Thomas Böllinghaus · Horst Herold ·

Carl E. Cross · John C. Lippold (Eds.)

Hot Cracking Phenomena

in Welds II

123
Prof. Dr.-Ing. Thomas Böllinghaus Prof. Dr. Carl E. Cross
Bundesanstalt für Bundesanstalt für
Materialforschung und–prüfung Materialforschung und–prüfung
Unter den Eichen 87 Unter den Eichen 87
12205 Berlin 12205 Berlin
Germany Germany
[email protected] [email protected]

Prof. Dr.-Ing. habil. Dr. E. h. Horst Herold Prof. John C. Lippold

Otto-von-Guericke Universität The Ohio State University
Postfach 4120 1248 Arthur E. Adams Drive
39016 Magdeburg Columbus, Ohio 43221-3560
Germany USA
[email protected] [email protected]

ISBN: 978-3-540-78627-6 e-ISBN: 978-3-540-78628-3

Library of Congress Control Number: 2005921916


c 2008 Springer-Verlag Berlin Heidelberg

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 any other way, 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. Violations are
liable to prosecution under the German Copyright Law.

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.

Cover design: deblik, Berlin

Printed on acid-free paper

9 8 7 6 5 4 3 2 1

springer.com
Preface

Failure of welded components can occur during service as well as during

fabrication. Most common, analyses of the resistance of welded
components against failure are targeted at crack avoidance. Such
evaluations are increasingly carried out by modern weldability studies, i.e.
considering interactions between the selected base and filler materials,
structural design and welding process. Such weldability investigations are
particularly targeted to prevent hot cracking, as one of the most common
cracking phenomena occurring during weld fabrication.
To provide an international information and discussion platform to
combat hot cracking, an international workshop on Hot Cracking
Phenomena in Welds has been created, based on an initiative of the
Institute for Materials and Joining Technology at the Otto-von-Guericke
University in Magdeburg and the Division V.5 – Safety of Joined
Components at the Federal Institute for Materials Research and Testing
(BAM) in Berlin, Germany. The first workshop was organized in Berlin
under the topics mechanisms and phenomena, metallurgy and materials,
modelling and simulations as well as testing and standardization. It
consisted of 20 individual contributions from eight countries, which were
compiled in a book that found a very ready market, not only in the welding
community. As a consequence of increasing interest, it has been decided to
establish the Workshop on Hot Cracking Phenomena in Welds as a regular
event every three years embedded in the International Institute of Welding
(IIW). Attached to the IIW Commission IX and II Spring intermediate
meetings, the second workshop was organized in March 2007.
The present book assembles 22 papers from 10 different countries,
which have again been published without any length restrictions on
individual contributions. The authors have attached great importance to
highlight the very recent developments and thus, the present book
particularly compiles the major worldwide hot cracking research
advancements of the latest years. In recent years, the research interest is
increasingly attracted to elucidation of the mechanisms, which differ
widely with the type of hot cracking. The book has thus been divided
correspondingly to the workshop sessions into chapters referring to the
VI Preface

various types of hot cracking, i.e. solidification cracking, liquation

cracking and ductility dip cracking.
It can only be emphasized that in combination with the previous book,
this edition represents a helpful tool for metallurgical, materials and
mechanical engineering students of higher semesters. For all welding
scientists and experts who want to get acquainted to the subject, the state
of knowledge can very easily be drawn from the various chapters instead
of gathering the necessary information piecewise from elsewhere.
The editors convey their sincere gratitude to all authors, session chairs
and participants of the second workshop for their engaged contributions
and for establishing a regular forum for exchanging the major research
advancements in hot cracking phenomena of welds. Special thanks go to
Ms. A. Cichon and Ms. I. Schülke for the tremendous work in organizing
the workshop.

Berlin, Magdeburg and Columbus Thomas Böllinghaus

May 2008 Horst Herold
Carl Cross
John Lippold
Contents

Part I: Solidification Cracking Theory

In Search of the Predicition of Hot Cracking in Aluminum

Alloys........................................................................................................... 3
L. Katgerman, D.G. Eskin

Application of the Rappaz-Drezet-Gremaud Hot Tearing

Criterion to Welding of Aluminum Alloys ............................................ 19
J.-M. Drezet, D. Allehaux

Weld Solidification Cracking: Critical Conditions for Crack

Initiation and Growth.............................................................................. 39
C.E. Cross, N. Coniglio

Consideration of the Welding Process as a Thermo-Physical

Mechanism to Control Cracking in Weldments ................................... 59
H. Herold, M. Streitenberger

Determination of Critical Strain Rate for Solidification

Cracking by Numerical Simulation........................................................ 77
M. Wolf, Th. Kannengießer, Th. Böllinghaus

Part II: Solidification Cracking of Ferrous and

Nickel-Base Alloys

Classification and Mechanisms of Cracking in Welding

High-Alloy Steels and Nickel Alloys in Brittle Temperature
Ranges....................................................................................................... 95
K.A. Yuschenko, V.S. Savchenko

Submerged Arc Welding – A Test for Centerline Cracking.............. 115

K. Håkansson
VIII Contents

Influence of Local Weld Deformation on the

Solidification Cracking Susceptibility of a Fully Austenitic
Stainless Steel ......................................................................................... 127
A. Kromm, Th. Kannengießer

Weld Solidification Cracking in Solid-Solution Strengthened

Ni-Base Filler Metals ............................................................................. 147
J.C. Lippold, J.W. Sowards, G.M. Murray, B.T. Alexandrov,
A.J. Ramirez

Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar

Metal Welds............................................................................................ 171
H. Hänninen, A. Brederholm, T. Saukkonen

Evaluation of Weld Solidification Cracking in Ni-Base

Superalloys Using the Cast Pin Tear Test ........................................... 193
B.T. Alexandrov, J.C. Lippold, N.E. Nissley

SAW Cold Wire Technology- Economic Alternative for

Joining Hot Crack Sensitive Nickel-Base Alloys................................. 215
U. Reisgen, U. Dilthey, I. Aretov

Part III: Solidification Cracking of Aluminium Alloys

Hot Tearing During Laser Butt Welding of 6xxx Aluminium

Alloys: Process Optimisation and 2D/3D Characterisation
of Hot Tears............................................................................................ 241
D. Fabrègue, A. Deschamps, M. Suéry, H. Proudhon

The Integral Approach- a Tailored Method to Optimize

Structural Behavior and Weldability................................................... 257
H. Gruss, A. Pshennikov, H. Herold

Weld Parameter and Minor Element Effects on Solidification

Crack Initiation in Aluminum .............................................................. 277
N. Coniglio, C.E. Cross

Using Simulation for Investigations of Hot Cracking

Phenomena in Resistance Spot Welding of 6xxx Aluminium
Alloys (AA6016 and AA6181) ............................................................... 311
A. Eder, S. Jaber, N. Jank
 Contents IX

Part IV: Liquation Cracking

Evaluating Hot Cracking Susceptibility of Ni-Base SAW

Consumables for Welding of 9% Ni Steel............................................ 329
L. Karlsson, E.-L. Bergquist, S. Rigdal, N. Thalberg

Assessment of HAZ Hot Cracking in a High Nitrogen

Stainless Steel ......................................................................................... 349
K. Stelling, M. Lammers, D. Meinel

Crack Appearance in Hot Rolled Billets.............................................. 371

S.T. Mandziej, G. Krallics

Part V: Ductility-Dip Cracking

Effect of Filler Metal La Additions on Micro-Cracking in

Multi-Pass Laser Overlay Weld Metal of Alloy 690 ........................... 389
K. Nishimoto, K. Saida, M. Sakamoto, W. Kono

Ductility-Dip Cracking in High Chromium, Ni-Base

Filler Metals ........................................................................................... 409
J.C. Lippold, N.E. Nissley

Thermodynamic and Kinetic Approach to Ductility-Dip

Cracking Resistance Improvement of Ni-base Alloy
ERNiCrFe-7: Effect of Ti and Nb Additions ...................................... 427
A.J. Ramirez, C.M. Garzón

Index ....................................................................................................... 455

 Part I
Solidification Cracking Theory
In Search of the Prediction of Hot Cracking
in Aluminium Alloys

L. Katgerman, D.G. Eskin

Delft University of Technology, Netherlands Institute for Metals Research,

Delft, The Netherlands

Abstract

Hot tearing remains a major problem of casting technology despite dec-

ades-long efforts to develop a working hot tearing criterion and to imple-
ment it into casting process computer simulation.
Existing models allow one to calculate the stress–strain situation in a
casting (ingot, billet) and to compare it with the chosen hot tearing crite-
rion. Two kinds of hot tearing criteria are available in literature: mechani-
cal and non-mechanical one. The mechanical criteria of hot tearing are de-
rived based on mechanical behaviour semi-solid, and the non-mechanical
one is based on other properties of semi-solid. In most successful cases,
the simulation shows a relative probability of hot tearing and the sensitiv-
ity of this probability to such process parameters as casting speed, casting
dimensions, and casting practice.
None of the existing criteria, however, can give the quantitative answer
on whether the hot crack will appear or not and what will be the extent of
hot cracking (position, length, shape). This chapter outlines the require-
ments for a modern hot tearing criterion as well as the future development
of hot tearing research in terms of mechanisms of hot crack nucleation and
propagation.

Introduction – Mechanisms of Hot Tearing

Various defects of as-cast product are still frequently encountered in cast-

ing practice. One of the main defects is hot tearing or hot cracking, or hot
shortness. Irrespective of the name, this phenomenon represents the forma-
tion of an irreversible failure (crack) in the still semi-solid casting.
4 L. Katgerman, D.G. Eskin

From many studies [1, 2, 3, 4, 5, 6, 7, 8] started already in the 1950’s,

and reviewed by Novikov [9] and Sigworth [10], it appears that hot tears
initiate above the solidus temperature and propagate in the interdendritic
liquid film. In the course of solidification, the liquid flow through the
mushy zone decreases until it becomes insufficient to fill initiated cavities
so that they can grow further. The fracture has a bumpy surface covered
with a smooth layer and sometimes with solid bridges that connect or have
connected both sides of the crack [7, 8, 11, 12, 13, 14, 15, 16].
Research studies show that hot tearing occurs in the late stages of solidi-
fication when the volume fraction of solid is above 85–95% and the solid
phase is organized in a continuous network of grains. It is also known that
fine grain structures and controlled casting (without large temperature and
stress gradients) help to avoid hot cracking.
During direct-chill (DC) casting of aluminium alloys, primary and sec-
ondary cooling cause strong thermal gradients in the billet/ingot, resulting
in uneven thermal contraction in different sections of the billet/ingot. As a
result, macroscopic stresses cause distortion of the billet/ingot shape (e.g.
butt curl and swell, rolling face pull-in) and/or may trigger hot tearing and
cold cracking in the weak sections. The terms “hot” or “cold” refer to the
temperature range where the cracking occurs – in the semi-solid mushy
zone or below the solidus, respectively. In DC casting, the name “mushy
zone” is frequently applied to the entire transition region between liquidus
and solidus, which is misleading, as the semi-solid mixture in the top part
of the transition region is actually a slurry. Only after the temperature has
dropped below the coherency temperature, a real mush is formed. On the
microscopic level solidification shrinkage and thermal contraction impose
strains and stresses on the solid network in the mushy zone. The deforma-
tion behaviour of the mush is very critical for the formation of hot tears.
The link between the appearance of hot tears and the mechanical properties
in the semi-solid state is obvious and has been explored for decades; see
for example reviews [9, 17].
Another important correlation – between the hot cracking susceptibility
and the composition of an alloy – has been established on many occasions.
A large freezing range of an alloy promotes hot tearing since such an
alloy spends a longer time in the vulnerable state in which thin liquid
films exist.
A lot of efforts have been devoted to the understanding of the hot tear-
ing phenomenon. Compilations of research in this field have been done by
Novikov [9], Sigworth [10], and Eskin et al. [17]. Several mechanisms of
hot tearing are already suggested in literature. Some of those are outlined
in Table 1.
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 5

Table 1. Summary of hot tearing mechanisms

Mechanism Suggested and developed by* Ref.**
Cause of hot tearing
Heine (1935); Pellini (1952);
Thermal contraction [18, 2, 19]
Dobatkin (1948)
Liquid film distribution Verö (1936) [20]
Liquid pressure drop Prokhorov (1962); Niyama (1977) [37, 39]
Vacancy supersaturation Fredriksson et al. (2005) [21]
Nucleation
Patterson et al. (1953, 1967); Niyama
[44, 22, 39,
Liquid film or pore as stress concentrator (1977); Rappaz et al. (1999); Braccini
38, 23, 48]
et al. (2000); Suyitno et al. (2002)
Oxide bi-film entrained in the mush Campbell (1991) [8]
Vacancy clusters at a grain boundary or
Fredriksson et al. (2005) [21]
solid/liquid interface
Propagation
Patterson (1953); Williams & Singer
Through liquid film by sliding (1960, 1966); Novikov & Novik [44, 24, 25]
(1963)
Pellini (1952); Patterson (1953);
By liquid film rupture [2, 44, 26, 27]
Saveiko (1961); Dickhaus (1994)
By liquid metal embrittlement Novikov (1966); Sigworth (1996) [9, 10]
Through liquid film or solid phase depend-
Guven & Hunt (1988) [28]
ing on the temperature range
Diffusion of vacancies from the solid to the
Fredriksson et al. (2005) [21]
crack
Conditions
Thermal strain cannot be accommodated by Pellini (1952); Prokhorov (1962);
[2, 37, 9, 29]
liquid flow and mush ductility Novikov (1966); Magnin et al. (1996)
Niyama (1977); Guven & Hunt
Pressure drop over the mush reaches a
(1988); Rappaz et al. (1999); Farup & [39, 28, 38, 41]
critical value for cavity nucleation
Mo (2000)
Strain rate reaches a critical value that Pellini (1952); Prokhorov (1962);
cannot be compensated by liquid feeding Rappaz et al. (1999); Braccini et al. [2, 37, 38, 23]
and mush ductility (2000)
Lees (1946); Langlais & Grizleski,
Thermal stress exceed rupture or local
(2000); Lahaie & Bauchard (2001); [30, 31, 32, 48]
critical stress
Suyitno et al. (2002)
Bochvar (1942); Lees (1946);
Stresses and insufficient feeding in the Pumphrey & Lyons (1948); Clyne & [33, 30, 34,
vulnerable temperature range Davies (1975); Feurer (1977); 35, 40, 36]
Katgerman (1982)
Thermal stress exceeds rupture stress of the
Saveiko (1961) [26]
liquid film
*
The list of the authors is by no means complete. The references have been chosen to represent the
development of ideas.
**
References are given in the same order as the authors in the second column.

One can see that, over the years, much more effort has been put on
the conditions required for hot tearing occurrence rather than on the
mechanisms of crack initiation and propagation. And when it comes to
the nucleation and propagation of hot tears, an educated guess
frequently replaces experimental proof.
6 L. Katgerman, D.G. Eskin

Over the years, different macroscopic parameters, such as stress and

strain, were considered as critical for the development of hot tearing. To-
day, the strain rate is believed to be the most important factor and some
modern models are based on it. The physical explanation of this approach
is that semi-solid material during solidification can accommodate the im-
posed thermal strain by plastic deformation, diffusion-aided creep, struc-
ture re-arrangement, and filling of the gaps and pores with the liquid. All
these processes require some time, and the lack of time will result in frac-
ture. Therefore, there exists the maximum strain rate that the semi-solid
material can endure without fracture during solidification. Prokhorov [37]
was the first to suggest a criterion based on this approach. More recently, an
elaborate, strain-rate based hot tearing criterion was proposed by Rappaz
et al. [38].
On the microscopic level, the important factor is believed to be the feed-
ing of the solid phase with the liquid. Within this approach, the hot tear
will not occur as long as there is no lack of feeding during solidification.
Niyama [39] and Feurer [40] use hindered feeding as a base for their po-
rosity and hot tearing criteria. The feeding depends on the permeability of
the mush, which is largely determined by the structure. Later, a two-phase
model of the semi-solid dendritic network [41], which focuses on the pres-
sure depression in the mushy zone, has been suggested to describe the hot
tear formation. This approach treats the semi-solid material as solid and
liquid phases with different levels of stress and strain. The pressure drop of
the liquid phase in the mush is considered as a cause of a hot tear. An ex-
tension of the two-phase model that includes plasticity of the porous net-
work is also reported [42]. Logically, bridging and grain coalescence,
which determine the transfer of stress and limit the permeability of the
mush, are the other important microscopic factors for the development of
hot tearing [43].
Yet the hot tearing theories that operate with macroscopic mechanical
behaviour and microscopic phenomena like feeding and porosity formation
do not take into account the mechanism of crack nucleation and propaga-
tion and, in this, are intrinsically weak. The other approach to the descrip-
tion of the hot tearing phenomenon is the application of fracture mechanics
that describes initiation and propagation of cracks. The liquid film sur-
rounding the grain at late stages of solidification is considered as a stress
concentrator of the semi-solid body [44, 45]. In this theory, a liquid-filled
cavity acts as a crack initiator. The propagation of the crack is determined
by the critical stress [45]. The critical stress can be estimated using the
Griffith energy balance approach modified by taking into account the
plasticity [46, 47]. Recently, a formulation of hot tearing as a phenomenon
related to micro-porosity has been proposed [48]. In this approach the
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 7

porosity and the hot tearing are considered as sequential events. As a

result, there is a possibility to predict simultaneously the occurrence of
micro-porosity and hot tearing. The model uses the feeding difficulties at
the last stage of solidification as a starting point of cavity nucleation. The
nucleus then grows and becomes at the end of solidification either a micro-
pore or a hot tear as determined by the Griffith model for brittle crack
growth. The fracture-mechanical concept of hot tearing can be enriched
with the application of a liquid-metal-embrittlement mechanism [10].

Fig. 1. Different length scales of equiaxed dendritic solidification along with

suggested hot tearing mechanisms (the initial sketch is adopted from [49])

It is obvious that actual hot tearing mechanism includes phenomena

occurring on two scales: microscopic (crack nucleation and propagation,
stress concentration, structure coherency, wet grain boundaries, feeding)
and macroscopic (stress, strain, or strain rate imposed on the structure).
Figure 1 illustrates these scales during equiaxed dendritic solidification.
8 L. Katgerman, D.G. Eskin

Current Hot Tearing Criteria and their Applicability

to DC Casting

The existing hot tearing criteria as reviewed elsewhere [17, 50] can be
conditionally divided into the two categories: non-mechanical and me-
chanical. The former type of criteria deals with vulnerable temperature
range, phase diagram, and process parameters; and is represented by the
criteria of Clyne & Davies [6], Feurer [40], and Katgerman [36]. The latter
type of criteria involves critical stress [9, 27, 31, 32, 45], critical strain [9,
29, 51], or critical strain rate [23, 37, 38, 39, 49, 52].
Different casting processes impose specific requirements on the
application of hot tearing criteria. That is why some criteria are working
better for shape casting whereas others are more suitable for direct-chill
casting. There is no doubt that a good hot tearing criterion for DC casting
should correctly respond to the casting parameters, e.g. casting speed,
ramping rate, and alloy composition; and predict the vulnerable section of
a billet or an ingot, e.g. the centre of a round billet. Most of the existing
criteria have been tested for the composition sensitivity by calculating the
hot tearing susceptibility of several binary alloys with an attempt to
reproduce the so-called lambda-curve showing the maximum susceptibility
at a certain composition. And most of the existing criteria can do this
successfully. However, dynamic parameters such as casting speed and
strain rate are usually kept constant upon such testing. Therefore, the
compositional sensitivity of a hot tearing criterion does not assure its
successful application to a particular casting technology.
The basic phenomena that lead to hot cracking are well established and
understood, but a generic criterion that will predict hot cracking under
varying process conditions is still not available. Although the earlier
simple criteria based on the thermal history of the casting have been
extended and improved to include shrinkage and deformation, they are still
unable to give reliable predictions under all process conditions. Most of
the existing hot tearing criteria do not incorporate the nucleation and
propagation of a hot tear, focusing more on the macroscopic and
microscopic conditions that may result in rupture.
The ultimate hot cracking criterion needs to combine aspects of thermal
history, shrinkage and porosity formation, and constitutive behaviour in
combination with the evolution of the semi-solid microstructure. Current
research efforts are aimed, in particular, at the quantitative description of
structure evolution and its correlation to cracking.
Recently, several mechanical and non-mechanical hot tearing criteria
were evaluated by implementing them into a thermo-mechanical model of
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 9

DC casting [50]. The criteria show different results in predicting the hot
tearing susceptibility as shown in Table 2.

