constitute the heat-affected zone of a subsequent pass;
an example is given in Fig. 34.
Fig. 34. An example of microfissure(s) appearing in inter-bead heat-affected zone
of a multi-bead weld, visible on a cross-section of the weld
To explain the formation of microfissures the following model [13] can
be used (Fig. 35) in which the inter-bead heat-affected zones are divided
into two portions: the first marked (a) representative to higher temperature
and coarse grain microstructure and the second marked (b) representative
to fine-grained microstructure and incomplete recrystallization.
While welding with the bead sequence: (1) → (2) → (3) → (4) → (5) →
(6) and so on, after laying down the bead #(6) the most potential site for
microfissure formation will be reheated by the bead #(6) coarse-grained
zone of the HAZ below the bead #(5) at the site where it overlaps with the
HAZ of the bead #(6). The (a) and (b) zones of the interbead HAZs can be
also treated as the upper one (a) in which the diffusion healing and the nilductility
appear, and the lower one (b) in which the ductility has recovered
substantially however strain hardening may occur. Thus at the bottom of
the lower layer the thermal cycling may results in a slight strengthening
due to the generation of dislocation and their incomplete annihilation /
recovery, without any substantial annihilation of the primary crystalline
lattice defects being mainly vacancies in the “as-frozen” weld metal.
In general, the microstructure of the as-solidified weld metal beads is far
from thermodynamic equilibrium by containing large amounts of
crystalline lattice defects, such as vacancies and dislocations as well as
planar defects, which are formed to compensate shrinkage effects and
372 Testing and Standardization
which tend to annihilate during subsequent thermal cycling. Thus, when
during the laying down of the bead #(6) such microstructure in the bead
#(5) is heated up to a temperature near to solidus, a large number of defects
annihilate and this process is affected at first by thermal expansion of
the whole interbead HAZ, and then by shrinkage restrained in some particular
directions. An intensive annihilation of vacancies and grain growth
causing local reduction of volume is expected to occur during heating in
the upper layer (a6) of the HAZ by the overlying bead (6), i.e. close to the
fusion line, due to restrained expansion.
Fig. 35. Schematic presentation of the multi-bead weld
with sites where the microfissures do form;
(1)÷(6) – sequence of weld beads,
(a1+b1) ÷ (a6+b6) – inter-bead heat-affected zones
Then on cooling, this zone shrinks and may be particularly prone to nucleation
of cracks if its ductility is low. However, the simultaneous
shrinkage of the already solidified weld metal in the bead #(6) acts to
compensate this effect.
The expansion and its resulting compressive strain may have a positive
effect that is the acceleration of homogenization and partial healing in the
upper layer (a6) of the HAZ, and these may reduce the sensitivity to liquation
cracking. Additionally, the shrinkage after solidification of the weld
metal bead #(6) may close partly the liquation cracks in the upper layer
(a6) of the HAZ.
When sensitization occurs, the microfissures nucleate in the upper or in
the lower layer of the interbead HAZ, depending on which effect prevails.
The critical sites for their formation are intersections with the coarse
Testing for Susceptibility to Hot Cracking on Gleeble™ 373
grained part of the HAZ (a5) of the previous bead #(5) where laying down
of this bead caused the liquation. In such site which is relatively remote
from the centre ‘C’ of the weld bead #(6) as is visualized in Fig. 35, the
shrinkage of the weld bead #(6) after its solidification and “anchoring” between
points ‘A’ and ‘B’, can result in localization of a substantial tensile
strain, thus opening the microfissure in the a5-b6 area. The location of
microfissures at the intersection with the HAZ of previous beads and frequent
propagation along boundaries of primary columnar grains into the
underlying bead, indicate that the segregation and liquation are important
factors for their formation.
To study the susceptibility of the multi-bead weld joints to microfissuring
as well as the influence of microfissures on hot strength and ductility
of the weld metals, SICO testing can be used on Gleeble. Samples for such
testing are cylindrical bars of 10 mm diameter by 90 mm long, and are
clamped from both sides in the Gleeble’s cold jaws. The simulated weld
heating cycle is then applied and the sample can be compressed immediately
at the peak temperature or at any temperature on cooling from the
peak.
To check for the presence of microfissures and their influence on hot
ductility SICO tests were run on samples taken from the upper half and
lower half of an austenitic stainless steel multi-bead weld, and the cylindrical
bar samples were machined transversely to the weld length, like
schematically presented in Fig. 36. Then in the Gleeble the samples were
heated at a heating rate of 150 °C/sec to a test temperature, and compressed
at a speed of 50 mm/sec to different reductions. The mean strain
rate was approximately 3 /sec at a free span of 30 mm.
Fig. 36. Schematic presentation of SICO specimens’ extraction from lower and
upper portions of multi-bead weld
374 Testing and Standardization
Fig. 37. Results of SICO testing for microfissure susceptibility of multi-bead austenitic
stainless steel weld metal [13]
For the SICO test specimens cut from the top layer of the multi-pass
weld metal, critical strains were lower at each temperature at the same testing
condition than that from the bottom layer with rare microfissures, as
shown in Fig. 37. Moreover, it was found that all the cracks appeared at
the specimen surface directed to the top of the weld, which indicates the
presence of microfissures in the top layers of the multi-pass weld joint.
The microfissures simply act as nuclei or microcracks for crack opening
during compression under secondary tensile stresses. This conforms with
the applied physical model of microfissuring. It has been concluded that
the strains at which the weld metal cracked at the temperature and strain
rate of the SICO test, or the microfissures extended to the surface of the
SICO sample and appeared visible, can be treated as characterizing the
susceptibility of weld metal to microfissuring.
Summary and Conclusions
The examples of Gleeble thermal-mechanical simulation procedures presented
in this paper, show the results of simulation that can be achieved if
the fundamental physical phenomena occurring during welding is well understood
and obeyed. They highlight an important role of thermal gradients,
which occur in all known industrial thermal processes and resulting
Testing for Susceptibility to Hot Cracking on Gleeble™ 375
from these gradients, thermal-mechanical strains and strain accommodation
phenomena which in turn affect strain hardening and recrystallization,
phase transformations and/or precipitation processes. The combinations of
different factors, and their importance, may vary substantially, and the correct
physical simulation of welding must take into account a number of interacting
phenomena appearing in the real application process, such as:
1. The balance between the heat input and electric current flow during
heating, and controlled by thermal gradient heat flux, cooling rate and
micro-deformation rate during the short thermal cycle of welding, in the
case of heat-affected zone during arc welding.
2. The changes of hot ductility and of hot strength of welded material on
heating and on cooling and its susceptibility to form liquid phase along
grain boundaries at elevated temperatures below solidus, in the case of
hot cracking.
3. The accommodation of strains occurring due to multiple welding thermal
cycles and due to annealing microstructural restraints in multi-bead
austenitic weld metals, in the case of micro-fissuring.
References
1. Lin W, Lippold JC, Baeslack III WA (1993) An Evaluation of Heat-Affected
Zone Liquation Cracking Susceptibility. Part I: Development of a Method for
Quantification. Welding Journal 72: 135–153
2. Savage WF (1955) An Investigation of the Hot Ductility of High-Temperature
Alloys. Welding Journal 34: 183–196
3. Nippes EF, Savage WF, Grotke G (1957) Further Studies of the Hot-Ductility
of High-Temperature Alloys. Welding Research Council, Bulletin 33
4. Savage WF, Lundin CD (1965) The Varestraint Test. Welding Journal 44:
433–442
5. Prokhorov NN (1968) Theorie und Verfahren zum Bestimmen der Technologischen
Festigkeit von Metallen beim Schweissen. Schweisstechnik 1: 8–11
6. Savage WF (1987) A Historical View of Weldability. In: Welding Metallurgy
of Structural Steels, TMS, Warrendale PA, USA, pp 3–19
7. Ferguson HS (1992) Fundamentals of Physical Simulation. In: Proceedings of
the International Symposium on Physical Simulation, TU Delft, The Netherlands,
pp 1–21
8. Lundin CD, Qiao CYP, Gill TPS, Goodwin GM (1993) Hot ductility and hot
cracking behavior of Modified 316 Stainless Steels designed for high temperature
service. Welding Journal 72: 189–200
9. Chen WC (1995) A Standard Procedure for Hot Cracking Test. Gleeble Application
Note, Dynamic Systems Inc, Poestenkill, NY, USA, (IIW Doc II-C-
042A/95)
376 Testing and Standardization
10. CEN ISO/TR 17641-3:2003 (2003) Destructive tests on welds in metallic materials
– Hot cracking tests for weldments – Arc welding processes – Part 3:
Externally loaded tests
11. Chen WC (1995) Hot Tensile Ductility Test for HAZ Liquation Cracking Susceptibility
Study. Report #950426, Dynamic Systems Inc, Poestenkill, NY,
USA, (IIW-Doc II-C-042/95)
12. Wilken K (1994) Investigation to Compare Hot Cracking Tests – external
loaded specimens. BAM, Berlin, Germany, IIW Doc IX-H-305/94
13. Mandziej ST (1997) Physical Simulation of Welding. In: Welding and Joining
Science and Technology, ASM International Europe, Brussels, pp 253–268
Scientific Bases of the
International Standardization Project
"Hot Cracking Tests for Welds"
B. Yakhushin, D. Semin
Baumann University, Moscow, Russia
Abstract
Microcracks which are hardly detectable by non-destructive test methods
represent a potential risk for the reliability of components or structures.