Table 2. Sensitivity of hot tearing criteria to the casting parameters and practice
upon direct-chill casting [50, 53]
Criterion Hot tearing More hot Ramping casting Correlation with
increases with tears in the speed during actual cracking
casting speed billet centre start-up of the observed in
casting reduces practice
hot tearing
Clyne & Davies [6] No No No N/A
Katgerman [36] Yes Yes No N/A
Feurer [40] Yes Yes No N/A
Novikov [9] No No No N/A
Magnin et al. [29] Yes No No No
Prokhorov [37] Yes Yes No No
Rappaz et al. [38] Yes Yes Yes No
Braccini et al. [23] Yes Yes No N/A
Suyitno [48] Yes Yes Yes Yes

The criteria of Clyne & Davies and Novikov give results that are
inconsistent with casting practice, not showing any sensitivity to the casting
speed and position within the billet volume. It is noteworthy to mention that
these criteria are very successful in predicting the compositional dependence
of hot tearing and are frequently used for shape casting. The criteria of
Feurer, Katgerman, Magnin et al., Prokhorov, Rappaz et al., and Braccini et
al. respond correctly to the casting parameters, demonstrating that the
increasing casting speed results in an increasing hot tearing susceptibility in
the centre of billet, which is in accordance with casting practice. However,
most of the tested criteria, except those by Rappaz et al. [38] and Suyitno
[48, 53], are not sensitive to the ramping of casting speed during the start-up
phase of casting (which is a usual practice to prevent hot cracking). And,
when confronted with casting practice, the criteria of Prokhorov, Magnin et
al., and Rappaz et al. predict the occurrence of hot cracks whereas no cracks
have been found in billets cast under given conditions. Only the criterion of
Suyitno adequately responds to all tested parameters, i.e. casting speed,
ramping rate, grain size, position in a billet, and casting practice. The
sensitivity of this criterion is, however, a function of correctly chosen values
of properties such as Young’s modulus of the mush, surface tension between
liquid and solid, and permeability of the mush. These parameters are
scarcely available and needed to be determined experimentally, while the
existing experimental techniques are not reliable.
10 L. Katgerman, D.G. Eskin

Outline of Hot Tearing Mechanisms as a Base for a New

Hot Tearing Criterion

Up to now, hot tearing is considered as a phenomenon linked to casting

and welding processes, and the existing theories, models and criteria are
biased by their applicability to solidification. Hot tearing is, however, just
another example of material failure. Therefore, it should be treated ade-
quately, using the well-developed apparatus of fracture mechanics. The
challenge nowadays lies not in the adequate description of macroscopic
and microscopic stress–strain situations and their correspondence to the
parameters and properties of the mushy zone, but rather in finding real fac-
tors causing the nucleation and propagation of a hot crack. In fact, some
existing theories and models of hot tearing partially describe these factors
with the crack initiator presented as a cavity filled with liquid or a pore
[38, 48], or an oxide bi-film [8] and with the crack propagation path
through the liquid film covering grain boundaries [10]. What is lacking is
the completeness of the model. A comprehensive model and a based-on-it
criterion should include nucleation and propagation of hot tears and con-
nect these processes to the microstructure evolution during solidification of
the semi-solid material; to the macroscopic and microscopic thermo-
mechanical situation in the mushy zone; and to the mechanical (or frac-
ture-mechanical) properties of the mushy zone. The last two components
are well covered by a large body of publications, though many mechanical
properties are still needed to be determined and the fracture mechanics
potential has not been fully exploited. The correspondence between the
structure evolution during solidification and the crack nucleation ad propa-
gation is studied in much less detail. Let us consider the possible mecha-
nisms of crack nucleation and propagation.
The mechanism of hot tear formation and propagation can be elucidated
from observations of fractures. Unfortunately, the reports on hot crack
fractures in metallic materials are rare. Most of such reports describe the
cracking of semi-solid alloys at relatively large fractions of liquid, when
grain boundaries are completely covered with liquid. In this case the
mechanism of crack propagation – through liquid film by grain separation
– is obvious. An example of such alloys is the classic Al–4% Cu alloy.
However, alloys with high fractions of liquid in the vulnerable solidifica-
tion range are in practice not susceptible to hot tearing [17, 54]. Cavities
and gaps between grains that may form in the mushy zone of such alloys
due to thermal contraction or external tension are easily filled with liquid
due to the adequate permeability of the mushy zone and sufficient amount
of available liquid that is represented in the final structure by non-equilibrium
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 11

eutectics [54]. Much more important is the mechanism of crack formation

and propagation in alloys containing little solute that are most susceptible
to hot tearing. But the information on semi-solid fracture in such alloys is
only starting to emerge. If one wanted to summarize the findings available
to date, it would appear that bridging of grain boundaries is an essential
feature of the fracture surface. Moreover, the closer the semi-solid material
gets to the temperature range of its maximum vulnerability to hot cracking,
i.e. 90–95% solid, the greater fraction of grain boundaries is connected to
each other, or coalesced [55]. In this case, there is no chance for the crack
propagation through a continuous liquid film as such a film does not exist.
Our recent observations [56] show that a hot tear apparently propagates
through the liquid film in more alloyed materials and through solid bridges
in less alloyed materials as illustrated in Fig. 2. Although in both cases the
fracture surface appears to be brittle, one can suggest that different crack-
ing mechanisms are acting. In real castings, cracks appear above the
solidus as well as below the solidus. In the latter case, these cracks are
called “cold”. There are reports, however, that these cracks, appearing at
high temperatures in the fully solid material, can originate in areas of local
remelting caused by intense local deformation [57]. Thus formed liquid
also assists in crack propagation. These observations make such “cold”
cracks in fact similar to hot tears.
The nucleation of hot cracks is an almost unexplored phenomenon. It is
obvious that under any stress-strain conditions, there should be a certain,
critical size of a defect (flaw, nucleus) that would enable crack growth.
The problem of the crack initiation is today solved by an educated guess,
as there are almost no experimental observations on the crack nucleation
upon natural hot tearing. Usually, the development of a hot crack is studied
on samples with a notch, hence – with the artificial crack initiator [43, 59].
Based on these observations and “post-mortem” examination of hot tear
surfaces, the following crack nuclei have been suggested: (1) liquid film or
liquid pool; (2) pore or series of pores; (3) grain boundary located in the
place of stress concentration; (4) inclusions that can be easily separated
from the surrounding liquid or solid phase, e.g. intermetallic particle or ox-
ide film. In should be noted that pores that are frequently cited as potential
hot tear nuclei can originate from gas precipitation [8], solidification
shrinkage [8], or vacancy supersaturation [21]. Figure 3 summarizes some
of the possible crack initiators.
12 L. Katgerman, D.G. Eskin

a b

Fractured
bridges

Fig. 2. Fracture surfaces of hot tears in 200-mm round billet produced by DC

casting at a casting speed of 200 mm/min: (a) Al–1% Cu and (b) Al–3% Cu

Fig. 3. Schematic illustration of possible hot crack initiators and some crack
propagation mechanisms (the initial sketch is adopted from [58])

Despite a large body of literature on hot tearing, only few efforts have
been spent on the mechanism of hot tear propagation. One can mention the
application of the Griffith criterion for brittle fracture [48, 53] and va-
cancy-diffusion-controlled growth [21]. The apparatus for the description
of crack propagation is well developed within fracture mechanics for
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 13

various situations, including those resembling hot cracking, i.e. liquid film
rupture, pore coalescence, high-temperature creep, and liquid-assisted
fracture [58]. It is clear that the propagation of the hot crack in the potential
presence of liquid (above or below the solidus) should involve the follow-
ing aspects: liquid feeding (involves permeability and, inevitably structure
evolution), pore coalescence, stress transfer by solid bridges, plastic de-
formation and creep of solid bridges in the absence of liquid, and brittle
fracture of solid bridges in the presence of liquid. It is also obvious that the
hot crack, like any other crack, can develop catastrophically (which is usu-
ally assumed), have sustained growth, or stop. Liquid feeding plays a dual
role. Firstly, adequate feeding of the shrinking material with liquid does
not eliminate the causes of hot tearing but rather “patches” the conse-
quences, which is reflected in the term “crack healing”. One can say that a
semi-solid alloy containing enough liquid at the last stage of solidification,
having a microstructure that enables adequate permeability of the mush,
and subjected to tensile stresses is a self-healing material. On the other
hand, the development of solid bridges between grains at high solid frac-
tions in the absence of any liquid would build up enough strength and duc-
tility to prevent any brittle rupture. Hence, the liquid feeding should be just
enough to supply some liquid to solid bridges that enables their liquid em-
brittlement, otherwise only the mechanisms of ductile fracture, e.g. high-
temperature creep and pore coalescence will be active [8]. The evidence of
plastic deformation during hot tearing has been observed in direct observa-
tions [59] and upon examination of fracture surfaces [9].
We can suggest the following approach to treating the nucleation and
propagation of hot cracks. Several distinct mechanisms are operational in
different temperature or compositional ranges or, in other words, at differ-
ent fractions of solid. Here we will consider only the decreasing temperature
as the factor affecting the solid fraction, though the composition is obvi-
ously the other factor that acts in a similar manner. It is important to note
that the microstructure, e.g. grain size and morphology, affects the critical
temperatures and fractions of solid. At relatively low fractions of solid
below the coherency temperature, the permeability of the mushy zone
enables adequate feeding of the solidification shrinkage and most of the
precipitating gas bubbles can float to the liquid part of the sample. In this
case, the tensile stresses caused by non-uniform thermal contraction of the
coherent dendrites may cause the formation of cavities and gaps that are
immediately filled with liquid, or “healed”. On further decreasing tempera-
ture, the tensile stress builds up to such an extend that the liquid film sepa-
rating grains ruptures and the formed gap cannot be filled with liquid due
to the already insufficient permeability of the mush and to the increasing
capillary pressures required to fill ever narrowing openings between grains.
14 L. Katgerman, D.G. Eskin

On further cooling, the bridging between solid grains replaces former entan-
glement and touching of grains, and the stress can now be transmitted over
larger distances through the rigid solid skeleton, hence the semi-solid body
acquires macroscopic strength. This critical temperature is called the
“rigidity temperature” and can be determined experimentally [9, 60]. Note
that the rigidity temperature strongly depends on the structure [60]. The
permeability at such fractions of solid is not enough to completely com-
pensate the solidification shrinkage, gas cannot escape from narrow pas-
sages and, as a result, porosity is formed at available interfaces, mainly on
grain boundaries. The non-uniform thermal stress causes rather significant
strains in the semi-solid material that can be or cannot be sustained by the
solid bridges. Even limited access of the liquid to the solid bridge will re-
sult in its brittle fracture by the mechanism of liquid-metal embrittlement.
This is partially reflected in the proposal of van Haaften et al. [61] to use
the fraction of grain boundaries covered with liquid rather than the fraction
of liquid in the constitutive equation for the mechanical behaviour or semi-
solid aluminium alloys. The same mechanism may act at subsolidus tem-
peratures when some amount of non-equilibrium liquid is present at grain
boundaries or other stress concentrators because of non-equilibrium char-
acter of solidification [8] or local remelting [57]. There is also a possibility
that the semi-solid material fails macroscopically in a brittle manner (be-
cause of film rupture and liquid-metal embrittlement) with ductile rupture
of some solid bridges on the microscopic level [59]. And finally, the frac-
tion of bridged grain boundaries becomes so overwhelmingly large and the
remaining liquid is so scattered in the solid network that the semi-solid
material behaves like completely solid and fails in a ductile manner by
ductile pore coalescence and high-temperature creep. The crack initiator in
all these cases could be represented by pore, liquid pool or film, interface
with an intermetallic particle, or non-metallic inclusion. The outline of
these mechanisms is given in Table 3. Figures 1 and 3 illustrate the corre-
lation between these mechanisms and the development of the structure
during solidification.
A criterion that can predict not the probability but the actual occurrence
and extent of hot tearing should be based on the application of multi-phase
mechanics and fracture mechanics to the failure of semi-solid materials,
which is today limited by the lack of knowledge about the actual nuclea-
tion and propagation mechanisms. The mechanisms outlined in Table 3 are
based on the common sense and interpretation of very few experimental
observations. What is needed is a thorough and systematic study of frac-
tures occurring in solidifying materials with the aim to single out the na-
ture and the critical dimensions of defects or structure features that can
cause the nucleation of hot cracks. It is also necessary, in our opinion, to
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 15

acknowledge that different mechanisms of crack propagation are possible

at different fractions of solid. Therefore, different models could be applied
to the development of hot tears in ingots, billets and castings in depend-
ence on the alloy composition, structure, and the level of stresses that are
present. For example, coarse-grained material with a large solidification
range and high coherency temperature is likely to fail due to liquid film
rupture. In the case of alloys with a considerable amount of eutectics, e.g.
foundry alloys of the Al–Si system, the healing of cracks is most probable.
In contrast, a fine-grained alloy that develops coherency late in solidifica-
tion and does not contain much of eutectics, e.g. a commercial wrought al-
loy, will undergo complex failure involving liquid-metal embrittlement
and plastic deformation of solid bridges with a resultant mixed brit-
tle/ductile fracture.

Table 3. Possible mechanisms acting within “hot tearing” phenomenon

Temperature Nucleation Propagation Fracture mode
range/fraction of solid of crack* of crack
Between coherency and Grain boundary a. Liquid film rupture a. Brittle, intergranular
rigidity temperatures covered with liquid; b. Filled gap b. Healed crack
50–80% solid shrinkage or gas pore.
Below rigidity Pore, surface of a. Liquid film rupture; a. Brittle, intergranular
temperature particle or inclusion, liquid metal b. Plastic deformation of
80–99% solid liquid film or pool, embrittlement of solid bridges possible
vacancy clusters bridges
b. Plastic deformation
of bridges
Close to the solidus Pore, surface of a. Liquid metal a. Brittle, transgranular
98–100% solid particle or inclusion, embrittlement propagation is possible
segregates at grain b. Plastic deformation b. Macroscopically brittle
boundary, liquid at of bridges, creep or ductile, transgranular
stress concentration propagation is possible
point, vacancy
clusters
*
The crack initiator should be located in the place of stress concentration.

Concluding Remarks

The existing hot tearing criteria based on different principles have limited
applicability to commercial casting processes, e.g. to direct-chill casting,
due to their probabilistic character. The best of the available criteria can
successfully predict the probability of hot tearing in its dependence on
some casting parameters but fail to quantify the event, in other words can-
not forecast the actual occurrence of hot cracks in ingots and billets. There
are two main challenges in this endeavour. Firstly, we are lacking the
16 L. Katgerman, D.G. Eskin

knowledge of the actual causes of crack nucleation. That is to say we do

not know exactly what defects or structure defects can act as crack initia-
tors under particular temperature–stress conditions. Secondly, there is a
possibility that different mechanisms of crack propagation and final failure
are acting in dependence on the fraction of solid at which the fracture oc-
curs and on the alloy structure. The application of multi-phase mechanics
and, eventually, fracture mechanics to the phenomenon of hot cracking
looks quite promising. The quest for a new hot tearing criterion should
focus on these two research areas

Acknowledgements

This chapter is written within the framework of the research program of

the Netherlands Institute for Metals Research (www.nimr.nl), projects
MP4.97014 and MC4.02134.

References

1. H.F. Bishop, C.G. Ackerlind, and W.S. Pellini: Trans. Am. Foundrymen’s
Soc., 1952, vol. 60, pp. 818–833.
2. W.S. Pellini: Foundry, 1952, vol. 80, pp. 124–133, 192–199.
3. J.C. Borland: Brit. Weld. J., 1960, vol. 7, pp. 508–512.
4. S.A. Metz and M.C. Flemings: Trans. Am. Foundrymen’s Soc., 1970, vol. 78,
pp. 453–460.
5. U. Feurer: Giessereiforschung, 1976, vol. 28 (2), pp. 75–80.
6. T.W. Clyne and G.J. Davies: in Solidification and Casting of Metals, Metals
Society, London, United Kingdom, 1979, pp. 275–278.
7. B. Rogberg: Scand. J. Met., 1983, vol. 12, pp. 51–66.
8. J. Campbell: Castings, Butterworth-Heinemann, Oxford, United Kingdom, 1st
edn. 1991; 2nd edn., 2003.
9. I.I. Novikov: Goryachelomkost tsvetnykh metallov i splavov (Hot shortness of
non-ferrous metals and alloys), Nauka, Moscow, USSR, 1966.
10. G.K. Sigworth: Trans. Am. Foundrymen’s Soc., 1996, vol. 104, pp. 1053–1062.
11. J.A. Spittle and A.A. Cushway: Met. Technol., 1983, vol. 10, pp. 6–13.
12. L. Ohm and S. Engler: Giessereiforschung, 1990, vol. 42 (4), pp. 149–162.
13. M.L. Nedreberg: PhD Thesis, University of Oslo, Oslo, Norway, 1991.
14. J.A. Spittle, S.G.R. Brown, J.D. James, and R.W. Evans: in Proc 7th Int.
Symp. on Physical Simulation of Casting, Hot Rolling and Welding, National
Research Institute for Metals, Tsukuba, Japan, 1997, pp. 81–91.
 In Search of the Prediction of Hot Cracking in Aluminium Alloys 17

15 W.-M. van Haaften, W.H. Kool, and L. Katgerman: in Continuous Casting,

K. Ehrke and W. Schneider, ed., Wiley-VCH, Weinheim, Germany, 2000,
pp. 239–244.
16. I. Farup, J.-M. Drezet, and M. Rappaz: Acta Mater., 2001, vol. 49, pp.
1261–1269.
17. D.G. Eskin, Suyitno, and L. Katgerman: Progr. Mater. Sci., 2004, vol. 49,
629–711.
18. R.W. Heine and P.C. Rosenthal, Principles of Metal Casting, McGraw-Hill,
New York, USA, 1955.
19. V.I. Dobatkin: Nepreryvnoe lit’e I liteinye svoistva splavov (Direct-Chill
Casting and Casting Properties of Alloys), Oborongiz, Moscow, Russia, 1948.
20. J. Verö: Met. Industry, 1936, vol. 48, pp. 431–494.
21. H. Fredriksson, M. Haddad-Sabzevar, K. Hansson, and J. Kron: Mater. Sci.
Technol., 2005, vol. 21, pp. 521–529.
22. W. Patterson, S. Engler, and R. Kupfer: Giessereiforschung, 1967, vol. 19 (3),
pp. 151–160.
23. M. Braccini, C.L. Martin, M. Suéry, and Y. Bréchet: in Modelling of Casting,
Welding and Advanced Solidification processes IX, P.R. Sahm, P.N. Hansen,
and J.G. Conley, ed., Shaker Verlag, Aachen, Germany, 2000, pp. 18–24.
24. J.A. Williams and A.R.E. Singer: Austral. Inst. Met., 1966, vol. 11, pp. 2–9.
25. I.I. Novikov and F.S. Novik: Dokl. Akad. Nauk SSSR, Ser. Fiz., 1963, vol. 7,
pp. 1153.
26. V.N. Saveiko: Russ. Castings Production, 1961, (11), pp. 453–456.
27. C.H. Dickhaus, L. Ohm, and S. Engler: Trans. Am. Foundrymen’s Soc., 1994,
vol. 101, pp. 677–684.
28. Y.F. Guven and J.D. Hunt: Cast Met., 1988, vol. 1, pp. 104–111.
29. B. Magnin, L. Maenner, L. Katgerman, and S. Engler: Mater. Sci. Forum,
1996, vol. 217–222, pp. 1209–1214.
30. D.C.J. Lees: J. Inst. Met., 1946, vol. 72, pp. 343–364.
31. J. Langlais and J.E. Gruzleski: Mater. Sci. Forum, 2000, vol. 167, pp. 31–337.
32. D.J. Lahaie and M. Bouchard: Metall. Mater. Tras. B., 2001, vol. 32B, pp.
697–705.
33. A.A. Bochvar: Izv. Akad. Nauk SSSR, Otdel. Tekhn. Nauk, 1942, (9), p. 31.
34. W.I. Pumphrey and J.V. Lyons: J. Inst. Met., 1948, vol. 74, pp. 439–455.
35. T.W. Clyne and G.J. Davies: British Foundrymen, 1975, vol. 68, pp. 238–244.
36. L. Katgerman: J. Met., 1982, vol. 34 (2), pp. 46–49.
37. N.N. Prokhorov: Russ. Castings Production, 1962, (2), pp. 172–175.
38. M. Rappaz, J.-M. Drezet, and M. Gremaud: Metall. Mater. Trans. A, 1999,
vol. 30A, pp. 449–455.
39. E. Niyama: in Japan–US Joint Seminar on Solidification of Metals and Alloys,
Japan Society for Promotion of Science, Tokyo, Japan, 1977, pp. 271–282.
40. U. Feurer: in Quality Control of Engineering Alloys and the Role of Metals
Science, H. Nieswaag and J.W. Schut, ed., Delft University of Technology,
Delft, The Netherlands, 1977, pp. 131–145.
41. I. Farup and A. Mo: Metall. Mater. Trans. A, 2000, vol. 31A, pp. 1461–1472.
18 L. Katgerman, D.G. Eskin

42. M. M’Hamdi and A. Mo: in Light Metals 2002, W. Schneider, ed., The Minerals,
Metals, and Materials Society, Warrendale, USA, 2002, pp. 709–716.
43. M. Rappaz, P.-D. Grasso, V. Mathier, J.-M. Drezet, and A. Jacot: in Solidification
of Aluminum Alloys, M.G. Chu, D.A. Granger, and Q. Han, ed., The Minerals,
Metals, and Materials Society, Warrendale, USA, 2005, pp. 179–190.
44. K. Patterson: Giesserei, 1953, vol. 40 (12), pp. 597–605.
45. J.A. Williams and A.R.E. Singer: J. Inst. Met., 1968, vol. 96. pp. 5–12.
46. J.J. Gilman: in Proc of 2nd Symp. on Naval Structural Mechanics, E.H. Lee and
P.S. Symonds, ed., Pergamon, London, United Kingdom, 1960, pp. 43–99.
47. E. Orowan: in Fatigue and Fracture of Metals, W.M. Murray, ed., Massachu-
setts Institute of Technology, MIT Press, Cambridge, USA, 1950, p. 139.
48. Suyitno, W.H. Kool, and L. Katgerman: Mater. Sci. Forum, 2002, vol. 396–402,
pp. 179–184.
49. J.F. Grandfield, D.J. Cameron, and J.A. Taylor: in Light Metals 2001, J.L. Anjier,
ed., The Minerals, Metals, and Materials Society, Warrendale, USA, 2001, pp.
895–901.
50. Suyitno, W.H. Kool, and L. Katgerman: Metall. Mater. Trans. A, 2005, vol.
36A, pp. 1537–1546.
51. L. Zhao, Baoyin, N. Wang, V. Sahajwalla, and R.D. Pehlke: Int. J. Cast Metals
Res., 2000, vol. 13 (3), pp. 167–174.
52. J.-M. Drezet and M. Rappaz: in Light Metals 2001, J.L. Anjier, ed., The Minerals,
Metals, and Materials Society, Warrendale, USA, 2001, pp. 887–893.
53. Suyitno: PhD Thesis, Delft University of Technology, Delft, The Netherlands,
2005.
54. Suyitno, D.G. Eskin, V.I. Savran and L. Katgerman: Metall. Mater. Trans. A,
2004, vol. 35A, pp. 3551–3561.
55. Y. Ju. PhD Thesis, Norwegian University of Science and Technology,
Trondheim, Norway, 2004.
56. D. Eskin, Suyitno, and L. Katgerman: in Aluminium Cast House Technology
2005, Proc. 9th Australasian Conference, J. Taylor, I. Bainbridge, and
J. Granfield, ed., Collingwood: CSIRO Publishing, 2005, pp. 77–84.
57. A. Deschamps, S. Péron, Y. Bréchet, J.-C. Ehrström, and L. Poizat: Mater.
Sci. Technol., 2002, vol. 18, pp. 1085–1091.
58. M. Janssen, J. Zuidema, and R.J.H. Wanhill: Fracture Mechanics. Delft
University Press Blue Print, Delft, The Netherlands, 2002.
59. W.-M. van Haaften, W.H. Kool, and L. Katgerman. J. Mater. Eng. Perform.,
2002, vol. 11, pp. 537–543.
60. D.G. Eskin, Suyitno, J.F. Mooney, and L. Katgerman: Metall. Mater. Trans.
A, 2004, vol. 35A, pp. 1325–1335.
61. W.-M. van Haaften, W.H. Kool, and L. Katgerman: Mater. Sci. Eng. A, 2002,
vol. 336, pp. 1–6.
Application of the Rappaz-Drezet-Gremaud Hot
Tearing Criterion to Welding of Aluminium Alloys

J.-M. Drezet1, D. Allehaux2

1
Federal Polytechnic School of Lausanne, Materials Simulation Laboratory,
Lausanne, Switzerland
2
EADS Corporate Research Centre-Paris, Suresnes, France

Abstract

Hot tearing, a severe defect occurring during solidification is the conjunc-

tion of tensile stresses which are transmitted to the mushy zone by the co-
herent solid underneath and of insufficient liquid feeding to compensate
for the volumetric change. The RDG (Rappaz Drezet Gremaud) criterion
for the appearance of hot tears in metallic alloys [1] is based upon a mass
balance performed over the liquid and solid phases and accounts for the
tensile deformation of the solid skeleton perpendicular to the growing den-
drites and for the induced interdendritic liquid feeding. When tackling the
problem of hot tearing in welding of aluminium alloys, the RDG criterion
can be used at three levels of increasing complexity by:
− ranking the alloy with regards to their sensitivity to hot cracking
− studying the risk of hot tearing in the process using only the thermal field
(thermal criterion),
− and studying the influence of the mechanical behaviour of the mushy alloy
on the risk of hot cracking (thermo-mechanical criterion).
Each level is illustrated by an example dealing with laser beam welding.
Nevertheless, one of the critical issues in the RDG approach is the defini-
tion of a coherency point which, in low-concentration alloys, corresponds
to the bridging or coalescence of the primary phase. To tackle this aspect, a
2D granular model is presented together with preliminary results.
20 J.-M. Drezet, D. Allehaux

Introduction

Hot tearing together with microporosity have always been recognized as

major defects in castings of aluminium alloys, the first one being typical of
semi-continuous casting whereas the second one is dominant in shape
castings. These two defects are also present in welds. As mentioned by
Campbell [2], they are interconnected, as both result from a lack of feeding
and nucleation of a pore/void in the remaining liquid. However, if porosity
is associated with solidification shrinkage and can occur within the grains
or at grain boundaries, hot tearing is clearly linked with tensile stresses in
the solid and is confined to grain boundaries.
An excellent review of all the hot tearing models has been published by
Eskin et al. [3] in 2004. Recent developments made both at the macro-
scopic and microscopic levels, have pointed out the difficulties in connect-
ing these two scales. Macroscopic approaches are based nowadays on two-
phase approaches [4, 5, 6]: they are the natural extension of the so-called
RDG criterion [1], but using a more rigorous rheological approach for the
compressible mushy solid phase. The main advantage of such methods is
that they can take into account the whole scale of the parts to be welded,
but their weakness is that averages can hardly account for the localization
of strains and feeding at grain boundaries. On the microscopic scale, coa-
lescence and percolation of grains can be accounted for at the scale of a
small volume element of the mushy zone (typically a few cubic centime-
tres) using granular approaches [7, 8, 9]. Although such techniques cannot
yet consider a whole solidification process, they provide a detailed and in-
teresting view of the phenomena occurring during hot tearing, in particular
localization of strains and feeding, gradual transition from a continuous
liquid film network to a fully coherent solid, etc. These methods are briefly
presented in the last section.

The RDG Hot Tearing Criterion

In the RDG criterion [1], a deformation perpendicular to the thermal gradi-

ent is applied to the solid phase, regardless whether it corresponds to co-
lumnar or equiaxed dendrites (see Fig. 1). The component of the strains
parallel to the thermal gradient is neglected on the basis that it is not a
component susceptible of inducing hot tearing. Defining then the average
density of the solid-liquid mixture as ρ = ρsgs + ρƐgƐ, where gν and ρν are
the volume fraction and the density of phase ν (ν = s, Ɛ), and using
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 21

Darcy’s equation describing the interdendritic flow in a mushy zone, the

following expression is derived for unsteady conditions [1, 10]:

∂g s §K ·
ρ + (1 + β ) g s (ε yy + εzz ) − div ¨ ( gradPl − ρl g ) ¸ = 0. (1)
∂t ©μ ¹
β = (ρs/ρƐ - 1) is the solidification shrinkage, ε yy and εzz are the two
components of the strain rate of the solid perpendicular to the thermal
gradient (see Fig. 1), K is the permeability of the mush, μ is the viscosity
and pƐ the local pressure in the interdendritic liquid, g being the gravity
vector. Note that ρs and ρƐ have been assumed constant, while gs is con-
stant in the directions perpendicular to the thermal gradient (directional
solidification). It is interesting to notice that although a deformation is
applied, it is the strain rate that appears in Eq. (1). This expression is
fairly general and can be interpreted as follows: solidification shrinkage
(1st term) and/or deformation of the solid (2nd term) have to be compen-
sated by liquid flow (3rd term) if pores or hot tears are to be avoided. In
the case a third phase (pores or hot tears) is considered, the right hand
term of Eq. (1) is simply replaced by - ∂gp/∂t, where gp is the fraction of
pores (or hot tears) [11].
Under steady directional solidification conditions at a velocity vT, the
maximum pressure drop, Δpmax, across the mushy zone and associated with
deformation and solidification shrinkage is given by:
L L
E g
Δpmax = (1 + β ) μ ³ dx + vT βμ ³ l dx (2)
0
K 0
K
where E is the cumulated strain rate E ( x) = g s (ε yy + εzz )dx . These two
³
integrals can be transformed into integrals over temperature, thus introduc-
ing a “competition” between strain rate and thermal gradient, G, for the
first contribution, and the standard vT/G ratio for the shrinkage term as
already derived by Niyama in his porosity criterion [12].
22 J.-M. Drezet, D. Allehaux

Fig. 1. Schematics of hot tearing formation and of the two-phase problem consid-
ered in the RDG approach [11]

The RDG criterion simply states that a hot tear forms if the local pres-
sure in the liquid, i.e., the metallostatic pressure minus the pressure drop,
falls below a given cavitation pressure. This criterion has nevertheless a
few limitations:
• using only the perpendicular component of the plastic strains is not
strictly valid, the longitudinal component also inducing some suction (or
expulsion) of the liquid;
• the lower bound of the integrals of Eq. (2) is ill-defined. As gs tends to-
wards unity, the permeability goes to zero and the calculation diverges.
In practical situations, this bound is set up to a value of gs at which the
solid is considered as coherent, i.e., the liquid remains only as liquid
pockets (no continuous liquid films).
• in relation to grain boundaries, the method does not consider any local-
ization of the strains and feeding.
Stating the local pressure in the liquid films must not fall below a cavita-
tion pressure if no hot tears should form is equivalent to state that the
strain rate must remain lower than a maximum value. In other words, the
mush can sustain some deformation but its rate should remain low enough
in order to permit liquid feeding. The deformation is indeed limited by the
ductility of the solid + liquid mixture. Ductility curves obtained by tensile
tests exhibit the typical shape of U when liquid and solid phases are pre-
sent as shown in Fig. 2 [13, 14]. Some ductility is present at temperatures
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 23

around the coherency temperature (beginning of mechanical resistance in

tension), then decreases when approaching the solidus temperature before
increasing again in the solid state. The brittle temperature range, BTR, cor-
responds to the temperature interval where the ductility is really low [15].
In order to avoid hot cracking, the thermomechanical path ε (T ) of the al-
loy must not cross the U ductility curve. In other words, the slope of ε (T )
is limited. Noticing that ∂ε = ∂ε = ε& & , the strain rate is
* ∂t
∂T ∂t T ∂T
limited too, as found in the RDG approach. Note that the quantity ε& & has
T
also been used as a hot tearing indicator [3].

Fig. 2. U ductility curve for the Al-Cu 4.5 wt. pct. alloy together with two ther-
momechanical paths (adapted from [14])

Finally, the RDG criterion, originally derived for aluminium alloys, has
been extended to steel by Drezet et al. [16] by taking into account the peri-
tectic reaction that transforms ferrite into austenite. Moreover, Rindler
et al. [17] have further worked out the criterion for steels by releasing the
assumption of a uniform strain rate acting over the mush.

Alloy Hot Tearing Susceptibility

Under the assumption that the strain rate applied to the mushy zone is
uniform ε = ε yy + εzz and that steady state is reached, the liquid pressure
drop through the mush is made out of two contributions, Δpsh due to the
24 J.-M. Drezet, D. Allehaux

solidification shrinkage and Δp mec due to deformation [10]. The two con-
tributions can be written as:
180μ ª (1 + β)Bε º
Δp sh +Δp mec = v βA +
2 « T »¼ (3)
Gλ 2 ¬ G
T
f s2 ³ f s dT
Tliq 2 Tliq Tcg
with A = ³
f sdT and B= ³ dT (4)
Tcg (1 − f s ) −
2 3
Tcg (1 f s)

λ2 , the secondary dendrite arm spacing, is the typical length of the micro-
structure used to define the permeability K. The two integrals A and B are
related to microporosity formation induced by the solidification shrinkage
and to hot tearing, respectively. The larger these quantities, the larger the
pressure drop and therefore the higher the risk to initiate a hot tear. Note
that the lower bound of the two integrals A and B are set to Tcg, the tem-
perature at which coalescence of the grains occurs.
Within the framework of the European Wel-Air project [18], the
RDG hot tearing criterion is used to analyse the influence of the alloy
composition per se and of the filler material on the hot tearing suscepti-
bility for new generation aircraft aluminium alloys. To do so, the solidi-
fication path of the alloy (plus the filler material) was obtained using
ProPHASE, a microsegregation program developed at Alcan Péchiney
[19]. As the solidification path should be relevant with the solidification
conditions undergone by the alloy during laser welding, a Scheil ap-
proach (no solid-state diffusion) is used owing to the very high cooling
rate experienced by the alloy. The grain coalescence temperature is not
easy to determine; the temperature corresponding to a solid fraction of
98% or the eutectic temperature if more than 2% eutectic has formed,
was adopted as it correctly predicted the hot cracking susceptibility for
binary alloys, the so-called Λ curve [1]. Figure 3 shows the solidification
paths computed with ProPHASE for five aluminium alloys. Their com-
position is detailed in Table 1.
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 25

Table 1. Composition for five selected aluminium alloys

As the end of solidification is particularly important, a zoom of the solidi-

fication paths is presented in Fig. 4. The ranking of the alloys with respect
to their hot cracking susceptibility HCS is presented in Fig. 5 where the in-
tegral A is represented, knowing that B varies in a very similar way. It ap-
pears that the alloys AA2139 and AA2098 exhibit the lowest hot tearing
susceptibility. Albeit, this trend must be balanced by the fact that these al-
loys present other shortcomings such as equiaxed grain zone (EQZ) and
possible liquation cracks.

1
0.9
0.8
2098
0.7
Solid fraction (-)

6013
0.6
0.5 6056

0.4 2139

0.3 2022
0.2
0.1
0
490 510 530 550 570 590 610 630 650
Temperature (°C)

Fig. 3. Solidification paths of the Wel-Air alloys as computed by ProPhase

26 J.-M. Drezet, D. Allehaux

1
0.99
2098
0.98
6013
0.97
6056
Solid fraction (-)

0.96
2139
0.95
2022
0.94
0.93
0.92
0.91
0.9
490 500 510 520 530 540 550 560 570
Temperature (°C)

Fig. 4. Zoom at solid fractions higher than 90% of the solidification paths

1.E+05

8.E+04 RDG (A)

HCS (integral A)

6.E+04

4.E+04

2.E+04

0.E+00
2098 6013 6056 2139 2022
Alloy

Fig. 5. HCS of the WEL-AIR alloys computed with the RDG criterion

To improve the weldability of the aluminium alloys, it is common to bring

a filler wire in different proportion to the weld bath, the idea being to in-
crease the Si content in order to get more eutectic. Albeit, the welded ma-
terial has lower mechanical properties and the use of filler material pre-
sents some technological constraints. The influence of the amount of filler
material on hot tearing susceptibility is computed for the alloy AA2098.
Figure 6 shows the computed solidification paths of the 2098 alloy to-
gether with 15%, 25% and 33% of 4047 filler material and Fig. 7, the cal-
culated hot cracking susceptibilities. It is interesting to notice that adding
some silicon completely changes the nature of the phases; in particular the
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 27

Al2Li3Si2 intermetallic phase appears when Si is present and the solidifica-

tion ends at 501°C with the deposition of Al2Cu. This has a large influence
on hot cracking as it decreases the HCS by almost a factor 6. Nevertheless,
this one remains almost unchanged when increasing the amount of 4047.
1
0.9
0.8
0.7
0.6
2098
0.5
2098 + 4047 15%
0.4
2098 + 4047 25%
0.3 2098 + 4047 33%
0.2
0.1
0
490 510 530 550 570 590 610 630 650

Fig. 6. Solidification paths of AA2098 alloy plus AA4047 filler

30000

25000

20000
HCS (Integral A)

15000

10000

5000

0
2098 2098 + 4047 15% 2098 + 4047 25% 2098 + 4047 33%
Filler content

Fig. 7. Influence of the filler content on HCS

Thermal Hot Tearing Criterion

The thermal criterion allows us to compute the hot tearing susceptibility of

a given alloy solidified under the conditions imposed by the process, i.e. in
a given thermal gradient and with a given solidification speed. The hot
cracking susceptibility, HCS, is therefore defined as the inverse of the
maximum strain rate sustainable by the mushy alloy. Under steady state,
28 J.-M. Drezet, D. Allehaux

this quantity is a function of the two integrals A and B but also of the local
thermal gradient G and solidification speed VT:
Δp max = Δpsh + Δp mec - ȡgh ≤ Δp cav
ε p ≤ ε pmax (G, VT , A, B, ǻpcav , Ȝ 2 ,...) (5)
1
HCS =
ε max
p

The computation of HCS for the AA6056 alloy (cf. Table 1) welded at a
speed of 1 m/min with a 3 kW laser is presented as an example. The steady
state temperature field in the butt joint is represented in Fig. 8. An eulerian
approach is used, i.e. the material is transported at the laser speed under
the fixed laser beam. The characteristics of the heat source can be found in
[20]. Using the local solidification conditions along the coalescence iso-
therm, HCS is computed with the help of CalcoSOFT3D [21]. The results
are shown in Fig. 9. As steady state is reached, HCS can be represented in
a plane perpendicular to the laser speed. A maximum appears close to the
bottom of the weld pool. Influence of the laser speed, power and heat
source parameters can be assessed with such an approach.
Finally, note that the present approach does not require any mechanical
computation, as the maximum strain rate the mush can sustain simply de-
pends on thermal quantities (cf. Eq. 5). However, only steady state can be
treated with the thermal criterion. The inclusion of a constitutive model for
the mushy alloy and the investigation of transients are presented in the
next section.

Fig. 8. Steady state temperature field during laser welding of the AA6056 alloy
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 29

Fig. 9. Distribution of HCS

Thermo-Mechanical Approach

By including the mechanical behaviour of the mush, the strain rate tensor
undergone by the dendrites during solidification can be computed and
therefore the induced pressure drop; the larger this value, the higher the
risk to initiate a hot tear [22]. The main challenge is to establish a reliable
constitutive model for the mechanical behaviour of the solidifying mate-
rial. Although the mush should be treated as a compressible medium with
the help of internal variable models [23], a simple incompressible model is
adopted as a first step and for sake of simplicity. Indeed, many mechanical
tests in the solidification interval are required to determine the numerous
parameters describing the compressibility of the mushy alloy [23].
In this section, the mechanical strain rates undergone by the mushy zone
at the rear of the weld pool are assessed in a simple configuration, the butt
joint, in order to predict how process parameters can decrease the risk of
hot tearing in steady state as well as transient regimes such as run-in and
run-out. The results presented here deal again with the AA6056 alloy. Its
BTR is considered to be 510°C–550°C and corresponds to solid fractions
higher than 95% (cf. Fig. 4).
Thermal contraction arises as soon as the dendritic network is well de-
veloped and interconnected, that is at a solid fraction of 85% i.e. at a tem-
perature of 600°C (coherency temperature). The rheology of the alloy is
given by the classic viscoplastic Ludwik’s model and is detailed in [20].
30 J.-M. Drezet, D. Allehaux

The mesh used for the butt joint configuration is presented in Fig. 10.
Due to symmetry, only one half of the whole domain is meshed. The laser
is supposed to travel along the z-axis and mesh is refined in the regions
that undergo melting and solidification. The dimensions of the domain are
given in Fig. 10. The length of the domain (dimension in z) was deter-
mined so that the transient run-in and run-out regimes are well separated
by a steady state regime in the so called on-going zone.
The laser heat input is modelled by a volume heat generation within a
cylinder that is moving over the surface of the parts to be welded [20]. As
the mechanical field does not influence the thermal field (owing to the
absence of any air gap formation), the thermomechanical computation is
un-coupled. The thermal field is computed first and then used as a loading
for the mechanical calculation. As the two parts are tack welded prior to
laser beam welding, they are considered to be part of the same continuum
and therefore no contact elements are used. All plastic deformations are
reset to zero above 600°C, using the *anneal temperature feature in
Abaqus 6.5 [24].

25

y 15
x 3
z

Fig. 10. FE mesh for the butt joint configuration (dimensions in mm)

Following the Rappaz-Drezet-Gremaud [1] hot tearing criterion and the

approach of Monroe and Beckerman [25], the larger the strain rate is,
the higher the risk to initiate a crack. The mean and maximum values of
the trace of the total (elastic + thermal + plastic) strain rate over the BTR
are taken as hot tearing (HT) indicators:
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 31

HTI1 =
Mean


 (trε )
510°C ≤ T ≤ 550°C
(6)
HTI2 = Max

 (trε )
510°C ≤ T ≤ 550°C

These values are saved during cooling only (i.e. during solidification)
and at each integration point using the user-subroutine UVARM of Abaqus
6.5. The larger this value, the higher the risk to initiate a hot crack.
The case presented here corresponds to a laser speed of 50 mm/s, i.e. 3
m/min, with a constant power all over the specimen length. The thermal
field when the laser reaches the location z = 18 mm is presented in Fig. 11.
The weld pool corresponding to temperatures higher than 650°C is repre-
sented in grey.

Fig. 11. Temperature distribution when the laser is located at z = 18 mm. The
liquid pool appears in grey

The distribution of the Von Mises equivalent plastic strain rate (positive
quantity) and the two hot tearing indexes, HTI1 and HTI2, is presented in
Fig. 12 as a function of the position, z, along the specimen slightly below
the top surface (y = 2.5 mm, x = 0). As expected, the three regimes of
welding are evidenced: in the run-in, the strain rate and the hot tearing in-
dexes exhibit high values; then they decrease and plateau in the on-going
zone. Finally, in the transient run-out regime, the three quantities increase
again. Both HTI1 and HTI2 present the same trends. One can notice that
the equivalent plastic strain rate cannot be considered as an indicator for
hot tearing since it is always positive by definition, whereas HTI1 and
32 J.-M. Drezet, D. Allehaux

HTI2 can have negative values (i.e. no risk of hot cracking). This is the
case when the strain rate reaches its maximum in the run-out whereas the
two HT indexes get negative. In the steady state regime, HTI1 is 0.05 /s
whereas HTI2 is close to 0.16 /s. HTI1 and HTI2 both exhibit a maximum
in the run-in and run-out regimes, which means that hot cracking is prone
to occur at those locations. This is in accordance with industrial observa-
tions, where no hot tears are observed in the steady state regime but appear
in the run-in and run-out. Finally, the HT indexes are larger in the run-in
than the run-out, which seems to be less prone to hot cracking.