More important than improving the postweld failure detection procedures
is the improvement of the material-inherent resistance against hot cracking
and the optimization of the welding process. This requires respective consistent
test procedures which are currently being internationally standardized.
The present contribution provides some remarks which might be considered
for this standardization project.
Introduction
For the assessment of hot cracking resistance, national standards for respective
test procedures have been released in many countries. However,
their operation principles vary greatly and the applied hot cracking criteria
are conditional and only to a limited degree quantitatively comparable. In
this connection it is worthwhile to notice that a comparison between hot
cracking test results of four different alloys investigated in the USA,
France, Austria, Germany, Slovakia and Russia is described in the final report
of K. Wilken [1].
Non-standardized test devices and the lack of statistical and scientific
data may even limit the free trade of base material and filler material
within the framework of the World Trading Society.
It is therefore the objective of the recent investigations to gather the
respective “hot cracking test experience” from different countries and to
378 Testing and Standardization
work out a scientific concept regarding the ongoing multinational standardization
discussion.
In principle, a general standard should allow comparative and quantitative
assessment of the hot cracking resistance of base materials, filler materials
and their common behavior under real welding conditions [2]. The
hot cracking theory has to be understood as a basis for such standardization
and a method to really measure the hot cracking resistance has to be
derived and welding technology has to be assessed.
The Hot Cracking Criterion and Theory
According to the frequently cited theories, hot cracks (HC) are caused by a
special kind of high-temperature brittleness originating from the nonequilibrium
states of a material during welding. Like any brittlenessrelated
material defect, a hot crack can thus be regarded as the consequence
of insufficient material plasticity. The following respective criteria
must occur simultaneously to initiate hot cracking:
1. Existence of a brittleness temperature range (BTR) at temperatures close
to or within the melting range
2. The minimal plasticity δmin is exceeded within this BTR (δmin > 0)
3. The critical deformation rate dε/dt is exceeded within the BTR, whereby
the deformation rate can also be related to the cooling rate by dε/dT.
The function of the temperature related brittleness is schematically plotted
in Fig. 1 following the theory of Prokhorov. It can be seen that the
straight line E1, representing the strain increase during cooling, intersects
the brittleness function, which means that the maximum ductility is exceeded
and hot cracking is initiated.
Every alloy has a particular range of plasticity within its BTR. The plasticity
can partly or completely be reduced by the thermomechanical influence
of the welding process. The reason is that the thermomechanically induced
material displacements lead to monotonously increasing tensile
deformation behind the weld pool caused by weld metal shrinkage during
cooling. The material-specific and quantitative criterion of any alloy is
thus the critical deformation rate (Bcr). Exceeding the Bcr limit causes a
complete loss of plasticity as well as hot crack initiation.
The value of Bcr combines the effects of the three basic criteria mentioned
above. The smaller the BTR, the higher the minimal plasticity and the
higher the critical deformation rate, the higher is the resistance of an alloy
against hot cracking during welding measurable by a higher Bcr.