3
run-in steady state run-out
2.5

2 Plastic strain rate

Strain rate,HTI (/s)

HTI1
1.5 HTI2

0.5

0
0 5 10 15 20 25
-0.5
Specimen length, z (mm)

Fig. 12. Distribution of the equivalent plastic strain rate and HT indexes along
the specimen

Granular Model for Hot Tearing Criterion

As mentioned previously, one difficulty in the RDG approach is to esti-

mate the temperature at coalescence, Tcg that appears in Eq. (4). Indeed,
average methods are unable to account for the localization of hot tears at
grain boundaries. This localization is essentially due to the fact that liquid
films remain to lower temperature as compared to those located in between
dendrites of the same grains (Fig. 1). In other words, the formation of a
coherent solid network by coalescence or bridging of dendrites arms oc-
curs earlier within the grains as compared with grain boundaries. Rappaz
et al. [26] have introduced for that purpose a coalescence or bridging un-
decooling which, for a pure metal, is given by:
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 33

1 γ gb − 2γ sl
ΔTb = (7)
Δs f δ
where γgb is the grain boundary energy, γsƐ is the solid-liquid interfacial en-
ergy, Δsf is the volumetric entropy of fusion and δ is the thickness of the
diffuse interfaces. For an alloy, coalescence is reached when a coalescence
line (or surface) parallel to the liquidus, but ΔTb below, is reached. Within
a grain, there is no grain boundary energy and interfaces become attractive
as soon as they get within interaction distance, i.e., distance δ. At “repul-
sive” grain boundaries, γgb > 2γsƐ, bridging is reached at some ΔTb > 0.
This concept of bridging undercooling has been tested by experiment
[27] where two crystals of a nickel-base superalloy are laser welded to-
gether under well defined conditions with increasing misorientations. At
small misorientation (typically less than 15º), no hot crack forms along the
weld centreline, whereas at larger values, a crack is initiated under the
same conditions, thus showing the influence of the grain boundary energy.
The simulations done by molecular dynamics or by phase field are interest-
ing but correspond to very small regions near the incoming interfaces,
typically a few tens of nanometers or micrometers, respectively. In practi-
cal situations, hot tears are indeed located at grain boundaries, but the con-
figuration of these boundaries associated with nucleation and growth of
grains is essential. For this reason, a simplified approach of coalescence
for a large population of equiaxed grains was undertaken first by Mathier
et al. [7] and then by Vernède et al. [8, 9]. In this granular approach, a
random set of nucleation centres with random orientations is first gener-
ated in a given volume. Considering that the grains nucleate at the same
time and that the temperature difference across each grain is small with re-
spect to the growth undercooling, the grain boundaries correspond to the
Voronoï tessellation of the nucleation centres (Fig. 13a), i.e., the grain
boundary between grains I and K is the median line. Assuming globular
grains, the smooth solid-liquid interfaces is first approximated by linear
segment in each triangle linking a nucleation centre (open circle) and two
vortices of the tessellation (open squares) (Fig. 13b).
34 J.-M. Drezet, D. Allehaux

Fig. 13. Schematics of the granular model used for the simulation of solidifica-
tion and feeding in a network of equiaxed globular grains: Voronoï tessellation
(a), microsegregation model (b), feeding KPL model (c)

Solidification is then calculated within each triangle using a microseg-

regation model [9]. Therefore, at any time, the remaining width of the
liquid channel in between two grains is known providing the thermal field
is known. When the two solid-liquid interfaces get within interaction dis-
tances, coalescence is accounted for, using Eq. (7) and a Read-Schockley
grain boundary energy model. The next step is to calculate feeding within
the network of liquid films. For that purpose, a Poiseuille flow was first as-
sumed within the channels (Fig. 13c) [8, 9]. This flow is not constant along
a given channel as it has to feed solidification shrinkage and the relative
movement of grains, i.e., the flow has some losses along each channel. Fi-
nally, at each vortex of the Voronoï tessellation, the sum of the (signed)
incoming flows must be zero according to Kirchhoff’s law (so-called
Kirchoff Poiseuille with Losses, KPL, model).
The granular model has been applied to the directional solidification of
an Al-1%Cu alloy (Fig. 14). The thermal gradient was 60 K/cm and the
cooling rate -1 K/s (i.e., velocity of the isotherms equal to 170 μm/s). The
average grain density was set to 108 mí2, i.e., average grain size of 100 μm,
and the computation domain which spans across the whole mushy zone
contains 14’000 grains. The central figure in Fig. 14 shows the grains with
various grey levels, grains in solid contact (clusters) being represented
with the same grey level while the liquid films are in black. The small fig-
ures on the left are magnifications of 4 typical regions of the mushy zone
which are discussed hereafter. On the right of Fig. 14, the evolution of the
volume fraction of solid as calculated in horizontal sections of the grain
structure is represented together with the imposed temperature profile.
In region (a), typically for 0 < gs < 0.89, most the grains are isolated
and surrounded by liquid films. For 0.89 < gs < 0.97 (region (b)), clusters
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 35

of a few grains are formed but the liquid films remain continuous and in-
terconnected. In region (c) characterised by 0.97 < gs < 0.99, larger clusters
are visible, with a few isolated liquid films remaining inside. Finally, in
region (d) (0.99 < gs < 1), the solid network is continuous and liquid only
remains as isolated regions. As can be seen, this granular model is able to
predict the gradual transition from a continuous intergranular film network
to a continuous fully coherent solid. It should be emphasised that cluster
formation is directly induced by the stochastic nature of the nucleation
centre locations, a feature that has not been considered in past simulation
works related to hot tearing. Further analysis of the transition regions is
given in [11].

Fig. 14. Calculated mushy zone for an Al-1wt%Cu alloy cooled down at –1 K/s
in a gradient of 6000 K/m. Grains in solid contact are shaded with the same grey
level [11]
36 J.-M. Drezet, D. Allehaux

Conclusion
Hot tearing is a complex defect that involves many phenomena, in particu-
lar thermal and solidification aspects, stress-strain in an increasingly co-
herent solid, feeding in a gradually disappearing liquid film network. The
RDG criterion provided the first two-phase approach, which was further
improved using a more rigorous formalism and the complex rheology of
porous media. Nevertheless, these approaches are still based on averages
and do not consider any localization of strains and feeding at grain
boundaries. Granular models, while still limited to small portions of a so-
lidification process, have certainly the potential to answer some of these
questions, once mechanical aspects will be fully built in and the model will
be extended to 3 dimensions. The numerical simplicity of such approaches
makes it feasible from a CPU time point of view.

Acknowledgements

The European Commission and the whole consortium of the Wel-Air project,
Specific Targeted Research Project from the 6th framework, are acknowl-
edged for financial support. Isabelle Bordesoules from Alcan Péchiney CRV is
greatly thanked for providing the Prophase computations.

References

1. M. Rappaz, J.-M. Drezet, and M. Gremaud: Met. Mat. Trans. 30A (1999), 449.
2. J. Campbell: Castings (Elsevier, 2003).
3. D.G. Eskin, Suyitno and L. Katgerman: Progr. Mater. Sci. 49 (2004) 629.
4. M. M'Hamdi, A. Mo, and C.L. Martin: Met. Mat. Trans. 33A (2002) 2081.
5. M. M'Hamdi, H.G. Fjaer, A. Mo, D. Mortensen, and S. Benum in TMS 2004.
6. V. Mathier, J.-M. Drezet, and M. Rappaz: Two-Phase Modelling of Hot
Tearing in Aluminium Alloys using a Semi-Coupled Approach, Modelling
Simul. Sci. Eng. 15 (2007) 121–134.
7. V. Mathier, M. Rappaz, and A. Jacot: Mod. Simul. Mater. Sc. Engng 12
(2004) 479.
8. S. Vernède and M. Rappaz: Transition of the Mushy Zone from Continuous
Liquid Films to a Coherent Solid, Phil. Mag. 86 (2006) 3779–3794.
9. S. Vernède, P. Jarry, and M. Rappaz: A Granular Model of Mushy Zones-
Formation of a Coherent Solid and Localization of Feeding, Acta Materialia
54 (2006) 4023-4034.
 Application of RDG Hot Tearing Criterion to Welding of Al Alloys 37

10. J.-M. Drezet and M. Rappaz: In the proceedings of the First EsaForm Conference
on Material Forming, Ecole des Mines de Paris, CEMEF, Sophia Antipolis,
France, (Mar. 1998) pp. 49–52.
11. M. Rappaz, J.-M. Drezet, V. Mathier, and S. Vernède: In Materials Science
Forum vols. 519–521 (July 2006) pp. 1665–1674, Trans Tech Publications,
Switzerland.
12. E. Niyama, T. Uchida, M. Morikawa, and S. Saito: AFS Int. Cast Metals J.
(Sept. 1982) 52.
13. I.I. Novikov and O.E. Grushko: In Materials Science and Technology, (Sept.
1995) vol. 11, p. 926.
14. B. Magnin et al.: In Materials Science Forum vols. 217–22 (1996) pp. 1209–1214,
Trans Tech Publications, Switzerland.
15. W. Rindler, E. Kozeschnik, and B. Buchmayr: Computer simulation of the
Brittle Temperature Range (BTR) for hot cracking in steels, in Steel Res.
2000, vol. 71 (11), pp. 460–465.
16. J.-M. Drezet, M. Gremaud, R. Graf, and M. Gaümann: A new hot tearing cri-
terion for steel, proceedings of the 4th European Continuous Casting Confer-
ence, IOM communications, Birmingham, UK, (Oct. 2002) pp. 755–763.
17. W. Rindler, E. Kozeschnik, N. Enzinger, and B. Buchmayr: A modified hot
tearing criterion for steels, in Math. Modelling of Welding Phenomena 6,
(2002) pp. 819–835.
18. EU- FP6 Project: WEL-AIR.
19. C. Sigli, R. Dif, R.B. Commet, and T. Warner: In Materials Science Forum
vol. 426–432, Issue 1, (2003) pp. 351–356.
20. J.-M. Drezet et al.: In Mathematical Modelling of Weld Phenomena 8, Ed.
H. Cerjak (2007) pp. 137–152.
21. Calcosoft3D user manual, distributed by Calcom ESI, http://www.calcom.ch.
22. J.-M. Drezet and M. Rappaz: “Prediction of hot tears in DC cast billets” in
Light Metals, Ed. J.-L. Anjier, TMS, (2001) New Orleans, pp. 887–893.
23. O. Ludwig, C.-L. Martin, J.-M. Drezet, and M. Suéry in Met. Mat. Trans, vol
36A, (June 2005) pp. 1525–1535.
24. Abaqus user manual, http://www.abaqus.com.
25. C. Monroe and C. Beckerman: Development of a hot tear indicator for steel
castings, in Mat. Sc. and Eng. A, vol. 413–414 (2005), pp. 30–36.
26. M. Rappaz, A. Jacot, and W. J. Boettinger: Last-stage solidification of alloys:
theoretical model of dendrite-arm and grain coalescence, in Met. and Mat.
Trans. vol. 34A, (Mar. 2003) pp. 467–479.
27. N. Wang, S. Mokadem, M. Rappaz and W. Kurz: Acta Mater. 52 (2004) 3173.
Weld Solidification Cracking: Critical Conditions
for Crack Initiation and Growth

C.E. Cross, N. Coniglio

Federal Institute for Materials Research and Testing, Berlin, Germany

Abstract

A perspective will be given that outlines important considerations in

evaluating and predicting weldability. An examination will be made of the
local conditions necessary for solidification crack initiation and growth in
a weld. This will be done in light of two prominent thermo-mechanical ap-
proaches involving critical strain and critical strain rate. Critical conditions
will be identified based upon values available in the literature. Methods
used to measure strain and strain rate will be compared. The interpretation
of crack length measurements commonly used to quantify weldability will
be questioned, based upon our current understanding of the problem.
Complications and problem areas needing better definition will be identi-
fied and discussed, including strain distribution in the mushy zone, segre-
gation at grain boundaries, effect of impurities, and effect of cooling rate
on solidification path. Finally, a suggestion will be made for a new ap-
proach to weld development using in-situ strain rate measurement and new
composition-strain rate maps that define the boundary between crack and
no-crack conditions.

Introduction

The topic of solidification cracking as it applies to weldability has for

many years been a highly contentious subject, particularly with regard to
defining its cause. This is due in large part to its complex nature, involving
interplay between thermal, mechanical, and metallurgical components.
Welding research has concentrated on specific aspects of this subject (e.g.
weldability testing and crack length measurement), while ignoring the un-
derlying cause. Weldability tests normally permit only a relative ranking of
crack susceptibility. No one mechanism appears to explain the behavior
of all alloys. The ability to predict, a priori, the cracking behavior of a
40 C.E. Cross, N. Coniglio

particular alloy has remained to date an unattainable goal. However, with

advancing ideas regarding cracking mechanisms, together with improved
methods of testing and simulation, welding science is moving towards a
point where cracking may be predictable for a given alloy and welding ap-
plication.
The goal of this chapter is to provide a perspective that both identifies
and quantifies the critical conditions important to solidification crack ini-
tiation and growth. A new way of looking at critical cracking conditions
will be presented that is critical of traditional 50 year old ductility curve
approach. Of practical interest is an examination of how critical conditions
are represented in conventional weldability testing. Also, a new methodology
will be proposed for predicting weldability using critical stain rate – dilution
maps.

Thermo-Mechanical Analysis

Assuming that all thermo-metallurgical conditions are held constant (e.g.

constant alloy composition and cooling rate), it is useful to concentrate on
what conditions are required to initiate and propagate cracking from a
purely thermo-mechanical aspect. This serves to decouple the problem
from variable metallurgical influences and allows a focus on mechanical
concerns. In this regard, cracking is caused from the strains that arise from
the weld thermal experience, including both thermal and solidification
shrinkage, affected by weld heat input and conditions of restraint.

Crack Initiation

Critical Strain
This topic has traditionally been approached assuming that the mushy
zone, and its associated liquid films at grain boundaries, can withstand a
limited strain before failure. This follows from early work of Pellini [1],
and has been further developed by Prokovorov [2] and Senda et al. [3] who
have established characteristic ductility curves for specific alloys. These
curves define the critical strain that can be tolerated over the solidification
range (brittle temperature range- BTR), bounded by liquidus and non-
equilibrium solidus temperatures. Examples of several different ductility
curves are given in Fig. 1 for different alloy systems. It is observed that
minimum ductilities vary between 0.1–0.5% for aluminum alloys [4],
2–5% for plain carbon steels [5], and 1–9% for stainless steels [5].
 Weld Solidification Cracking 41

TL TS

BTR

B
Strain
A

εmin dε/dT
C

Temperature
(a) (b)

(c) (d)

Fig. 1. Ductility curve comparison showing (a) general schematic, (b) aluminum
alloys [4], (c) plain carbon steel alloys [5], and (d) austenitic stainless steel alloys [5]

Developers of the ductility curve concept believed that strain accumula-

tion is an important factor in determining whether cracks will form. It was
argued that strain accumulates from the beginning of solidification and, if
accumulated in sufficiently high amounts during solidification (as repre-
sented by line A in Fig. 1a), the ductility limit will be exceeded and crack-
ing will result. This rate of strain accumulation with temperature drop
(dε/dT) can be related to the strain rate (dε/dt) accordingly:

dε dt dε ε
= ⋅ = (1)
dT dT dt T
42 C.E. Cross, N. Coniglio

Consequently, numerous tests have been developed specifically to

measure and define the critical strain rate for different alloys. Among these
tests are the variable deformation rate (VDR) test [6], programmable de-
formation crack (PVR) test [7], and controlled tensile weldability (CTW)
test [8].
Considering Eq. 1, it follows from the U-shape of the ductility curves in
Fig. 1 that ductility must be dependent upon strain rate. In fact, this is how
these curves are typically generated, using a controlled strain rate (slow
bend) varestraint test to determine the minimum strain to cause cracking
when tested at different strain rates (see Fig. 2). Different behavior is ob-
served in Fig. 2 for aluminum and steel, where increased strain rate re-
duces ductility for aluminum, but increases ductility for steel [4, 9]. This
behavior is reflected in the shape of the respective ductility curves, with
aluminum having more of a truncated shape at high temperature (e.g.
Fig. 1b versus 1c).

(a) (b)

Fig. 2. Strain rate dependence of hot crack ductility for (a) aluminum alloys [4]
and (b) plain carbon steel alloys [9]

In addition to strain rate effects, Pumphrey and Jennings argued that a

large solidification range permits the build-up of more strain (assuming a
constant strain rate) and a greater likelihood that a critical strain will be
reached [10]. While alloys with a large solidification range are often found
more susceptible to cracking, this is not always the case. The most notable
exception is the Al-Mg alloy system, which demonstrates both good weld-
ability and a large solidification range. Although this exception brings into
serious question the validity of the strain build-up model, other arguments
 Weld Solidification Cracking 43

have been proposed to explain this inconsistent behavior, among them

including liquid wettability conditions that promote solid-solid bridging
between dendrites [11].
In more recent analyses, numerical simulations have been used to model
cracking behavior based upon strain in the mushy zone, building upon the
limited ductility concept. Feng simulated weld centerline strain for welds
made on aluminum 2024 plate (911–775 K solidification range) as shown
in Fig. 3 [12]. Here it is observed that welds started at the plate edge (x = 0
mm) accumulate significantly more strain than welds started 12.5 mm
from the edge (1/4 of plate length). Such behavior is commonly observed
in practice, particularly for low restraint conditions [13]. Note that strain
values are around 2% at the coherent temperature of 865 K for the
edge-started weld. Ploshikhin et al. have also simulated weld metal
strain and predict cracking will occur when grain boundary deformation
exceeds 15 μm [14].

Fig. 3. Predicted strain along centerline of aluminum 2024 weld for two different
weld starting positions: at plate edge and 12.5 mm from plate edge [12]

Critical Strain Rate

Following from the ductility curve-strain buildup concept outlined above,
there appears to be a critical strain rate that defines the boundary between
crack and no-crack conditions. This critical condition is represented by line
B in Fig.1a, where the deformation curve becomes tangent to the ductility
curve. Although critical strain rate characterization is limited, critical strain
rate for temperature drop (CST) have been reported [15, 16], expressed in
44 C.E. Cross, N. Coniglio

terms of strain per temperature (dε/dT). Measured values for critical strain
rate are compared in Table 1, where it is observed that values range be-
tween 0.1 and 5.0%/s. These values can be taken as a direct comparison of
weldability, with low values representing poor weldability.

Table 1. Critical strain rate to initiate weld solidification cracking derived from [4]
Critical Strain
Test Aluminum Alloy
Rate (%/s)
2017 0.15
5083 0.47
Slow Bend Trans-Varestraint 2219 0.50
Test 5052 0.64
5154 0.70
1070 5.00

Meanwhile, with significant advancements in solidification theory ap-

plied to castings, there is now a better understanding of a mechanism for
solidification cracking, which points to a direct connection with strain rate.
In particular, attention has been paid to the pressure drop occurring in the
liquid film between dendrites, and its resulting consequences on liquid
fracture. Fracture itself may be tied to decohesion of oxides, pore forma-
tion, or possibly even cavitation. This subject has been addressed by
Feurer [17], Campbell [18], and most recently Rappaz, Drezet and Gre-
maud (RDG) [19].
In the RDG theory [19], an account is made of the local thermal strain
conditions in addition to the solidification shrinkage and liquid feeding. In
this analysis, the maximum pressure drop (Δpmax) over the dendrite length
(L) is predicted to increase with a higher transverse strain rate ( ε ):
(1 − f S )
L L
E
Δpmax = (1 + β ) μ ³ dx + vβμ ³ dx (2)
0
K 0
K

§ ρS ·
where E ( x) = ³ f εdx , ß = shrinkage factor ¨©
S ρ L − 1¸¹ , K = perme-
ability, v = growth velocity, fS = fraction solid, and μ = viscosity.

The significance of applying RDG theory to welding is profound, as it

gives elevated importance to strain rate and it brings into serious question
the relevance of ductility curves as well as numerous existing crack predic-
tion simulations based upon strain. Although ductility concepts have become
firmly entrenched and are widely accepted in the welding community, they
 Weld Solidification Cracking 45

have no foundation in any known liquid fracture mechanism. Furthermore,

the existence of ductility curves can be explained using RDG concepts.
The U-shaped ductility curves of Fig. 1 can be taken to represent time-
temperature nucleation curves (see Fig. 4), replacing the strain axis with a
time axis. The application of higher strains means longer test duration, and
hence more time for crack initiation and growth. The C-curve kinetics for
crack nucleation follows from a low driving force at high temperatures (i.e.
low pressure drop Δp), and an increased resistance to deformation at low
temperature (i.e. increased coherency).

TL
low driving
force (Δp)
TC
increased
coherency
TS

time (for strain application)

Fig. 4. Schematic showing C-curve kinetics for hot crack nucleation and growth

Crack Growth

Only a limited amount of work has been devoted to understanding the fun-
damentals of crack growth in a liquid film. This curious lack of attention
may come from a presumption that conditions appropriate for crack initia-
tion must also result in crack growth. This assumption finds support in
many weldability tests (e.g. circular patch test), where it is observed that
once a crack is initiated, the crack will follow behind the weld pool until
the torch is extinguished. Another common problem is the failure to rec-
ognize crack initiation and growth as two distinguishable events, involving
two different mechanisms.
Two approaches for modeling liquid crack growth can be found: one
involving breaking of bonds [20]; and one involving the flow of liquid
[21]. Considering the relatively slow speed of crack propagation in
welding (i.e. at torch velocities around 4 mm/s), it may be difficult to
justify a mechanism based upon bond breakage. For the liquid flow
model, within a differential period of time, the transverse strain will be
46 C.E. Cross, N. Coniglio

compensated by both advancement of the crack and the back-flow of

liquid, as defined by a simple mass balance [21]:
ε (λ − x ) = x h + v L h , (3)

where ε is the rate of strain applied transverse to the growth direction, x is

the position of the crack relative to the solidus temperature, λ is the size of
the mushy zone, x is the rate of crack growth, h is the liquid film thick-
ness, and vL is the flow velocity of liquid to feed displacement.
Based upon direct observation of weld crack growth, x appears to re-
main at a fixed position within the mushy zone (i.e. at a constant solid
fraction, perhaps defined by the coherent temperature) and must therefore
grow at the same velocity as the weld torch [22]. It follows that if the crack
were to grow faster than the torch, it would advance toward the weld pool
and encounter conditions of hydrostatic pressure (i.e. more conducive for
porosity rather than a planar defect). If it were to grow slower than the
torch and drop behind the solidus temperature, the crack would immedi-
ately terminate. This suggests that there exists a critical strain rate, below
which the growth cannot be maintained for a given liquid flow rate. Data
given in Table 2 from Matsuda et al. supports this idea [6].

Table 2. Critical strain rate to stop cracking as a function of the aluminium base
metal, filler, and welding speed, as measured by Matsuda et al. [6]
Critical
Welding Speed Aluminium Alloy
Test Deformation
(mm/min) (Base Metal/Filler)
Rate (mm/s)
240 1100/1070 0.15
400 1100/1070 0.25
VDR
600 1100/1070 0.40
Variable
800 1100/1070 0.50
Deformation Test
400 5052/1070 0.05
400 5083A/5183B 0.18

Measuring Local Strain

Key to the establishment of critical cracking conditions, verification of

numerical simulations, and eventual prediction of cracking behavior is the
ability to measure transverse strain rate in the mushy zone behind a mov-
ing weld pool. This poses some unique challenges primarily associated
with the high temperatures encountered in welding. Different methods
 Weld Solidification Cracking 47

have been used over time, including real-time observation of scribe marks
[23] and moiré-fringe analysis of grid patterns [24].
Use has also been made of extensometers spanned across the weld, at-
tached either above or below the plate surface [25]. When applied to the
top surface, small diameter pins are attached on opposite sides of the joint,
and extension arms connect these pins to a remote extensometer (i.e. re-
moved from the torch path and possible exposure to heat). This allows
measurement of strain as the welding torch passes between the affixed
pins. When applied below the surface, the extensometer can be placed di-
rectly at the point of measurement, but this requires use of through-
thickness welds and a thin plate (i.e. plane strain) to provide any useful in-
formation. An example of one such measurement for an autogenous Al
6060 weld is given in Fig. 5a, where an extensometer (10.5 mm gage) has
been placed at the mid-span of a 100 mm long joint and the torch passes
over the extensometer at approximately 15 s. A compression cell is ob-
served ahead of the weld pool (negative strain), followed by a tensile cell
behind the weld pool, with a steadily increasing strain rate (approximately
0.5%/s in mushy zone).

(a) (b)

Fig. 5. (a) Output from extensometer placed across the root of an autogenous, full-
penetration GTA weld (10.5 mm gage) made on 4 mm aluminum 6060-T6 plate,
(b) first derivative of strain-time curve in (a) showing transverse strain rate

Extensometer measurements are limited to a fixed location and gage

length. A much more thorough characterization can be made using a digi-
tal image correlation (DIC) technique, which employs the computer aided
tracking of a random speckle pattern painted onto the plate surface prior to
welding [26]. This allows transverse strain to be monitored during welding
over a large area as demonstrated in Fig. 6a, and permits strain across the
mushy zone to be examined as shown in Fig. 6b. When using light optics,
this method is best suited for use on the bottom surface of the weld, where
arc light and gas fumes cannot interfere with speckle resolution. Use of a
laser based DIC system, however, would not suffer from this limitation.
48 C.E. Cross, N. Coniglio

One technique has been developed to measure strain directly in the weld
mushy zone making use of small discontinuities found on the weld surface
[27]. The relative movement of these discontinuities is tracked using high
speed photography, typically providing an effective gage length of around
1 mm. Measurements made using this technique, referred to as measure-
ment by means of in-sito observation (MISO), are given in Table 3 for
values representing the minimum strain to initiate cracking (at high strain
rate) in a tensile test. When compared against corresponding values based
upon applied strain in a trans-varestraint test (TVT), it is observed that
MISO values are significantly higher. This discrepancy may be due in part
to a hinge effect, discussed below in the weldability section, and reflects a
difficulty to control applied strain in the TVT.

(a) (b)

Fig. 6. (a) Light optical DIC measurements made on bottom side of plate during
autogenous, full-penetration, bead-on-plate GTA welding of aluminum 6060 plate.
Note black paint speckle pattern sprayed onto plate prior to welding. (b) Strain
across mushy zone of weld pool showing a constant outward displacement with
distance from weld centerline

Table 3. Minimum strain for cracking (in %) comparing MISO measurements

with applied strain in TVT test [27]
Alloy MISO Trans-Varestraint
plain carbon steel 2.0 <0.5
(30C-P)
304L 9.0 0.5–0.75
310S 2.3 <0.15
Al 5083 2.1 <0.15
 Weld Solidification Cracking 49

Weldability Testing

A broad variety of weldability tests have been developed over the years
that rely upon special conditions of joint configuration or applied strain to
generate cracks as outlined in Table 4. In most cases, weldability is deter-
mined based upon an evaluation of crack length measurements (i.e. crack
length parametrics). In other cases specifically involving application of
controlled strain rate, weldability can be related directly to a critical strain
rate (crack versus no-crack) for a given alloy and welding condition. As
argued above, this latter case holds the greatest promise for predicting
cracking because it relates to an actual cracking mechanism. Crack length
measurements, on the other hand, can only be used to rank the cracking
susceptibility of different alloys (i.e. provide a relative comparison), and
even this has come under increased scrutiny as discussed below.

Table 4. Comparison of Weldability Tests and Method of Evaluation

weldability tests
intrinsic
circular patch test (CPT)
houldcroft test
extrinsic crack length
longitudinal varestraint test (LVT) parametrics
transverse varestraint test (TVT)
modified varestraint test (MVT)
sigma-jig test

variable tensile strain test

variable speed varestraint
variable deformation rate test (VDR) critical
program. verformungsrisstest (PVR) strain rate
cont. tensile weldability test (CTW)

Crack Length Parametrics

Crack length measurements are most often reported in terms of total accu-
mulated crack length (TLC), as it is generally assumed that for fixed test-
ing conditions, an alloy with higher susceptibility must somehow result in
more extensive cracking. In longitudinal varestraint specimens, for exam-
ple, the TCL constitutes the sum of several smaller crack lengths, where
the number of cracks is bounded by the number of grain boundaries and
the length of each crack is bounded by size of the mushy zone. It is found
50 C.E. Cross, N. Coniglio

that cracking increases with applied strain up to some saturation limit, as

each grain boundary is fractured and opened to its maximum length. Duc-
tility theory proponents would argue that higher applied strain results in
more grains exceeding the critical strain value. However, with regard to
critical strain rate theory, this ductility argument can be countered by not-
ing that the application of higher strain also means longer test duration,
and hence more time for crack initiation and growth.

Fig. 7. Weldability comparison of aluminum alloys 6061 (4043 filler) and

Weldalite 049 (2319 filler) using the trans-varestraint test run at high strain rate
(50%/s) [28]

Additional problems are encountered when attempting to rank weldabil-

ity in terms of crack length measurement, particularly when comparing al-
loys with different thermal properties. For example when comparing alu-
minium alloys 6061 against 2094 (Weldalite 049) in Fig. 7, conditions of
constant heat input (i.e. constant current) suggest that alloy 6061 is more
weldable [28]. However, the exact opposite conclusion is reached for con-
ditions of constant penetration. To achieve the same penetration, Alloy
6061 requires higher heat input (i.e. higher current) which results in a lar-
ger mushy zone and corresponding higher TCL values.
To avoid the above noted problems associated with TCL comparisons, it
has been found useful to relate the maximum crack distance (MCD) in a
varestraint test (i.e. obtained at high applied strain, measured normal to
isotherms) to a corresponding temperature range. This requires the aid of a
thermocouple inserted into the weld pool. The resulting value, or solidification
cracking temperature range (SCTR) [29], is believed to be independent of
welding parameters and appears to represent a characteristic property
 Weld Solidification Cracking 51

reflecting relative weldability. As noted earlier, a large temperature range

suggests the possibility for more strain build-up, but it also represents a
more difficult condition for liquid feeding and a subsequent higher liquid
pressure drop.
In normal varestraint testing, both TVT and LVT, strain is applied at
relatively high rates, with different values reported as presented in Table 5.
It should be noted that the local weld metal strain rate, measured with the
MISO technique, is more than twice the applied strain rate. Under normal
circumstances, a fixed strain is applied over a duration of approxi-
mately100 ms, which has led to a general assumption that these tests in-
volve instantaneous strain application. Also, these strain rates far exceed
critical values for crack initiation by several orders of magnitude (recall
Table 1), which brings into question how results from this type of test can
ever be used to simulate real-world behavior.

Table 5. Comparison of weldability test strain rates

Applied Strain MISO Strain Weldability Test Reference
Rate Rate
50%/s – TVT [28]
40%/s 100%/s LVT [22]
– 130%/s Tensile [5]

Work by Robino et al. [22] has recently demonstrated a problem with

assuming strain to be instantaneous in a LVT test. Using high speed pho-
tography, a crack was observed to initiate in the mushy zone (at fs=0.78) at
a local strain of 1% during the application of a 4% augmented strain. This
crack then proceeded to grew simultaneously in opposite directions, both
towards the weld pool and away from it. The advancing crack tip grew at
the velocity of the weld torch (3.3 mm/s), thus maintaining its relative po-
sition at fs=0.78 and terminating as the applied strain becomes dissipated.
The retreating crack tip grew at a higher velocity and terminated when it
intersected the advancing solidus temperature. During the period of strain
application (approx. 120 ms), the weld pool advanced 0.4 mm, which is
the same order of magnitude as the measured crack length of 0.7 mm.
What this means is that MCD measurements coming from varestraint
tests should be expected to vary with torch travel speed and applied strain
rate and cannot be related to a characteristic temperature range (e.g.
SCTR) in any straight forward manner. It also suggests the importance of
standardizing strain rates and welding speeds for varestraint testing for
52 C.E. Cross, N. Coniglio

purposes of crack length and SCTR comparisons. Varestraint tests run at

sub-normal speeds will likely result in MCD values that are even more un-
representative of a characteristic temperature range.
One additional problem with varestraint testing is worth pointing out
here, as it helps to explain the discrepancies observed in MISO meas-
urements given in Table 3. When using the TVT test, it has been found
that the strain experienced in the mushy zone can far exceed the applied
strain. This is believed due in part to a hinging effect, whereby plastic
deformation is concentrated in hot material along the weld seam as
shown schematically in Fig. 8a. This is further demonstrated in strain
gage measurements compared in Fig. 8b and 8c, showing behavior of
wrought material versus weld material for aluminium 6061 tested at
room temperature. The wrought material strain (0.6%) closely matches
the applied strain (0.5%). However, the weld metal shows much higher
strain values (over 2%), because it is weaker than the base metal (50%
joint efficiency).

(b)

(a)

(c)

Fig. 8. (a) Schematic illustration showing hinge effect in TVT testing, (b) strain
gage placed on Al 6061 plate for TVT test with 0.5% strain die block, (c) strain
gage place on weld metal in pre-welded Al 6061 coupon, TVT tested with 0.5%
die block [28]
 Weld Solidification Cracking 53

Topics of Limited Understanding

The following topics are believed to have major influences on weld solidi-
fication cracking, but are themselves not well understood and need further
clarification in order to expand our understanding of cracking behavior.

Strain Rate Partitioning

Most of the local strain rate measurements considered above, using exten-
someter and DIC methods, have been made in the vicinity of the mushy
zone, but not in the mushy zone itself and not across a single grain bound-
ary. In order to relate local measurements to actual strain conditions across
a single grain boundary, for ideal application of theory, it is important to
know how strain (and hence strain rate) is partitioned within the mushy
zone. For this purpose, it becomes convenient to treat the mushy zone as a
composite material consisting of parallel strips of liquid and solid phases,
with behavior proportioned to the relative amount of each phase:
ε = f S εS + f LεL (4)

This becomes further complicated if it is considered that the strain

partitioned to grain boundaries is different from the strain partitioned
interdendritically.

Grain Boundary Segregation

Microsegregation of alloying elements due to partitioning between den-

drites during solidification has been extensively modeled using the Scheil
equation and its various different forms (e.g. to account for back-diffusion)
[31]. However, the concentration of segregates at weld metal grain
boundaries is not well understood or quantified. It has been well docu-
mented that segregates (e.g. sulfur in steel) migrate to the weld centerline
and that this behavior contributes significantly to weld cracking, particu-
larly for tear-drop shaped weld pools [32]. While models for cracking so
far have considered idealistic interdendritic conditions for reasons of sim-
plicity, it is clear that intergranular conditions are more appropriate.
54 C.E. Cross, N. Coniglio

Role of Impurities

In ferrous alloys, the impurities sulfur and phosphorous play a major role
in determining weldability and this has been fairly well documented. In
aluminum alloys, however, the role of impurities is less clear and this re-
mains a little studied subject [33]. Impurities oxygen and hydrogen, for ex-
ample, could conceivably affect weldability; where oxide films may serve
as crack initiation sites, and hydrogen may serve to reduce these oxides.
Hydrogen also results in pore formation that may help feed shrinkage. Dis-
solved hydrogen could also make it easier to initiate cracks through cavita-
tion. The impurity iron has been shown to affect both porosity and crack-
ing behavior in aluminum castings [34], but its affect in welding aluminum
remains largely unexplored.

Solidification Path

The solidification path of an alloy, and its corresponding solid fraction

versus temperature curve, is an important input to the RDG model. While
there is now popular use of thermodynamic based software predictions, it
is of course important to also perform experimentation to confirm solidifi-
cation behavior. It is of particular interest to understand the effect of cool-
ing rate on phase formation. For example, solidification microstructures
are compared in Fig. 9 for aluminum 6060 showing a significant change in
constituent phase distribution with cooling rate, in addition to the finer
dendrite spacing and grain structure. A single sensor-differential thermal
analysis (SS-DTA) technique developed specifically for welds, has proven
useful in identifying phase reactions and solidification ranges [35, 36].

(a) (b)

Fig. 9. Solidification structure for Al 6060 (a) casting (9°C/s) and (b) gas tungsten
arc weld (54°C/s). Shown in (a) are dendrites with interdendritic constituents and
in (b) grain boundaries with finer constituents [35]
 Weld Solidification Cracking 55

Crack Prediction

Future approaches for predicting the onset of cracking for a given indus-
trial application will involve applying material information to a cracking
mechanism with the aid of numerical simulation. Inputs to this simulation
will include pertinent welding information (heat input, travel speed), alloy
composition (base metal and filler dilution), and restraining conditions
(weld geometry, fixture design). The ability to accurately predict cracking
behavior will require a thorough understanding of the cracking mechanism
itself, something that does not yet exist.
Meanwhile, and until a reliable cracking model is developed, a more
pragmatic approach may prove useful. This could, for example, involve
development of composition-strain rate maps as demonstrated in Fig. 10
[8]. Based upon Controlled Tensile Weldability (CTW) testing, a boundary
has been established for the aluminum 6060/4043 alloy system that defines
how much 4043 filler metal is required to avoid cracking for a given local
strain rate. Using this map, a welding engineer can make informed deci-
sions regarding weld procedure development (including filler dilution)
based upon local strain rate measurements. Such measurements could con-
ceivably be made using a laser scanner across the mushy zone. While the
given data is for the Al 6060-4043 alloy system, a similar type of map
could be generated for any ferrous or non-ferrous alloy.

Fig. 10. Critical strain rate – ductility map showing conditions required for weld
solidification cracking in Al 6060 + 4043 filler using gas-tungsten arc process
(full penetration, 4 mm plate, 4 mm/s travel speed) [8]
56 C.E. Cross, N. Coniglio

Summary

The long-held belief in the welding community that solidification cracking

is a ductility limited phenomena, and that ductility curves represent a char-
acteristic critical strain behavior, has been questioned. Now that mecha-
nisms are being developed to explain how a liquid fails, a mechanism is
now available to help guide how we look at weld cracking. Strain rate, be-
fore just a secondary effect needed to achieve a critical strain, has now be-
come a key controlling factor for both crack initiation and growth. Com-
mon weldability tests involving the application of a fixed strain at high
rates, far exceeding critical strain rate values, are likewise coming under
scrutiny as to just what it is they represent.

References

1. Pellini WS (1952) Strain theory of hot tearing. Foundry 80: 125–199

2. Prokhorov NN (1956) The problem of the strength of metals while solidifyin-
during welding. Svar Proiz 6: 5–11
3. Senda T, Matsuda F, Takano G (1973) Studies on solidification crack sus-
ceptbility for weld metals with trans-varestraint test. J Japan Weld Soc 42:
48–56
4. Nakata K, Matsuda F (1995) Evaluations of ductility characteristics and
cracking susceptibility of Al alloys during welding. Trans JSRI 24: 83–94
5. Matsuda F, Nakagawa H, Kohmoto H, Honda Y, Matsubara Y (1983) Quantita-
tive evaluation of solidification brittleness of weld metal during solidification by
in-situ observation and measurement (report II). Trans JWRI 12: 73–80
6. Matsuda F, Nakagawa H, Nakata K, Okada, H (1979) The VDR cracking test
for solidification crack susceptibility on weld metals and its application to alu-
minum alloys. Trans JWRI 8: 85–95
7. Herold H, Streitenberger M, Pchennikov A (2001) Modelling of the PVR test
to examine the origin of different hot cracking types. In: Mathematical model-
ling of weld phenomena 5, Inst Metals, London, pp 783–792
8. Coniglio N, Cross CE, Michael Th, Lammers M (2006) Defining a critical
weld dilution to avoid solidification cracking in aluminum. In preparation for
Welding Journal, presented at AWS, Atlanta
9. Matsuda F, Nakagawa H, Tomita S (1986) Quantitative evaluation of solidifi-
cation brittleness of weld metal during solidification by in-situ observation
and measurement (report III). Trans JWRI 15: 125–133
10. Pumphrey WI, Jennings PH (1948) A consideration of the nature of brittleness
at temperatures above the solidus in castings and welds in aluminum alloys.
JIM 75: 235–256
11. Borland JC (1960) Generalized theory of super-solidus cracking in welds and
castings – an initial development. Brit Weld J 7: 508–512
 Weld Solidification Cracking 57

12. Feng Z (1994) A computational analysis of thermal and mechanical conditions

for weld metal solidification cracking. Welding in the World 33: 340–347
13. Cross CE, Böllinghaus Th (2006) The effect of restraint on weld solidification
cracking in aluminium. Welding in the World 50: 51–54
14. Ploshikhin V, Prikhodovsky A, Makhutin A, Ilin A, Zoch HW (2005) Inte-
grated mechanical-metallurgical approach to modeling of solidification crack-
ing in welds. In: Hot cracking phenomena in welds, Springer, pp 223–244
15. Arata Y, Matsuda F, Nakata K, Shinozaki K (1977) Solidification crack sus-
ceptibility of aluminum alloy weld meatls (report II). Trans JWRI 6: 22–35
16. Herold H, Pchennikov A, Streitenberger M (2004) Assessment of hot cracking
initiation by laboratory test procedures and FEM simulation associated ex-
perimental measurements during welding of large weld components. J Japan
Weld Soc 22: 211–217
17. Feurer U (1977) Influence of alloy composition and solidification conditions
on dendrite arm spacing, feeding, and hot tear properties of aluminum alloys.
In: Proc international symposium engineering alloys, Delft, pp 131–145
18. Campbell J (1991) Casings, Butterworth-Heinemann
19. Rappaz M, Drezet JM, Gremaud M (1999) A new hot-tearing model. Met Mat
Trans 30A: 449–455
20. Murakawa H, Serizawa H, Shibahara M (2005) Prediction of welding hot
cracking using temperature dependent interface element. In: Mathematical
modelling of weld phenomena 7, Maney, pp 539–554
21. Braccini M, Martin CL, Suery M, Brechet Y (2000) Relation between mushy
zone rheology and hot tearing phenomena in Al-Cu alloys. In: Modeling of
casting, welding, and advanced solification processes IX, Shaker Verlag, Ger-
many, pp 19–24
22. Robino CV, Reece M, Knorovsky GA, DuPont JN, Feng Z (2005) Prediction
of maximum crack length in longitudinal varestraint testing. In: Proc. 7th int.
conf. trends in welding research, ASM Int, pp 313–318
23. Chihoski RA (1972) The character of stress fields around a weld arc moving
on aluminum sheet. Weld J 51: 9s–18s
24. Johnson L (1974) Moire techniques for measuring strains during welding. Exp
Mech 14: 145–150
25. Kannengiesser T, McInerney T, Florian W, Böllinghaus T, Cross CE (2002)
The influence of local weld deformation on hot cracking susceptibility. In:
Mathematical modelling of weld phenomena 6, Maney, pp 803–817
26. Sutton MA, Wolters WJ, Peters WH, Ranson WF, McNeill SR (1983) Deter-
mination of displacements using an improved digital correlation method. Im-
age and Vision Computing 1: 133–139
27. Matsuda F, Nakagawa H, Nakata K, Kohmoto H, Honda Y (1983) Quantita-
tive evaluation of solidification brittleness of weld metal during solidification
by means of in-situ observation and measurement (report I). Trans JWRI 12:
65–72
28. Cross CE, Tack WT (1993) Factors affecting the weldability evaluation of
aluminum alloys. unpublished research, Martin Marietta Aerospace, Denver
58 C.E. Cross, N. Coniglio

29. Lippold JC (2005) Recent developments in weldability testing. In: Hot cracking
phenomena in welds, Springer, pp 271–290
30. Unpublished research (2006), BAM, Germany
31. Brooks, JA (1990) Weld microsegregation: modelling and segregation effects
on weld performance. In: Weldability of Materials, ASM, pp 41–47
32. Savage WF, Aronson AH (1966) Preferred orientation in the weld fusion
zone. Welding J 45: 85s–89s
33. Coniglio N, Cross CE (2007) Weld parameter and minor element effects on
solidification crack initiation in aluminium. In: this conference proceedings,
pp. 277–310
34. Lu L, Dahle AK (2005) Iron rich intermetallic phases and their role in casting
defect formation in hypoeutectic Al-Si alloys. Met Mat Trans 36A: 819–835
35. Coniglio N, Cross CE (2006) Characterization of solidification path for alu-
minium 6060 weld metal with variable 4043 filler addition. Welding in the
World 50: 14–23
36. Alexandrov BT, Lippold JC (2006) A new methodology for studying phase
transformations in high strength steel weld metal. In: Proc 7th int conf trends
in welding research, ASM Int, pp 975–980
Consideration of the Welding Process
as a Thermo-Physical Mechanism to Control
Cracking in Weldments

H. Herold, M. Streitenberger

Otto-von-Guericke University, Institute for Materials and Joining

Technology, Magdeburg, Germany

Abstract

Prevention of cracking in weldments is the result of understanding and

evaluating extensive experience from cases of damage. The welding
process-con trolled mechanics for the initiation of weldment cracking by
extensive strain due to thermal expansion of the moving weld pool with its
surrounding temperature field, and a local characterisation of the dynamic
non-linear behaviour during solidification in, around and behind the weld
pool in interaction with the weldment restraint are described in terms of
thermo-physical mechanics. Different experience from solidification
cracking assessment results in a concept for controlling weldment cracking
by means of thermo-physical mechanics.