Scientific Bases of the International Standardization Project 379
Fig. 1. Illustration of Prokhorov´s theory of hot cracking formation
Method for Measuring the Hot Cracking Resistance
During the past decades, different machine and technological test methods
have been developed in which a specimen is deformed during welding at
constant or increasing strain rates. Internal and external material deformations
are thus superimposed. Such externally loaded hot cracking tests allow
the determination of each of the three criteria separately.
But the measurement of the mechanical high-temperature properties and
the calculation of the criterion Bcr = δ min/BTR - Δe/BTR is complex.
Therefore, such methods which might be applicable in research cannot be
recommended as a general standard.
There are especially some technical difficulties concerning the measurement
of the minimal high-temperature plasticity. This means that the
determination of criteria-related parameters under welding conditions is
inexact.
A second variant by which defined specimen deformation is generated
under isothermal conditions is more precise, but does not correspond to
real welding processes [2].
A third variant of plasticity testing consists in creating high-speed
deformation during welding, thus reducing the blurring influence of the
380 Testing and Standardization
moving heat source [3]. However, the plasticity of a material under highspeed
deformation in solid-liquid state might be decreased to zero, which
does not correspond to the theory as stated above. The contradiction between
theory and experiment becomes especially clear under the consideration
that hot cracks would be engendered in any alloys if zero plasticity
exists, which has not been proven in practice, yet.
Integrated quantitative assessment seems to be more promising, however.
The respective criterion is represented by the sum of all influencing
factors and can be determined by a machine method if the above stated
theory visualized by Fig. 1 is taken as a basis.
For this purpose, a specimen is deformed during welding with different
strain rates, so that the weld metal and the heat affected zone (HAZ) are
subjected to different mechanical loadings.
This means that from the upper limit of the BTR, which corresponds to
a temperature at which the crystals get in touch with each other, up to the
lower limit, which is between 100 °C and 200 °C below the solidus temperature,
a material is subjected to different strain rates which gives clear
and real information of its hot cracking resistance. This test variant preserves
the kinetic adequacy of the real welding process and must therefore
be considered as the superior method.
In this connection it has to be remarked that one fundamental dilemma
of hot cracking testing still remains undissolved, which is the detection of
hot cracks, the measurement of their geometries and the determination of
the number of cracks.
Furthermore it has to be noticed that there might be a further classification
for hot cracks, because cracks emerging within BTR1 are differently
oriented compared to those emerging within BTR2 [4].
Additionally to weld metal testing, the machine test methods allow also
the measurement of the critical deformation rate of the heat affected zone
(HAZ) in the base material. The determination of the critical deformation
rate of the HAZ is possible by subjecting a specimen to a simulated weld
thermal cycle up to the temperature of "nil strength" (NST) and by applying
variable strain rates during cooling.
The respective critical deformation rate is determined according to the
following equation:
Bcr = vcr / vbtr [mm/°C], (1)
whereby vcr represents the critical deformation rate (lowest rate at which
hot cracks occur) and vbtr represents the cooling rate within BTR [4].
Scientific Bases of the International Standardization Project 381
The Assessment of Welding Technology
The application of the Bcr criterion allows the assessment of the influences
of different welding parameters on the hot crack formation.
The diagram of weldability shows schematically the functional dependency
of Bcr on welding parameters (Fig. 2).
Fig. 2. Diagram of weldability
The general criterion of the welding condition q·vw was applied for determining
the crystallization formation and this function turned out to correlate
with the critical deformation rate of an alloy, which has been proved
for a wide range of welding conditions.
It was found, that the greatest influence on hot crack formation is exerted
by the crystallization pattern which is described in detail elsewhere in
this book [4].
The application of standard machine and technological methods allows
the assessment of the hot cracking probability. For this purpose, testing of
the corresponding standards specimens is performed using different welding
conditions in order to detect critical welding parameters causing hot
cracks.
382 Testing and Standardization
Conclusions
The technological methods developed so far might be suitable for the assessment
of technological parameters (especially welding parameters)
causing a susceptibility to hot crack formation. But, by application of machine
methods, in particular in terms of different, i.e. increasing, strain
rates, an assessment of a material-inherent hot cracking susceptibility
might be achieved providing the following advantages:
− The hot cracking resistance can be defined separately for deformations
along and across the weld direction. The anisotropy of the crystallization
pattern is therewith taken into account.
− The hot cracking resistance for homogeneous welds is defined both in a
solid-liquid stage and in a temperature range below the solidus temperature.
Especially hot cracking below the solidus temperature depends on
specific alloying elements [4].
− The hot cracking resistance can be determined both for the weld metal
and the HAZ, allowing the selection of an optimal base material / filler
material combination
For these reasons, a general standard should specify a hot cracking test
device capable of introducing universal external loads.
References
1. Wilken K (2000) Investigation to Compare Hot Cracking Tests – Externally
Loaded Specimen. IIS/IIW – Subcommission IX-H/II-C, Final Report 2000,
Doc.: IX-1945-99
2. GOST 26389 - 84 Welded joints: Test methods of metals on hot cracking resestivity
during welding. By Nikolaev G, Yakushin B et al., M.: USSR Standards
1984- 23
3. Prokhorov N, Yakushin BF, Prokhorov NN (1968) Theorie und Verfahren
zum Bestimmen der technologischen Festigkeit von Metallen während des
Kristallisationsprozesses beim Schweißen. Schweißtechnik vol.19, N1: 8–11
3. Stan T, Mandziej. Physical Simulation of Welding. Advanced Materials
Analysis, P.O. Box 3751, 7500 DT Enschede, Netherlands
4. Yakhushin B (2004) Morphology of Hot Cracks in Single-Phase Weld Metal.
In this book
Discussion and Evaluation of Some
Extraordinary Cases of Hot Cracking
K. Wilken
Germany
Abstract
Reasons for hot cracking formation are frequently present. Metallurgical
aspects represent only one conceivable contingency. Some years after the
active period there is time to clear up the writing desk. It is amazing and
interesting how many unanswered and unclear cases exist. Some of them
are worth to be reflected again. Four of them will be presented in this
contribution:
1. What sort of influence on hot cracking is created by the welder himself?
2. Hot cracking occurs in a material which normally is hot crack-proof.
What is the reason?
3. Is it possilble to weld an extremly hot crack-sensitive material nearly
without hot cracks?
4. What is the reason for hot cracking in the heat affected zone of the base
material?
Case 1 – What sort of influence on hot cracking is created
by the welder himself?
Some years ago, a test program was carried out by IIW SC IX H and SC II
C for comparative investigations and scatter studies with the new Longitudinal
Bend Test (LBT) [1], now standardized in CEN ISO 17641 Part 2.
Test coupons were welded in seven different countries. A straight and precise
welding instruction had been established before. Each participant had
to follow this instruction. The welding consumables to be used and additional
instructions were handed out to each participant. Everybody had to
384 Testing and Standardization
cut out his own test specimen and to check the microfissure sensitivity indicator
(MSI).
The results revealed a lot of differences [2], as shown in Table 1.
Table 1. Longitudinal Bend Test ( LBT). Comparison of MSI results
Test coupon
welded by
Number of runs
top layer
MSI 10-3
mm-1
Soudometal 3 0
Avesta 6 1.3
Esab 5 2.0
Böhler 4 14.3
BAM 4 16.6
MSI microfissure sensitivity indicator.
For the moment, the SC´s were helpless and it seemed that LBT should
be condemned. But before doing that, it was decided to prepare cross sections
of all test coupons.
Fig. 1. Correct weld cross section following the instructions
Some of the welders followed the instructions, others did not. They used
their ability to weld successfully without hot cracks. One conclusion on
this is: The influence of the welder and the influence of the welding process
are greatly significant [3]. All other factors commonly named „influencing
factors“ are less important. A flat weld pool produced by electrode
weaving and clever handling of heat during the welding process prevented
hot cracking.
Discussion and Evaluation of Some Extraordinary Cases of Hot Cracking 385
Fig. 2. Different number of top layer runs
Case 2 – Hot cracking occurs in a material which normally
is hot crack-proof. What is the reason?
Hot cracking occurred in the heat affected zone inside a welded pipe. It
was detected on occasion of a routine inspection in a nuclear power plant.
The austenitic parent metal X 6 CrNiTi 18 10, material number 1.4541
normally is weldable without problems in the heat affected zone. The
chemical analysis of the material was customary.
The calculated ferrite number FN was 4. The result of an MVTspecimen
taken from the cross section of the pipe was: The material is not
sensitive to hot cracking.