Introduction

Cracking in weldments is an unwished side effect of the fusion welding

process when applied to metals in manufacturing structural components
and providing them with special-use properties for lifetime. Historically, in
any metal-working process priority has been given to high toughness in
combination with all other aspects of material resistance –just to mention
the Damascus blade and all types of novel steel-makers’ products to which
a special metal-working treatment was applied to exclude brittle failure
and any cracking. Thus, weldment cracking should be avoided, but if
occurring it will indicate that in the respective operation, e.g. fusion
welding, tolerances have been exceeded.
60 H. Herold, M. Streitenberger

Failure analysis and assessment concepts are made during the lifetime of
welded structures if cracks have formed and grown to macroscopic scale,
endangering the reliability of structural components. Cracks originate at
the weakest link of material behaviour in the microstructure, caused by
locally positioned thermo-mechanical overstress within the welded
structural component on which general thermo-mechanical stress is
imposed by welding.
On the example of solidification cracking, the paper follows the
philosophy: When we can describe and analyse the hot cracking
phenomena in both material behaviour and welded structural
components, concentrating on the crack – initiating thermo-mechanical
processes due to welding and on the state of art in hot-cracking
assessment, then we can develop more efficient engineering tools for
their prevention in stages of manufacture.

Weldment Cracking

The welding procedure supplies the thermal energy for local fusion with
the intensity and efficiency of the heat source (arc, beam). The material of
the assembled weld component absorbs the heat by its thermal capacity
and mass (volume), with heat losses due to heat transfer by resistance,
convection and emission. The heat dissipates thermodynamically into the
structures’ volume by heat conductivity, changing the physical properties
of the metal by thermal expansion. Quasi-steady temperature fields and
temperature gradients around the moving welding pool characterize the
changing local behaviour, which results in superposition of both the
microstructural material transformation and thermo-mechanical structural
kinetics, balancing the local thermal, mechanical and constitutional
yielding.
Thermal cycles are used to describe the constitutional and thermo-
dynamic reaction of the composition of the alloy on solidification and
HAZ. They also form the basis for calculation programs for temperature-
time-transition, correlation of microstructure and mechanical behaviour,
grain growth, and cooling time concepts in general. Quasi-steady
temperature fields describe the 2D or 3D thermo-physical and thermo-
dynamical heat transport around the weld pool in relation to the welded
component, which is used for the calculation of temperature-induced
mechanical reactions as distortion, displacement, stress and strain with the
help of Finite Element Analysis (ANSYS, ABACUS).
 Thermo-Physical Mechanism to Control Cracking in Weldments 61

Whether a designed component of a selected material is weldable or not

cannot be measured directly. It is the result of balancing equilibrium crite-
ria which decides whether the welding-process treated material of the
component to will yield wether micro-fissuring will be initiated, or
whether break at usage will occur.

Classification of Weldment Cracking

welding

material structure

When ? Where ? Why ? What?

crack mechanical thermal

welding consider
location load load

post weld Stress allotrope

base metal weldment
treatment Stress rate reaction

service micro strain physical

HAZ
condition structurs strain rate reaction

transition distance to displace chemical

life time fusion line lattice
precipitation fusion line ment composition

ambient residual
weld metal atom
media stress

super-
position

Fig. 1. Classification of weldment cracking

The classification of weldment cracking in Fig. 1 verifies not only welding

processes in fabrication but also post-weld heat-treatment like stress relief,
special cladding in automotive industry and higher temperature service
conditions on lifetime elicited by manufacture. Each further heat-
treatment, post-weld treatment, repair welding, service conditions at higher
temperatures will result in local material reactions, superposing that of the
prior welding and damaging the material locally, and will initiate
weldment cracking.
When we investigate how a crack is initiated, where and why the crack
occurs and what the crack’s path or way is, than we consider existing
cracks at weldments. The different weldment cracking types can be traced
back to the weakest link of their microstructure where, on the one side, the
local material failure of the welding procedure conditions occur due to
62 H. Herold, M. Streitenberger

metal-constitutional, metal-physical and thermo-mechanical conditions,

and, on the other side, metal-physical laws of dislocation movement in
heat-affected and deformed material act. The failure of a welded compo-
nent is always related to local micro-structural disruptions of the weakest
link of the whole structure.

Tools for the Characterisation of Solidification Cracking

Finite Element Analysis (FEA)

In the sixties, the analysis of hot cracking proved that crack initiation
depends on superposition of a critical internal deformation or deformation
rate (thermal and external) during the existence of a material-specific
cracking-sensitive temperature range (brittleness temperature range (BTR)
or ductility loss temperature rage (DTR) [1, 2, 3].
During one-sided welding processing, the liquid weld metal solidifies
behind and/or at the sides of the molten weld pool, while at the same time
the constant heat input of the welding arc produces non-uniform thermal
deformation and shrinkage in the plates, especially at transient stages of
the structure. The features of high density heat input and large shrinkage
were experimentally investigated in the seventies for hot cracking
prevention in shipbuilding, focussed on the effect of external deformation
behaviour, transient strain, seam-end cracking and its prevention, and for
modelling and simulation by Finite Element Analysis (FEA) [2, 3, 4, 5, 6,
7, 8, 9, 10, 11].
In the eighties, mostly experimental and simulative investigations
followed, concentrated on the effect of root gap changing during welding
of butt-weld joints [2, 3, 4, 5], with partially systematic and quantitative
theoretical considerations. The results of welding shrinkage computed by
FEA were compared with experimental results in the nineties, when the
effects of local heating, welding sequences and constraint conditions on
the transverse shrinkage of large plates [3, 4, 5, 6] as well as the effect of
precision ship assembling came under discussion. All these investigations
described and calculated the correlation between solidification cracking,
heat input, welding processing and deformation fundamentally and in
detail, but the results were not related to each other for assessing the hot
crack risk during one-sided welding of real components due to transient
stages.
Further FEA simulation [4] was executed to determine the local residual
stress and strain distribution along the seam in a welded component with
respect to the position of hot cracks. Hot cracking prevention has been
carried out in this principle case by welding-procedure associated effects,
minimizing the non-thermal high-temperature transverse strain to between
 Thermo-Physical Mechanism to Control Cracking in Weldments 63

1400 and 1000°C by reduced heat input per unit length and increased weld
speed all over the welded seam as well as by special designing and
assembling for welding [5, 6].
Welding reliability in structural transient stages has become more
important as refers to hot cracking prevention and hot cracking assessment.

Microscopy and Fractography

Microscopy offers increasingly precise and fundamental contribution in

identifying micro-structural and allotrope constituents (by analysing mixed
crystals, precipitations, inter-metallic and other special phases), their size
and distribution within the grain or at the grain boundaries. Furthermore,
micro-fractography gave progress to classification and better
understanding of grain-boundary constituents and interactions that bring
about solidification cracking.

Weldability Tests

Numerous special weldability test procedures were created to quantify the

various effects of sensitivity on hot cracking. The different hot cracking
test procedures and their comparability were discussed on international
scale [7], which resulted in successful European standardisation. While the
theoretical hot cracking research is focussed on local thermo-mechanical
kinetics and metallurgical dynamics during solidification, the hot-cracking
test methods applied in practice are based on global concepts with
measurable constitutional, mechanical, thermal and time-affected test
criteria [8]. Their development has changed from descriptions of hot
cracking phenomena by yes/no criteria (test procedure with self-loading
specimens [9]) to quantification of hot cracking sensitivity by thermo-
physical simulation as well as directly in weldment during welding
procedures, using external loading for the specimens [10].
First, the PVR-test [11] was applied to quantify hot cracking sensitivity
of weld and base metal. In this, the material was evaluated, ranking from
hot cracking sensitivity to solidification cracking, liquation cracking as
well as ductility dip cracking. Crack sensitive bands in the HAZ (for
steady stages) with determined distances to the fusion line in accordance
with the crack sensitising temperature ranges induced by the welding proc-
ess were defined [12]. Second, the PVR test was modelled with the help of
FEA [13] to describe the origin of hot cracking in welding. That is the cor-
relation of the moving temperature field and internal deformation within
the BTR, derived mathematically as the changed deformation BTR within
the BTR dependent on time. Third, the PVR-test was applied to qualify
64 H. Herold, M. Streitenberger

welding procedures by tailored-to-material heat input, suited filler metal

and consumables, examining the effect of shielding gases and dilution
level on hot cracking [14]. The PVR test was compared with other test
procedures applied in Magdeburg, like MIS Test and DSI hot ductility test
which work with externally loaded specimens [15].
The analysis of different hot cracking test procedures with externally
loaded specimens proved that for the formation of all hot cracking types
superposition of a critical deformation speed within a material-specific
temperature range is necessary. The premises for any hot cracking
assessment in welding are defined equilibrium criteria, quantifying the
local processing conditions for crack initiation of metal-physical
associated crack mechanisms at the moment of instability. The critical
values for weldment reliability cannot be directly examined in hot cracking
tests.

Role of Fracture Mechanics in Respect to Weldment Cracking

The guide Fracture mechanical assessment of cracks on welded

components was intended for life-time reliability assessment of locally
damaged structures to exclude critical failure [16]. Measurement of weld-
zone fracture toughness by KIC, CTOD and critical J-Integral value was
recommended by AWS and BS [17] not only for failure assessment.
About 60% of applied fracture-mechanical failure analyses were related
to weld imperfections which had grown into macroscopic scale during
lifetime, thus causing local failure at the weakest link in the microstructure
in a once thermo-mechanically overstressed area within the entire
mechanically loaded welded component. In classic fracture mechanics,
different crack-criteria balance the equilibrium conditions relating to
stress, stress-intensity factor and crack size. If we can understand the
special criteria for the initiation of welding cracks or hot cracking, then we
can, based on this knowledge, develop tools for its prevention.
In the early nineties, the need for higher guarantees for reliability of
welded structural components resulted in the IIW-case study collection on
the assessment of significance of weld imperfections In this, based on
examples, recommendations on significance and assessment of weld
imperfections with respect to structural stress-strain analysis were offered,
comparisons between predicted and true failure loads were made, material
and welding-process conditions were analysed, and conclusions about
repair and other were drawn [18].
Figure 2 considers the question: Which aspects of the classic fracture-
mechanic associated assessment can be transferred to a welding process
associated assessment? On the left-hand side, a fracture-mechanic test
procedure for measuring fracture toughness (KIC, CTOD and critical
 Thermo-Physical Mechanism to Control Cracking in Weldments 65

J-Integral) as a method for the assessment of failure due to measurable

cracks that have grown in a welded component under load, and related
Failure Assessment Diagrams FAD for calculation of the failure risk can
be see. Failure assessment is a special tool for defining component reli-
ability, derived from fracture mechanics, in order to consider failure due to
special service conditions (as relate to lifetime reliability of plants at high-
temperature, creep, corrosion, e.g.).

Crack at Mechanism: weldment cracking

lifetime
mechanical Cross displacement
Elasto-plastic load thermal-deformation
Fracture mechanics process mechanics

tear- processing
zones Time,

weld procedure
associated load

Continuum Welding process associated mechanics

mechanics depending on weld zone location

Fig. 2. Comparison of classic fracture mechanics and welding-process associated

assessment

On the right, the welding-process associated test procedure determines

the material resistance to solidification cracking in welding procedures as a
method for assessing solidification-cracking initiation on the structural
component due to the welding procedure applied.
Both assessment concepts differ, as a rule, in the way a load is applied
(externally applied mechanical load and welding-process induced complex
thermal-physical and thermo-mechanical load), and in the type the
inhomogeneous processing zone (estimation of crack initiation in welding
as the critical the crack stage or crack propagation due to crack growth at
lifetime) is defined The concepts agree with the rule of geometrical
relations between the moving processing zone and the structural
component during welding.
66 H. Herold, M. Streitenberger

It also follows that material properties as toughness defined in notch

impact tests as well as weldment cracking sensitivity defined in weldability
tests are not suitable for assessing structural failure or crack initiation at
welded structural components.

Assessment of Solidification Cracking

Modelling Solidification Cracking Mechanisms and Criteria

In the past, Pellini [1] was the first to describe the effects of solidification
cracking by the hot spot in casting. The residual melt, enriched with
segregating elements, lowers the effective solidification temperature and
delays solidification time. The assessment quantifies the stress-concentration
over the length of the casting due to the linear expansion coefficient during
cooling down from liquid`s temperature, thus producing strain by
shrinkage. Campbell [19] describes the local conditions for solidification
crack formation in castings with five feeding mechanisms fundamentally,
by liquid feeding, mass feeding, interdendritic feeding (eutectic, solid
solution), solid feeding (elastic, plastic) and burst feeding. These
descriptions initiated two different directions for further investigation of
solidification mechanisms, giving priority either to the effects of thermo-
mechanical changes at solidification, or to the changing constitution during
solidification.

Criteria Determined by Constitutional Solidification

The following models of solidification (without considering solidification

cracking) quantify equilibrium conditions in respect to their constitutional
change. Hansen correlates solidification speed with shrinkage speed as
hot-ductile crack criteria [20]. The microstructure-process-relation by Hunt
[21] quantifies the secondary dendrite arm distance as a function of the
time for solidification within the brittleness temperature range (BTR).
Feurer [22], defines the hot tearing susceptibility by lacking volume due to
solidification shrinkage and continuous feeding. In the crack-susceptible
liquid-solid state, the cracking susceptibility coefficient by Clyne et al. [23,
24] correlates the time needed for liquid flow and mass flow at solidifica-
tion with the time for stress relaxation and tearing between the dendrites
depending on its solid fraction, using the solidification model by Scheil.
The welding pool can be understood as a small and rapidly solidifying
casting between components, or rather as continuous casting, if we consider
 Thermo-Physical Mechanism to Control Cracking in Weldments 67

different types of feeding mechanisms and microstructure-process-relations,

quantifying the secondary dendrite arm distances as a function of solidifi-
cation time during the BTR.
Rappaz-Drezet-Gremaud established the hot tearing initiation model
(RDG criterion), combining local strain with mass balance between
dendrites and solidifying interdendritic liquid [25]. This was applied in
laser beam welding with Al-alloys, with special attention to preventing
solidification cracking at the transient stages at weld start and weld end
[26]. The RDG criterion was extended to steel by Drezet et al. [27], taking
into account the peritectic reaction that transforms δ-ferrite into austenite.
Further developments of this RDG criterion by Rindler et al. [28] release
the assumption that the uniform strain-rate acts troughout the mush state
(BTR).
For all models, the strain-rate acting throughout the mush state (within
the BTR, during existing time of BTR) is the most important factor in the
solidifying weld metal both for local constitutional and thermo-physical
mechanical determination.

Criteria Determined by Structural Mechanisms

As established by Prokhorov [2] and Matsuda [3], in opposite to castings,

the whole welded structural component including weld metal absorb all
thermo-mechanically induced local deformation, strain and strain-rate
during solidification shrinkage within the BTR, which are associated to
the welding-process, for the prevention of hot cracking. According to the
theory of technological strength by Prokhorov [29], solidification cracks
will always occur when a minimum deformation δ[%] and a critical
deformation rate Bcr [%/°C] is reached or exceeded in the BTR. The
deformation rate BBTR is to be understood as modification of the
deformation dεi/dT in the BTR (B=dεi/dT [%/°C]). The amount of
deformation εI corresponds to the sum of thermal shrinkage εT and
external deformation εext (εi = εT + εext).
For easier application, Prokhorov’s more theoretical criteria were
transformed to centreline solidification cracking, defining the critical speed
of cross displacement within the BTR vcr [23, 24], and modified by the
amount of cross displacement changing during time of the existing BTR
[30]. Both criteria are useable for solidification cracking assessment,
balancing the critical value of criteria with the FEA computed welding
process, material and structure induced values of local cross displacement
and their speed.
68 H. Herold, M. Streitenberger

Assessment Concepts

Hot cracking occurs when as well the critical cross displacements during
the existence time of the BTR Δuy min are exceeded by the total cross
displacement Δuy BTR, due to component design and assembly for
welding, and/or the critical speed of cross displacement is exceeded by the
local speed of cross displacement [31, 32, 48].
The welding-process associated local speeds of cross displacement vq
are calculated by FEA with the help of the program ANSYS at discrete
points along the seam during one-side welding of the actual welding
components, taking into account the effects of design and assembly by
dimension of run-outs, tacks and magnetic clamping. The speed of cross
displacement vq in Fig. 3 will maximize systematically when the arc comes
near the plate’s end, and the far side of the weld pool passes from the plate
to the run-out due to the stiffness jump. It increases locally during fusion
of each tack, additionally releasing assembling stresses. These short and
rapid increases of cross displacement with regard to time and versus the
welded seam are the most important factors to initiate centerline
solidification cracking of longer butt weld seams. However, the estimation
of the critical vcr needs an empirical assessment diagram for determination
of the critical speed of cross displacement vcr in correlation with the
applied material and welding process.
The assessment diagram in Fig. 4 offers the complexity of hot
cracking by three sub-diagrams. The assessment diagram for centreline
solidification cracking can be seen on the right side. The multitude of
calculated curves of the weld component and process-associated speed of
cross displacements vq, measured with the help of the welding-process
associated hot-cracking test method, are plotted here as cooling time of
the weld metal versus the change of the measured and calculated cross
displacement ΔU (t) in [mm]. The critical speed of cross displacement vcr
is defined here as the threshold curve, separating the regions with
occurring hot cracking from crack-free regions, established by empirical
approximation. The sub-diagram in the centre shows the correlation with
the hot cracking theory by Prokhorov, setting up internal deformation
versus temperature. The left sub-diagram shows the thermal cycle of the
applied welding procedure.
 Thermo-Physical Mechanism to Control Cracking in Weldments 69

Fig. 3. Distribution of flat displacements (arrows value amount and direction) in

large plates during one-side submerged arc welding (sub-fig. above) and the speed
of cross displacement after passing the plate end by the welding pool vq,
calculated with the help of FEA simulation (sub-fig. middle) in correlation with
the applied welding procedure with taking into account designing and assembling
for welding in principle (sub-fig. below) [49]

The multitude of calculated curves of the weld component and process-

associated speed of cross displacements vq, measured with the help of the
welding-process associated hot-cracking test method, are plotted here as
cooling time of the weld metal versus the change of the measured and
calculated cross displacement ΔU (t) in [mm]. The critical speed of cross
70 H. Herold, M. Streitenberger

displacement vcr is defined here as the threshold curve, separating the

regions with occurring hot cracking from crack-free regions, established
by empirical approximation.
The sub-diagram in the centre shows the correlation with the hot cracking
theory by Prokhorov, setting up internal deformation versus temperature.
The left sub-diagram shows the thermal cycle of the applied welding
procedure.

Fig. 4. Assessment diagram for empirically determining the material- and

welding-process associated speed of cross displacement vcr for critical initiation
of centreline solidification cracking during one-side submerged arc welding by
measuring the component-and welding-process associated speed of cross
displacement vq for thickness of about t=10 mm and welding parameter:
I1 = 730 A, U1 = 32 V, I2 = 680 A, U2 = 38 V, vw = 95 cm/min [48]

The assessment diagram in Fig. 4 is applicable in correlation with FEA

simulation of large components in order to develop new methods of crack-
free design and assembling for welding relevant and large components, as
well as for refining metallurgical effects and modifying welding proce-
dures. However, all variations in welding procedures will change the
location of the threshold curve of the tested welding procedure, thus af-
fecting the material and metallurgical effects as well as the component and
 Thermo-Physical Mechanism to Control Cracking in Weldments 71

assembling effects. This high sensitivity to welding-procedure induced

effects underlines the limited expressiveness of the laboratory test proce-
dures in view of the reliability of the weldment.
The computational procedure of solidification cracking assessment
combines thermal and structural analyses, due to the applied welding
procedure on the structural component of a selected material, with the
implementation of local analysis of the balancing solidification cracking
criteria (modified cross displacement and speed of cross displacement,
existing during time of BTR). Crack initiation is proved by checking
exceeded critical criteria values for each discrete FEA noodle point [33].
Solidification assessment concepts under the use of strain rates are
considerably gaining in importance if the indication of different
assessment concepts and criteria, applied independently of one another to
similar structures (butt-joints and T-joints of the same material) agree in
principle [44, 51] and follow different realization. The FEA model by
Drezet et al. [44] based on the RDG criteria is applied in studying the
influence of laser beam welding on the strain-rate in the BTR by FEA to
prevent solidification cracking at transient stages on weld start and weld
end. Successful optimization of the welding procedure was achieved by
increased laser power at run-in, constant power in linear run and decreased
power at run-out with effective clamping in this case. While [44] realizes
compensating laser beam power along the seam, [51] uses structural design
and assembling for welding in combination with thermal effects by
welding direction and sequence.
The thermo-mechanical conditions for weld-metal solidification
cracking of Al-alloys were modelled in the simulation model by Feng,
Zaharia and David [34], the local stress-strain evolution within the BTR at
the weld centreline solidification crack site having also been calculated.
The combination of heat transfer model with mechanical model and
solidification behaviour of the weld pool results in a concept of a
computation model, which acts as key concept for quantitative assessment
on thermo-mechanical FEA formulae. That means that the microscopic
deformation, which relates directly to the cracking process, is controlled to
a significant degree by the macroscopic deformation process, which obeys
the principles of continuum solid mechanics and heat transfer. The concept
was used to prove the relation between the macroscopic thermo-
mechanical condition and the local effects produced by the trailing edge of
the welding pool due to the welding operation (welding parameter, joint
design, component restraint).
Murakawa et al. [35, 36] developed a Temperature Dependent Interface
Element, as it is not possible to apply a simple thermal elastic-plastic FEA
for analysing deformation and residual stress in welding in the Houldcroft
72 H. Herold, M. Streitenberger

test under the use of fish bone specimens. In their model, they adopted
(crack) opening displacement from elastic-plastic fracture mechanics, con-
sidering the crack propagation as the formation of new surfaces, without
taking into account the effects of temperature on the strength of the inter-
face element. The influence of BTR width and the opening displacement at
the crack were examined, from which followed that the opening displace-
ment must be adjusted together with the BTR width, so that the tempera-
ture-dependent interface element represents the solidification cracking sus-
ceptibility of the material.
Ploshikhin et al. [37] developed a thermo-mechanical and metallurgical
model for the solidification cracking mechanism in welds. These
solidification cracking criteria are based on the tensile strain accumulated
in the segregating inter-granular liquid film at the weld centreline by the
solidified dendrites from the completely solid region of the base material
until cracking appears. The local critical parameter δacc represents the
ability of the segregating liquid film to accumulate strains without tearing,
measured by the micro-structural morphology (by fraction of residual
liquid, primary and secondary dendrite arm spacing). Solidification
cracking will result when the maximum accumulated strain exceeds the
critical value.

Conclusion

The analysis of weldment cracking especially solidification cracking in

welding proved that the crack initiation depends on the superposition of a
critical deformation rate (strain rate, cross displacement, accumulated
strain) within a material-specific temperature range, e.g. brittleness
temperature range (BTR) with intergranular changing yielding capacity
due to constitutional- time-temperature- and shrinking conditions which
exceed the critical value of the defined cracking criteria.
The premises for any hot cracking assessment on welding are defined
equilibrium criteria, quantifying the local processing conditions for crack
initiation of metal-physical associated crack mechanisms of instability at
the moment of passing the BTR. The critical values for weldment
reliability cannot be directly examined in hot cracking tests.
The possibility and validity of solidification crack initiation assessment has
been proved by different assessment concepts and criteria, which were created
independently of one another and applied to similarly structured butt-joints,
T-joints of aluminium alloys, and one-side welding of special structured
components of ferritic steels, and which resulted in a similar tendency.
 Thermo-Physical Mechanism to Control Cracking in Weldments 73

Solidification cracking assessment for welded structural components of

austenitic steel or nickel-base alloys have not been investigated, so far.
If we can analyse the cracking processes due to welding procedure more
in detail, focussing on the local microscopic conditions, then we can also
develop new engineering tools, based on the state of art in actual welding-
process-associated tools, by computational combination of the data of
thermo-physical, mechanical, metallurgical and structural behaviour, with
the aim of prevention of weldment cracking.

References

1. Pellini WSF (1952) Foundry 80:125–133, 192, 194, 196, 199

2. Satoh K, Terasaki T et al. (1980) Changes of root gap during welding in case
of butt weld joints (1st Report). J Japan Weld Soc vol 49 7:478–483 (in
Japanese)
3. Ueda Y, Murakawa H et al. (1992) Simulation of welding deformation for
precision ship assembling (Report I) – In-plane deformation of butt welded
plate. Trans JWRI vol 21 2:125–135
4. Andersson B, Karlsson L (1981) Thermal stresses in large butt-welded plates.
J Thermal Stress 4:491–500
5. Herold H, Streitenberger M, Pchennikov A, Makarov E (1998) Modelling of
one sided welding to describe hot cracking at the end of longer butt weld
seams. (1999) Weld in the World vol 43 2:56–64
6. Herold H, Streitenberger M, Pchennikov A (2001) Prevention of centreline
solidification cracking during one-side welding. Doc IIW-Doc. IX-2000-01
7. Wilken K (1999) Investigation to compare hot cracking tests – Externally
loaded specimen. (1999) Doc IIW IX-1945–99
8. 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)
9. PrEN ISO 17641-2 (Nov. 2003) Destructive tests on welds in metallic
materials – Hot cracking tests for weldments – Arc welding processes – Part
2: Self-restraint tests (ISO/FDIS 17641-2:1003)
10. prCEN ISO/TR 17641-3 (Nov. 2003) 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)
11. Folkhard E (1984) Metallurgie der Schweißung nichtrostender Stähle.
Springer, Wien New York, pp 153
12. Herold H, Schulze S, Streitenberger M, Spieler S (1998) Mikroschädigungen
von NiMo-Werkstoffen. In: Petzow G. (eds) Praktische Metallographie 29
Fortschritte in der Metallographie. DGM Informationsgesellschaft mbH, pp
175–180
74 H. Herold, M. Streitenberger

13. Herold H, Streitenberger M, Pchennikov A (2001) Hot Cracking Theory by

Prokhorov and Modelling of the PVR-test. Weld World vol 45:3/4, 17–22
14. Herold H, Zinke M, Hübner A (2004) Investigation on the use of nitrogen
shielding gas in welding and its influence on hot crack behaviour of high-
temperature resistant fully austenitic Ni- and Fe-base Alloys. IIW-Doc IX-
2110-04
15. Herold H, Pchennikov A, Streitenberger M (2005) Influence of deformation rate
of different tests on hot cracking formation. In: Böllinghaus T, Herold H (eds)
Hot cracking phenomena in welds. Springer, Berlin Heidelberg, pp 328–346
16. Merkblatt DVS 2401 (Aug. 2004) Fracture mechanical assessment of cracks
on welded components. Fachbuchreihe Schweißtechnik Bd. 101. DVS-Verlag
GmbH, Düsseldorf
17. BS 7448 Part 2: Seventh draft (revised) Draft British Standard (1996) Fracture
mechanics toughness test. Part 2: method for determination of KIC, critical
CTOD and critical J valued of weld in metallic materials
18. Blauel JG (1993) IIW-case study collection on the assessment of the
significance of weld imperfection. IIW-Doc. X-1280-93, XE 35-93
19. Campbell J (1991) Castings, Butterworth-Heinemann-Ltd. pp 175–240
20. Hunt JD (1979) Solidification and Casting of metals. The Metals Society
London
21. Hunt JD (1979) Solidification and Casting of metals. The Metals Society
London
22. Feurer U (1976) Gießerei 23:75–80
23. Clyne TW, Davies GJ (1981) Brit Foundry vol 74 4:65–73
24. Clyne TW, Kurz W (1981) Metallurgical trans vol A 12:965–971
25. Rappaz M, Drezet J-M, Gremaud M (1999) A New Hot-Tearing Criterion.
Metallurgical and Materials trans vol 30A 2:449–455
26. Drezet J-M, Mathier V, Allehoux D (2006) The modeling of laser beam
welding of Aluminium alloys with special attention to hot cracking in
transient regimes. In: Cerjak H, Bhadeshia HK, Kozeschnik E (eds) Math.
Modelling of Welding Phenomena 8, presented
27. Drezet J-M, Gremaud M, Graf R Gräumann M (2002) A new hot tearing
criterion for steel. In: Proc 4th International Continuous Casting Conference,
IOM communications, Birmingham, UK, pp 755–763
28. Rindler W, Kozeschnik E, Enzinger N, Buchmayr B (2002) A modifyied hot
tearing criterion for steels. In: Cerjak H, Bhadeshia HK (eds) Math. Modelling
of Welding Phenomena 6, IOM Communication, UK, pp 819–835
29. Prokhorov NN, Prokhorov N (1968) Schweißtechnik (Berlin) vol 19 1:8–11
30. Pchennikov A, Streitenberger M, Herold H (2004) Application and further
development of Prokhorov`s solidification-cracking theory. IIW-Doc. IX-
2160r1-05
31. Herold H, Pchennikov A, Streitenberger M (2004) Assessment of hot cracking
initiation by laboratory test procedures and FEA simulation associated
experimental measurements during welding of large weld components.
Quarterly Journal of Japan Welding Society, vol 22 2:211–217
 Thermo-Physical Mechanism to Control Cracking in Weldments 75

32. Pchennikov A (2005) Entwicklung von Maßnahmen zur Heißrissvermeidung

beim Einseitenschweißen langer Nähte. Diss, Otto-von-Guericke Universität
Magdeburg
33. Pchennikov A, Gruss H, Herold H (2006) Avoidance of hot cracking at
unsteady weld start on laser beam welding of Aluminium. In: Cerjak H,
Bhadeshia HK, Kozeschnik E (eds) Math. Modelling of Welding
Phenomena 8, presented
34. Feng Z, Zaharia T, David SA (1997) On the thermomechanical conditions for
weld metal solidification cracking. In: Cerjak H, Bhadeshia HK (eds) Math.
Modelling of Welding Phenomena 3, The Institute of Materials, Cambridge
UK, pp 114–150
35. Shibahara M, Serizawa H, Murakawa H (1999) Finite Element Method for hot
cracking using temperature dependent interface element. Trans JWRI vol 28
1:47–53
36. Shibahara M, Serizawa H, Murakawa H (2000) Finite Element Method for hot
cracking using temperature dependent interface element (Report II). Trans
JWRI vol 29 1:59–64
37. Ploshikhin V, Prikhodovsky A, Makhutin M, Llin A, Zoch H-W (2005)
Integrated mechanical-metallurgical approach to modelling of solidification
cracking in welds. In: Böllinghaus T, Herold H (eds) Hot Cracking Phenomena
in Welds 1, Springer, Berlin Heidelberg, pp 223–244
Determination of Critical Strain Rate
for Solidification Cracking by Numerical
Simulation

M. Wolf, Th. Kannengießer, Th. Böllinghaus

Federal Institute for Materials Research and Testing, Berlin, Germany

Abstract

Hot crack prevention during welding processing of metallic materials is an

essential prerequisite of component safety. The causes of solidification
cracking can be attributed to the occurrence of metallurgical and ther-
momechanical effects. Even though numerous hot cracking test procedures
have been developed until now, unexpected solidification cracking during
component welding cannot be avoided, since specifically in the hot
cracking test the thermomechanical conditions and the local crack-critical
strain rates around the weld pool may be very different from those in the
component.
In the presented numerical analyses, contrary to the well-known energy
distribution models (according to Goldak, for example), the experimentally
determined weld pool geometry has directly been implemented as a 3D-
function into the numerical simulations and thus conduced to high compu-
tational accuracy.
The investigations were carried out using nickel-base Alloy 602 CA,
since its solidification resistance exhibits a significant dependence on the
shielding gas, which enabled studies of hot crack-critical as well as hot
crack-uncritical material behaviour. Validation was carried out with the
help of the MVT-test which allowed additional variation of the specimen
loading rate. Numerical calculation of the solidification crack-critical lim-
iting temperature versus various welding parameters and the investigated
shielding gases could be performed. In order to provide a transferability of
hot cracking tests to components, critical specimen loading rates were de-
termined and the local crack-critical strain rates in close vicinity of a weld
pool were calculated. It is demonstrated that the local critical strains and
strain rates represent a crack criterion for a transferability.
78 M. Wolf et al.

Introduction

Hot crack prevention during fabrication and processing of metallic materi-

als is an essential pre-requisite of component safety. The causes of hot
cracking can be attributed to the occurrence of metallurgical and ther-
momechanical effects.
In order to define hot cracking, a solidification crack-critical limiting
temperature TSR can be drawn upon according to which solidification
cracks occur beyond 0.5 TM (TM melting point) [1]. Such solidification
cracks appear at a stage where the melt is distributed like a film along
grain boundaries. The precise processes involved in solidification crack
formation are, however, difficult to quantify experimentally. For examin-
ing metallurgical influences on hot crack formation, hot cracking tests us-
ing externally loaded specimens are an obvious choice enabling deliberate
variation of the loading conditions [1, 2].
In order to ensure transferability of hot cracking test result to real
components, it is exactly the action of thermomechanical effects and of
external geometric boundary conditions (restraint intensity, tacks) that
needs to be quantified. [3, 4]. Specifically, local critical strains and
strain rates, respectively, are difficult to determine experimentally, on
account of the high temperatures arising in the immediate vicinity of a
weld pool [5]. Because of the low chances of measured thermomechanical
data collection, numerical simulation of hot cracking is gaining in impor-
tance. In order to take account of the fluid-mechanical behaviour of the
melt in numerical calculations, this study presents a new method of di-
rectly implementing the weld pool geometry in the numerical simulations.
This allows it to use the reconstructed weld pool geometry as a given
function exactly considering the change of the weld pool shape versus
the welding speed. In the next step, numerical MVT-test simulations
will be performed in order to calculate local crack-critical strains and
strain rates, respectively.

Experimental Procedure

Test Materials and Welding Parameters

The investigations were carried out selecting Ni-base Alloy 602 CA (Ma-
terial No. 2.4633) as test material with composition given in Table 1 [6]. It
was available in solution-annealed condition. In preceding MVT- and
PVR-hot cracking tests, the metallurgical effect of N on hot cracking resis-
tance was already identified [7]. It was hence possible, by the use of two
 Determination of Critical Strain Rate by Numerical Simulation 79

different shielding gases, to investigate a hot crack-critical (shielding gas:

pure Ar) as well as a hot crack-uncritical (shielding gas: 1% N2 + 99% Ar)
material behaviour for Alloy 602 CA.

Table 1. Chemical composition of Alloy 602 CA, percent by weight

Ni Cr Fe Mn Ti C Zr Cu Mo Si Al Y
62.1 25.6 9.45 0.07 0.14 0.165 0.1 0.01 0.05 0.04 2.35 0.06

The numerical calculations were verified with the help of the MVT-
Test [8], in which the specimen loading rate was varied additionally [1, 2].
For quantification of the mechanical specimen loading, strain gauges
were applied to the specimens subjected to MVT-testing using various
die radii. The size of MVT-specimens was 100 mm × 40 mm × 10 mm.
The various welding parameters can be seen from Table 2. The tests P1 –
P4 were conducted in combination with the shielding gases Ar as well as
Ar + 1% N2.

Table 2. Welding parameters for TIG-welding

Welding Current Voltage Welding speed Heat input ES

parameter I [A] U [V] vs [mm/s] [kJ/cm]

P1 250 18 5 9.0

P2 182 17 3 10.3

P3 205 17 1.8 19.4

P4 200 17 6 5.7

Numerical Approach

The numerical simulations for the MVT-2-Test were performed using

ANSYS® FEM-software. In order to take account of the mechanical
specimen deformation, special contact and target elements were employed.
The material behaviour within each element was assumed to be homoge-
neous and isotropic. The heat conductivity and the specific enthalpy of Al-
loy 602 CA are largely characterized by a linear temperature dependence.
80 M. Wolf et al.

As temperature-dependent material characteristics for the thermomechani-

cal calculations, the heat expansion coefficient, the E modulus as well as
σ-ε-functions were available. A bilinear stress-strain behaviour was as-
sumed. The tangential modulus for Alloy 602 CA was determined in ten-
sile tests using 2.13 GPa.

Test Results and Discussion

Measurement and Simulation of Mechanical Specimen

Deformation in the MVT-Facility

Figure 1 gives an example of the measured strain history depending on the

cross-beam travel for a bending strain of 2% in the Varestraint as well as
Transvarestraint mode. The cross-beam travel speed was 0.1 mm/s. The
strain increase per millimetre of the traverse travel Δε/ΔH required for
quantifying the mechanical specimen deformation was also determined.
The strain increases for εtot = 2% are given in Table 3.

Table 3. Strain increase per mm of cross-beam travel H (MVT-Test)

Specimen Strain increase per mm of cross-beam travel Δε/ΔH [1/mm]

thickness
10 mm

Bending Measurement Simulation

strain εtot
Varestraint Transvarestraint Varestraint Transvarestraint

2% 0.040 0.040 0.041 0.042

Calibration and validation of the numerical calculations accompanying

the MVT-Test was carried out using the Transvarestraint model repre-
sented in Fig. 2.
 Determination of Critical Strain Rate by Numerical Simulation 81

Fig. 1. Simulated and measured strain in the Varestraint and Transvarestraint

mode (MVT), specimen thickness: 10 mm, die radius RM = 250 mm

Fig. 2. Numerical model for the Fig. 3. Strain distribution in x direction

MVT-Test (Transvarestraint mode) after bending (Transvarestraint mode,
RM = 250 mm)

Figure 3 represents the numerically calculated strain distribution in an

MVT-specimen after bending around a die with the radius RM = 250 mm.
This corresponds to a bending strain of 2%. Figure 1 and Table 3, respec-
tively, give a comparison between the results of simulation and measurement
by the example of 2% bending strain which shows a deviation of < 5%.
82 M. Wolf et al.

Determination of Weld Pool Geometry and Calculation

of Temperature Field

Exact consideration of weld pool geometry and heat distribution around

the weld pool is essential for accurate numerical calculation of solidifica-
tion cracking. In order to calculate a weld pool geometry as much real as
possible, the exact shape and height of the distribution functions must be
matched to a real arc welding process on an iterative trial and error basis.
Doing this using energy distribution models as described in [9, 10] would
be extremely laborious. Therefore, the weld pool geometry was recon-
structed with the help of transverse sections and of the weld end crater ge-
ometries and was directly implemented in the numerical simulations.
As an analytical basic equation, Eq. (1) by Prokhorov [11] was adopted
to describe the 3D-weld pool geometry (leading edge or trailing edge of
the weld pool). Equation (1) was enlarged in order to differentiate geome-
try and shape parameters (Table 4). Whereas the geometry parameters rep-
resent the dimensions of a weld pool with the values of maximum width,
maximum depth and maximum length, the shape parameters detemine the
weld pool shape in the symmetry plane, the surface plane and the cross-
sectional plane.

§­ 1
½ ·
μ
1
¨ °ª ºω ° ¸
ª § τ
· º ¨ °« φ » ¸
«1 − ¨ z ¸ » ¨ °« § x ·
¨ ¸ »
°
¸
« ¨© Lx , SB ¸¹ » ¨B ¸ °
¨ °« » ° θ¸
T ( z , x) = − H SB ⋅ ¬ ¼ © SB ¹
⋅ ¨ ®«1 − » − Δ¾ + Δ ¸
(1 − Δ )θ + Δθ ¨ °« ª ν η
º
φ
» ° ¸ (1)
¨ °« « §¨ z ·¸ » » ° ¸
¨ °« «1 − ¨ L ¸ » » ° ¸
¨¨ °«¬ ¬ © x , SB ¹ ¼ »¼ ° ¸¸
©¯ ¿ ¹
 Determination of Critical Strain Rate by Numerical Simulation 83

Table 4. Geometry and shape parameters for Eq. (1)

Symbol Definition of the geometry parameters
HSB Maximum depth of the weld pool
Lx,SB Maximum length of a weld pool segment
BSB Maximum width of the weld
Definition of the shape parameters
μ, τ Shape of the crystallization front in the symmetry plane of the weld
η, ν Shape of the crystallization front in the surface plane
φ, ω Shape of the crystallization front in the cross-sectional plane
Δ, θ Factors for the determination of turning point coordinates

Figure 4 gives an example of an analytically calculated three-dimensional

weld pool trailing edge geometry.

Fig. 4. Parameters and three-dimensional function of a weld pool geometry calcu-

lated from them

The geometry parameters BSB, Lx,SB and HSB were determined by meas-
urement of the weld end crater length, penetration depth and weld width.
The method of directly implementing the weld pool geometry in the nu-
merical simulations is particularly suitable with regard to the defined
specimen geometries and welding conditions in MVT-Tests. It was, for
example, possible to generate reproducible weld pool geometries by varia-
tion of test parameters such as cross-beam travel speed vtrav or total strain
εtot without the need of geometry and shape parameter matching [12, 13].
In order to determine the geometry and shape parameters, TIG-remelt
welds were produced on MVT-specimens using the parameters given in
Table 2.
84 M. Wolf et al.

Fig. 5. Weld end crater (left hand side) and transverse section (right hand side) of
a TIG-remelt weld of Alloy 602 CA; welding parameters P1; shielding gas:
Ar + 1% N2

A raster was projected with a grid spacing of 0.2 mm (Fig. 5). This
raster enabled light-microscopical analyses of the weld pool geometry pa-
rameters for the individual welding parameters P1–P4 (penetration depth
HSB, weld width BSB, total weld pool length Lx,SB = LV,SB + LR,SB). For
shape parameter determination, the coordinates of the weld end crater
functions and of the fusion line functions were measured in the transverse
section. Figures 6 and 7 show a comparison between the measured coordi-
nates and each matched curve by respective examples relating to the weld-
ing parameters P1 with Ar + 1% N2, for both the crater trailing edge and
the fusion line. The weld pool shape determined by this method was de-
pendent on the shielding gas.

Fig. 6. Comparison of the crater Fig. 7. Comparison of the fusion line

trailing edge function for P1 with function in the transverse section for
Ar + 1% N2 P1 with Ar + 1% N2
 Determination of Critical Strain Rate by Numerical Simulation 85

Figure 8 exemplifies a reconstructed weld pool trailing edge geometry

for P1 with Ar + 1% N2.
The analytical three-dimensional weld pool geometry function was im-
plemented in the numerical model (see Fig. 9). This was accomplished for
each time step by a complete selection of the nodes which are existent in
the weld pool geometry at the point in time ti. The movement of the heat
source was subdivided into a finite number of time steps using a time
increment Δt.
The selected nodes representing the weld pool geometry were subse-
quently set to the liquidus temperature Tliq (Tliq = 1400°C).
The numerical temperature field calculations performed were validated
by temperature measurement close to the fusion line. Figure 10 shows a
comparison between simulation and measurement at distances of 2.7 mm
and 1.0 mm, respectively, from the fusion line for P2 with Ar + 1% N2.
The maximum deviation in the transition area of temperature rise is ap-
proximately 15%.

Fig. 8. Three-dimensional representation of the determined weld pool geometry

(trailing edge) using the geometry parameters for P1 with Ar + 1% N2
86 M. Wolf et al.

Fig. 9. Implementation of the weld Fig. 10. Temperature history of

pool geometry function (Tz,x) in the simulation/measurement at distances
numerical model of 2.7 mm und 1 mm, resp., from the
weld for P2 with Ar + 1% N2

Fig. 11. Numerically calculated temperature distribution in the cross-section (top)

and at the surface (bottom) for P2 with Ar + 1% N2
 Determination of Critical Strain Rate by Numerical Simulation 87

In Fig. 11 shows an additional projection of the calculated temperature

distribution to the transverse section and to the crater at the end of the
weld. It is found that the weld pool geometry used in the simulation repro-
duces with adequate exactness the real weld pool geometry in the cross-
section as well as at the surface.

Numerical Calculation of Local Strain and Strain Rates

Before presenting the numerical simulation results of the MVT-Test, the

numerically determined results of the thermomechanical calculations were
verified.
In order to provide transferable solidification crack criteria from the
MVT-Test, the local critical strain rates must be determined taking account
of the critical specimen loading rates.