386 Testing and Standardization
Fig. 3. Cross section of a pipe; segregation zone at the inner side
But in a detailed metallographic examination, a high amount of segregation
was found on the inner side of the seamless manufactured pipe.
Fig. 4. Sampling plan
Discussion and Evaluation of Some Extraordinary Cases of Hot Cracking 387
Specimens from the inner side of the pipe were taken and were prepared
in order to be able to carry out the MVT-Test along a line of the inner pipe
surface (Fig. 4). Together with this pieces a special MVT-specimen was
prepared (Fig.5).
Fig. 5. Special MVT-specimen
Fig. 6. Cross section of an MVT-specimen;
hot cracking in the segregated inner side of the pipe in the heat affected zone
388 Testing and Standardization
The following result was obtained from the MVT-Test: The material is
susceptible to hot cracking in the heat affected zone. Metallographic examination
showed that primary austenitic solidification occured. And in
this area hot cracks started [4]. This is shown in Fig. 6 to Fig. 8.
Fig. 7. Detail from Fig. 6; hot cracking in the primary austenitic zone
Fig. 8. Detail from Fig. 6; primary ferritic solidification above, primary austenitic
solidification below
This damage at that time caused a lot of trouble in Germany. As a tentative
conclusion one cannot trust in normal material data. Sometimes it is
possible that other proportions exist in local areas. In this case the hot
cracks inside the pipes were the starting points of crack growth through the
wall thickness of the pipes.
Discussion and Evaluation of Some Extraordinary Cases of Hot Cracking 389
Case 3 – Is it possible to weld an extremely hot cracksensitive
material nearly without hot cracking?
Railway tracks are subjected to abrasion, especially on bracking tracks.
Therefore, surfacing by welding on top of the rails is necessary. The
welding is possible in some breaks of the traffic or at night. Early replacement
of a rail can thus be ensured. But the welding consumables used are
extremely hot crack-sensitive.
Such a huge number of cracks is unacceptable. They are starting points
of cracking across the rail. How is repair welding realized on the rail top
without or nearly without hot cracks?
Hot cracks only occur in the Brittleness Temperature Range (BRT). It is
necessary to have strains higher than the critical strain to cause hot cracks.
That means: If strains are avoided, no hot cracking will occur.
Fig. 9. Welding of rail material; MVT-specimen showing an extremely high
amount of hot cracking
In general, welding of rail material is a complex process, because of the
high carbon content of up to 0.7 %. Depending on the rail material, preheating
between 350 °C and 500 °C and postheating between 350 °C and
500 °C for 5 min to 10 min are necessary.
But, a rail is a simple workpiece. So a simple bending device was built
to initiate compression stresses in the top of the rail during welding. The
preheating and postheating procedure was regulated in such a way that the
rail foot was under tension and the rail top was under compression. Using
this method was very successful.
390 Testing and Standardization
Fig. 10. Experimental welding of hot crack-sensitive rail material
without hot cracking
The microcracking factor (similar to MSI, case 1) dropped down from
the maximum of 7.56⋅10–2 to 0. But one specimen showed a small amount
of cracking in the heat affected zone of 0.2 x 10–2 mm.
Now the question was: What was the reason for not being completely
successful? In the heat affected zone of the rail material, hot cracking occurs
depending on the consumables used. However, this is the next case.
Case 4 – What is the reason for hot cracking in the heat
affected zone of the base material?
Fig. 11 is about 30 years old. At that time, my comment was the following:
The grain boundaries of the heat affected zone of the base material were
remelted because of the heat of the weld metal. At the same time, the additional
thermal strains initiated hot cracking. After that, the molten
austenitic weld metal filled the hot cracks which is also called eutectic
healing.
Now two other examples:
Fig. 12 shows the nickel base alloy 718, NiCr19NbMo, material number
2.4668. It seems that the molten material of the weld pool flowed into the
opened crack in the heat affected zone of the parent metal. There was not
enough liquid weld metal for complete healing, so that a liquation crack
remained.
Similar to that figure is Fig. 13 which shows the copper nickel alloy
CuNi10Fe1Mn. The molten material of the weld pool filled up the crack in
the heat affected zone of the base material.
Discussion and Evaluation of Some Extraordinary Cases of Hot Cracking 391
Fig. 11. Heat affected zone hot cracking in a rail steel,
18/8/6 CrNiMn Type electrode
Fig. 12. Heat affected zone cracking in nickel base alloy 718
But there was a lack of material. The growing dendrites of both sides of
the crack could not come together and a liquation crack remained. But are
these explanations really true?
The next example may give the answer.
In a chemical plant, cracks occurred in the heat affected zone of welds
between an austenitic steel (material number 1.4541) and a zinc coated
mild steel. They were identified as liquid metal embrittlement (in German:
„Lotbrüchigkeit“).
392 Testing and Standardization
Fig. 13. Heat affected zone cracking in CuNi10Fe1Mn
It was intended to investigate this phenomenon using hot cracking tests
i.e. the MVT-Test. Specimens of mild steel (S 355) and stainless steel were
used which are normally unsensitive to hot cracking. These specimens
were galvanized with copper (thickness: 2 μm) and zinc (thickness: 5 μm),
respectively. Then the specimens were tested in the MTV-machine by producing
a melt run and 2 % strain. Extreme cracking was found in the heat
affected zone of both specimen types (Fig. 14).
On the specimen surface, the molten copper or zinc attacked the grain
boundaries of the parent metal. This happened only in the bending zone of
the specimens. In areas without bending, no cracking was observed.
Fig. 14. Surface of an MVT-specimen, parent metal S355, coated with copper,
hot cracking in the heat affected zone
(position in the cross section marked by the line)
Discussion and Evaluation of Some Extraordinary Cases of Hot Cracking 393
The specimens exhibit extreme heat affected zone cracking. Metallographic
examination revealed that copper is present on the surface and at
the end of the cracks. Zinc is more difficult to be seen on the cracks, because
most of the zinc was eliminated by the requisite etching of the
specimen.
Fig. 15. Cross section (position see Fig. 14) of a copper coated specimen,
copper situated at the end of the cracks
The testing time for an MVT-Test is about 0.3 s. During this short time
the process of cracking is completely finished. The behavior of molten
copper or zinc was extremely aggressive. The molten metal attacked the
grain boundaries of the normally hot crack-unsensitive parent material
S355 and X 10 Cr NiTi 18 9 (Fig. 16).
Fig. 16. Cross section of (a) a zinc-coated S355 MVT-specimen and (b) stainless
steel (X10 CrNiTi 19 9);
zinc situated on the cracks and at the crack ends
394 Testing and Standardization
Would it be possible that this process is similar to stress corrosion
cracking where a medium, i.e. an acid, attacks the grain boundaries and
initiates a crack under tension condition? Could it be that these are the
conditions for heat affected zone hot cracking?
This could be a future question for metal experts.
References
1. Wilken K (1995) Review of experiences with the Longitudinal Bend Test
(LBT) for microfissuring susceptibility of deposited weld metal. IIW/IIS –
1208-93 (Ex doc IX-1722-93). Welding in the World, vol 35, no 5: 307–326
2. Wilken K (1994) Scatter of LBT-results. IIW Doc IX-H-228-94, II-C 009-94
3. Wilken K, Heuser H, Hoffmann Th (1997) Heißrißverhalten von Chrom-
Nickel-Stählen und Nickelbasiswerkstoffen. DVS-Berichte, Band 183: 29–35
4. Wilken K, Bauer S (1994) Heißrißprüfung mit dem MVT-Test an Rohrabschnitten
des Grundwerkstoffes X6 CrNiTi 18 10 (Mat.-No. 1.4541) zur Ermittlung
des Chargeneinflusses. Unpublished report, BAM-6.43
5. Wilken K (1979) Untersuchungen zur Heißrißbildung neuartiger Schienenstähle.
Unpublished report BAM-6.43
6. Ehmig R, Wilken K (1994) Nutzungsmöglichkeiten von Farbniederschlagsätzungen
zur Erkennbarkeit und Bewertung von Heißrissen in Schweißverbindungen.
In: Sonderband „Praktische Metallografie“, Band 25, Carl Hanser
Verlag, pp 33–37
7. Pohle C,Wilken K, Schuster J (1996) Prüfen der Lotbruchanfälligkeit beim
Schweißen von Baustählen und austenitischen Chrom-Nickel-Stählen mit verzinkter
Oberfläche. Schweißen & Schneiden, Band 48, Heft 9: 684–689