Fig. 12. Strain distribution for P1 Fig. 13. Strain distribution for P4
with Ar, vtrav,crit = 0.05 mm/s with Ar, vtrav,crit = 0.1 mm/s

Figure 12 illustrates as an example the strain distribution transversely to

the welding direction (x-direction) for P1, 2.3 s after the onset of mechani-
cal loading at the point in time t = 10.3 s. The specimen loading rate for P1
(Ar) was vtrav,crit = 0.05 mm/s. It is clearly seen from Fig. 12 that the strains
accumulate at the weld axis behind the weld pool due to the prevailing
high temperatures. Figure 13 shows by comparison the strain distribution
for P4 at the point in time t = 9.4 s. Strain accumulation is observed at the
weld axis directly behind the weld pool. Contrary to P1, excessive
strain concentrates on a considerably narrower zone behind the weld
88 M. Wolf et al.

pool, because increasing welding speed (P4) involves a more tapered weld
pool trailing edge geometry [14, 15].
This example shows the advantages of direct implementation of the
weld pool geometry in the numerical simulation. Geometric influences of
the weld pool geometry on the local strain rates can thus be adequately
considered in the numerical calculation.
Further quantification of the relationship between the global specimen
loading rate vtrav and the local strain rate was performed by numerical
simulation of four different specimen loading rates between 0.05 mm/s and
0.4 mm/s for the parameter set P1, where 0.05 mm/s represents the critical
specimen loading rate vtrav,crit.

Fig. 14. Determination of the solidification crack-critical limiting temperature TSR

 Determination of Critical Strain Rate by Numerical Simulation 89

In order to determine the solidification crack-critical limiting tempera-

ture TSR [1], the weld pool geometries were reconstructed from the craters
at the weld end and from the transverse sections for all experiments
(Fig. 14) [12]. With the help of the weld pool geometries reconstructed by
this method and of numerical simulations, the temperature distribution in
the immediate vicinity of the weld pool was exactly calculated. Further-
more, the maximum solidification crack lengths were experimentally as-
certained with the help of the MVT-Test using rapid loading and related to
the calculated temperature distributions. As a result it was found that the
crack length in the temperature field calculation corresponds to the dis-
tance between the liquidus isotherm and the solidification crack-critical
limiting temperature TSR (Fig. 14c).
Figure 15 depicts the strain history versus the specimen loading rate
with additional plotting of the temperature history at the position of the
crack origin. The temperature history shows the transition from the melting
condition at 1400°C to the cooling phase. In investigations carried out by
[12] it was found for P1 with Ar 4.8 that the solidification crack-critical
limiting temperature TSR is 1195°C and that no solidification cracks occur
below this temperature. With increasing heat input ES, there is a significant
rise in the solidification crack-critical limiting temperature. Accordingly,
the area between the liquidus temperature and TER represents the solidifica-
tion crack-critical temperature interval (STI).
The local strain functions depending on the time were derived for de-
termining the local strain rate according to the time. The functions ob-
tained by this procedure are represented in Fig. 16. As can be seen from
Fig. 16, a maximum level of local strain rate occurs in the solidification
crack-critical temperature interval. This can be attributed to the fact that
the material directly behind the weld pool exhibits the highest temperature
in the continuum-mechanical condition. It must be presumed furthermore
that there is a pronounced notch effect directly behind the weld pool due to
the weld pool geometry.
The calculated values for the local strain rates can be seen from Table 5.
For comparison, the calculated global strain rates are also listed in Table 5
[12].
It is found from Table 5 that the local strain rates εloc for the welding
parameter set P1 with pure Ar are higher by the factor of 2.5–3.3 than the
global strain rates ε glo . Thus, the local strain rates in the MVT-Test differ
appreciably from the global strain rates. Note here the largely proportional
relationship between specimen loading rate, global strain rate and local
strain rate. From this it follows that exactly the critical specimen loading
90 M. Wolf et al.

rate describes the solidification cracking resistance of materials linear-

quantitatively. This shows that the local critical strains and strain rates can
be a crack criterion for component transferability.

Fig. 15. Local strain versus speci- Fig. 16. Local strain rate versus speci-
men loading rate for P1 with Ar 4.8 men loading rate for P1 with Ar 4.8

Table 5. Local strain rate depending on the global strain rate for P1
vtrav global strain rate local strain rate εloc
εglo [1/s] εloc [1/s] εglo
0.50 0.020 0.065 3.3
0.25 0.010 0.031 3.1
0.10 0.004 0.013 3.3
0.05 0.002 0.005 2.5

Summary and Conclusions

Hot crack prevention during welding processing of metallic materials is an

essential prerequisite of component safety. The causes of solidification
cracking can be attributed to the occurrence of metallurgical and ther-
momechanical effects. The conclusions reached are the following:

• A new method could be developed which was used to reconstruct the weld
pool geometry from transverse sections as well as from the crater of the
weld end and to incorporate it as a three-dimensional analytical function
into numerical simulation. The comparisons between numerical simulation
and the measured temperature distributions show high correlation.
 Determination of Critical Strain Rate by Numerical Simulation 91

• Validation was carried out using the MVT-hot cracking test in which the
specimen loading rate was varied additionally. The solidification crack-
critical limiting temperature could be calculated for Alloy 602 CA de-
pending on various welding parameters and on the investigated shield-
ing gases.
• In view of a transferability of hot cracking tests to components, critical
specimen loading rates were determined and the local crack-critical
strain rates in the immediate vicinity of a weld pool were calculated.
• It could be demonstrated that the local critical strains and strain rates
can be a crack criterion for a transferability.

References

1. Hemsworth B, Boniszewski T, Eaton NF (1969) Classification and definition

of high temperature welding cracks in alloys. Metal Construction and British
Welding Journal 1, pp 5–15
2. Wolf M, Kannengießer T (2005) Moderne Prüfverfahren zur Bestimmung der
Heißrisssicherheit beim Schweißen von hochlegierten Stählen und
Nickelbasislegierungen. DVM Report 641, In: Werkstoffprüfung 2005, ISSN
1861-8154, pp 419–426
3. Kannengiesser T, McInerney T, Florian W, Böllinghaus T, Cross CE (2002)
The Influence of local weld deformation on hot cracking susceptibility. In:
Mathematical Modelling of Weld Phenomena 6, Book 0784 Maney Publish-
ing, ISBN: 1-902653-56-4, pp 803–818
4. Kannengiesser T, Cross CE (2006) Effect of tack placement on local weld
displacement and solidification cracking during arc welding of aluminium al-
loy 6083. In: Proceedings to IIW conference, 21.-22.11.2006, Bangkok, Thai-
land, pp 480–491
5. Matsuda F, Nakagawa H, Nakata K, Kohomoto H, Honda Y (1983) Quantita-
tive evaluation of solidification brittleness of weld metal during solidification
by means of in-situ observation and measurement. Transactions of JWRI 12
(1), pp 65–72
6. Corporate Publication (2001) ThyssenKrupp VDM, Werkstoffblatt 4137
Nicrofer 6025H/HT – Alloy 602/602 CA, Edition March
7. Zinke M, Hübner A, Herold H, Hoffmann T (2003) Erhöhung der
Heißrisssicherheit von hochwarmfesten Ni-Basislegierungen durch die
Anwendung stickstoffhaltiger Schutzgase beim Schweißen. ISBN: 3-87155-
683-1, DVS-Report 225, pp 249–256
8. Herold H, Pshennikov A, Hübner A, Slyvinski A, Krafka H (2001) Was sagen
die Heißrissprüfungen mit dem PVR- und MVT-Verfahren über die
Schweißbarkeit aus. DVS-Report 216, pp 255–260
9. Goldak J, Chakravarti A, Bibby M (1984) A new finite element model for
welding heat sources. Metallurgical Transaction B (15B), June, pp 299–305
92 M. Wolf et al.

10. Wei Y, Dong Z (2005) Simulation and predicting weld solidification cracks.
In: Hot Cracking Phenomena in Welds, Springer-Verlag Berlin Heidelberg,
ISBN: 3-540-22332-0, pp 185–222
11. Prokhorov NN, Prokhorov NN (1969) A general equation for the surface of
the solidification front during welding. Svar. Proiz. 8, pp 1–4
12. Wolf M (2006) Zur Phänomenologie der Heißrissbildung beim Schweißen
und Entwicklung aussagekräftiger Prüfverfahren. Dissertation, Verlag für
neue Wissenschaft GmbH, ISBN-10: 3-86509-599-2
13. Wolf M, Schobbert H, Böllinghaus T (2005) Influence of the weld pool geome-
try on solidification crack formation. In: Hot Cracking Phenomena in Welds,
Springer-Verlag Berlin Heidelberg, ISBN: 3-540-22332-0, pp 243–268
14. Katayama S (2000) Solidification phenomena of weld metals (1st report).
Characteristic solidification morphologies, microstructures and solidification
theory. Welding International 14 (12), S. 939–951, 2000
15. Hunziker O, Dye D, Reed RC (2000) On the formation of a centreline grain
boundary during fusion welding. Acta Materialia 48, pp 4191–4202
 Part II
Solidification Cracking of Ferrous
and Nickel-Base Alloys
Classification and Mechanisms of Cracking
in Welding High-Alloy Steels and Nickel
Alloys in Brittle Temperature Ranges

K.A. Yushchenko, V.S. Savchenko

E.O. Paton Electric Welding Institute, Kiev, Ukraine

Fabrication of structures from stable-austenitic high-alloy steels, including

stainless steels, and nickel alloys may involve much difficulties caused by
their high sensitivity to cracking during welding and during short- or long-
time heating.
Cracks during the welding process may be formed in different zones of
a welded joint, depending upon the chemical composition and thermal-
physical properties of the materials welded, as well as upon the thermal-
deformation conditions within the weld zone and welded joint as a whole
(Figs. 1, 2).
The most common classification of cracks that may be formed in stable-
austenitic welds and heat-affected zones is as follows:
• solidification hot cracks;
• ductility dip cracks.

Fig. 1. High-temperature ductility of metal with temperature ranges of formation

of characteristic types of hot cracks [1]
96 K.A. Yushchenko, V.S. Savchenko

Fig. 2. Schematic view of the location of solidification cracks in welds

According to this classification, cracks in the HAZ metal can be of the

following types:
• liquation or segregation hot cracks;
• ductility dip cracks.
Given a wide variety of types of hot cracks and, hence, the mechanisms
of their formation, one of the main tasks is to identify and classify the
cracks. Depending upon the location and conditions of formation of
cracks, they can be classified by external attributes, as well as by using
modeling, metallography and fractography examinations.
Metallography is used to reveal internal structure of the weld metal dur-
ing solidification and subsequent cooling, as well as character of location
of hot cracks relative to the microstructure during solidification. It is a well
known fact that constitutional supercooling ahead of the solidification
front, due to a difference in solubility of impurity elements in solid and
liquid states of metal, results in a dendritic structure (Fig. 3) that forms
interlayers in the dendrite spacings with a decreased solidification tem-
perature. The process of cooling from the solidification temperatures (Ts)
is accompanied by migration of high-angle grain boundaries in the solidi-
fied metal, which initially, during solidification, occupied the place of
inter-dendrite boundaries (Fig. 4).
 Classification and Mechanisms of Cracking 97

Fig. 3. Dendritic structure in stable-austenitic weld

In the high-temperature region of the brittle temperature range (BTR)

(Fig. 1) the solidification cracks In the high-temperature region of the
propagate, as a rule, along the solidification subgrain boundaries or solidi-
fication grain boundaries (Fig. 5) [2], whereas in the low-temperature re-
gion of BTR the solidification cracks propagate along the migrated grain
boundaries [2].

Fig. 4. Microstructure of weld metal with migrated boundary [2, 3]

98 K.A. Yushchenko, V.S. Savchenko

Fig. 5. Structure of hot solidification crack in BTR

Further cooling below TS leads to formation of a thermodynamically

stable structure consisting of macro grains, with an identical crystallo-
graphic intragranular structure. The neighbouring macro grains are con-
fined by the high-angle boundaries, and they are identical to grains in the
base metal, which act as a substrate in the beginning of solidification of
grains (dendrites) near the fusion line (epitaxial solidification). The ductil-
ity dip cracks initiate in welds with a stable-austenitic structure mostly at
the migrated grain boundaries (Fig. 6).
Analysis of structure of the fracture surface resulting from cracking is
another characteristic method, in addition to metallography, which can be
used to classify the cracks and, therefore, reveal the mechanisms of their
formation.
Examinations of the fracture surface in solidification cracking occurring
in the solid-liquid state, according to the classification [3], show that this
surface can be of the type of D, D+F or F, depending upon the temperature
at the fracture location (Fig. 7).
 Classification and Mechanisms of Cracking 99

Fig. 6. Ductility dip crack

1 – crack;
2 – grain boundary

As found by the examinations, surface D that copies the shape of the

secondary dendrite arms is formed in a region of the highest temperatures
of the solidification crack range, when intergrowth of the dendrites pre-
sents a barrier to penetration of liquid metal from the weld pool to inter-
dendritic positions.
The fracture surface becomes smooth and flat (type F) (Fig. 7c) with
formation and migration of the boundaries. Surface D+F is formed at an
intermediate position of the migrating boundary (Fig. 7b). It can be con-
cluded from summarizing the fractography results obtained for the surfaces
of solidification cracks [2, 3, 4], that liquid planes may exist over the entire
brittle temperature range, including zones D, D+F and F, i.e. not only in
the interdendritic regions (zone D) but also at the migrated boundaries
(zone F). These conclusions, undoubtedly, require verification, as no co-
herent suggestion is available that could explain why a boundary migrated
to a new position remains liquid. In our opinion, investigation of strength
and ductility characteristics of metal within a range of subsolidus tempera-
tures, including the zero strength and zero ductility characteristics, can
help to clarify an extremely important issue of a real temperature at the end
of complete solidification, Ts, and fracture of the weld metal, including in
the presence of both the remaining matrix solution in solidification and
other types of liquid interlayers. This will make it possible to get an insight
into a very complicated multifactor mechanism of the formation of cracks
in a range of solidification temperatures. Using criteria Tz.d. (zero ductility
temperature) and Tz.s. (zero strength temperature) makes it possible to al-
low for the effect of thermal-deformation conditions on the formation of
cracks (Fig. 8).
100 K.A. Yushchenko, V.S. Savchenko

Fig. 7. Structure of solidification crack surface

Fig. 8. Schematic of temperature dependence of variations in strength and ductil-

ity of metallic materials in melting temperature range
1 – ductility variation curve ψ=f(T);
2 – grain body strength variation curve σ=f(T);
3 – curve of variations in strength of solid body with liquid interlayer σ=f(T);
Tz.d. – temperature of transition from intercrystalline to intracrystalline fracture
Tz.s. – temperature of zero strength
 Classification and Mechanisms of Cracking 101

Experimental studies of zero strength and zero ductility characteristics

of high-alloy stable-austenitic cast steel (Fig. 9a) [5] and heat-resistant cast
nickel alloy (Fig. 9b) [5], as well as fractograms of the fracture surfaces in
tensile testing, show that the zero ductility temperature corresponds to a
state before complete solidification of metal takes place. At the same time,
it can be shown that there exists a temperature range in which metal, still
being in the solid-liquid state during cooling, has certain strength. The zero
strength temperature corresponds to the condition when the majority of
metal is in the solid-liquid state, and surface tension forces are insufficient
to compensate for stresses in the welded joint as a whole.
The generalizing scheme that reveals the mechanism of formation of so-
lidification cracks in welds with a stable-austenitic structure is shown in
Fig. 10.
According to this mechanism, hot cracks are formed in BTR, the upper
bound of which is below TL and corresponds to the temperature of forma-
tion of the dendrite frame, which hinders penetration of liquid metal from
the pool to the interdendritic regions. This is a zero strength temperature
(Tz.s.), and it can be reliably determined using specialized equipment.
Given that liquid metal is present at the subgrain boundaries, the crack sur-
face in this case corresponds to type D (Fig. 7). The lower bound of BTR
corresponds to the temperature of solidification of liquid boundary inter-
layers, and it is determined by temperature Tz.d.. It should be emphasized
that the possibility of migration of liquid grain boundaries is confirmed by
[6], as well as by our investigations (Fig. 11).
As the weld metal cools down and all its structural components finally
solidify, ductility of the weld metal dramatically increases (in a range of
1250-1000oC), amounting to 50–70% or higher.
There are no hot cracks in this temperature range, owing to a high duc-
tility of metal (Fig. 1). They reappear during further cooling in a range of
moderate temperatures (0.4–0.7 Tmelt) in a region of decreased ductility of
metal. Such cracks are classified as ductility dip cracks.
The character of fracture is micro brittle and intergranular in the pres-
ence of traces of micro plastic deformation on the fracture surface. Cracks
may initiate both directly during the process of cooling of the welded joint
to room temperature, and (after a certain holding of the joint up to the
point of its service) during service.
It should be emphasised that many structural materials having a face-
centred cubic (fcc) lattice are sensitive to embrittlement in this temperature
range, including:
• stable-austenitic single-phase high-alloy iron-base steels;
• nickel and nickel alloys;
102 K.A. Yushchenko, V.S. Savchenko

b
Fig. 9. Variations in strength and ductility: (a) – high-alloy cast metal with austenitic
structure [5]; (b) – heat-resistant nickel alloy
 Classification and Mechanisms of Cracking 103

Fig. 10. Flow diagram of the process of solidification of metal with stable-
austenitic structure in the presence of the concentration overcooling zone (a),
variations in character of dendrite growth in bi-crystalline metal (b) [1], and varia-
tions in strength and ductility of metal in solidification (c) [5]: Tz.d. – zero ductility
temperature; Tz.s. – zero strength temperature. Characteristics of structure of hot
crack surface [6]: D – dendritic; F – flat; D+F - transient
104 K.A. Yushchenko, V.S. Savchenko

Fig. 11. Microstructure of welded joint in stable-austenitic steel with migrated

boundary (shown by arrows)

• aluminium and aluminium alloys;

• copper and copper alloys;
• cobalt, etc. (Figs. 12, 13, 14, 15).
Therefore, the above ductility dip temperature range at (0.4–0.7) Tmelt is
characteristic of a whole class of structural polycrystalline materials with
an austenitic structure, which means that the causes of their embrittlement
are of a common nature and fundamental character.
Results of numerous studies allow generalisation of the effect of main
parameters on the processes of embrittlement in this ductility dip range.
Embrittlement is aggravated under the following conditions:
• presence of coarse-grained structure in the weld or HAZ metal;
• presence of segregation elements in the form of impurities, uncontrol-
lable contaminants, such as [O], S, P or surface active B, Se, etc.;
• effect on metal by external energy sources, penetrating energy, in-
cluding neutron radiation;
• decrease in the rate of deformation and its localisation.
 Classification and Mechanisms of Cracking 105

Fig. 12. Temperature dependence of variations in ductility of steel AISI310 with

standard content of impurity elements (S ≤ 0.02%, P ≤ 0.035%)

Fig. 13. Temperature dependence of variations in ductility (A5) and strength Rm of

99.99% pure copper with oxygen content at a level of 0.01% [7]
106 K.A. Yushchenko, V.S. Savchenko

Fig. 14. Temperature dependence of variations in ductility of alloy Al-5Mg with

different content of Na (ppm) [8]

Fig. 15. Temperature dependence of variations in ductility RA of low-alloy high-

strength steel with different content of hydrogen (cm3/100 g of metal) [9]

So far, no commonly accepted opinion exists on the nature of embrit-

tlement or formation of cracks at high, moderate and comparatively low
temperatures (room). We believe that consideration of sensitivity of a
welded joint to cracking with time should not be confined only to the proc-
ess of cooling of the weld to room temperature during solidification. And
on the whole, classification of cracks into hot and cold ones is very relative
and has no clear physical explanation. At the same time, it can be stated
that the presence of stresses and strain zones has a decisive effect on the
 Classification and Mechanisms of Cracking 107

processes of embrittlement and cracking of a welded joint. Besides, the de-

creased rate of deformation, which intensifies brittle fracture, suggests that
the creep mechanism contributes to the processes under consideration.
In this case, the use of mathematical analysis for investigation of creep
processes is very important in terms of specifying mechanisms of general
and local plastic deformation for the class of materials considered and for
formation of hot cracks. Comparing mathematical generalisation of the re-
sults obtained with known data [10] on the mechanisms of plastic deforma-
tion during high-temperature creep, and assuming similarity of the proc-
esses of creep and deformation in welds allows a conclusion on the nature
of this type of the hot cracks in the welds. A large amount of experimental
data on the effect of stresses and temperature on the rate of steady-state
creep has been accumulated up to now [11].
At very low stresses, the creep rate changes in a linear manner from
stress ε∼σn, where n = 1. This dependence is characteristic of the tempera-
tures close to solidus, and allows description of the process of diffusion
creep at temperatures above 0.9 Tmelt, where slip along the grain bounda-
ries is the main form of deformation (Fig. 16). In a range of lower tem-
peratures the main mechanism of deformation is movement of dislocation.
Index “n” in this case amounts to high values [11].

Fig. 16. Schematic of the main types of steady-state creep [12]:

(a) – diffusion; (b) – dislocation
108 K.A. Yushchenko, V.S. Savchenko

Variations in the creep rate with the values of stresses at fixed values of
temperature for stable-austenitic steel AISI310 were evaluated to find
causes and determine the main mechanism of creep in the ductility dip
temperature range. Experimental points obtained as a result of experiments
are shown in Fig. 17.
Exponential dependence of variations in the creep rate for the selected
temperature of 650oC is proved by the straight line describing variations of
ε with σ in the selected coordinate system.
Empirical dependence of the steady-state creep rate upon the stresses
was plotted on the basis of the experimental results using the following
formula:
Q − mσ
ε = A ⋅ e RT

where:
ε is the variation in the deformation rate;
A is the constant;
Q is the creep activation energy, kJ/mole;
σ is the stress at a given temperature, MPa;
R is the gas constant, J⋅mole–1⋅K–1;
m is the exponent, m=f(T);
T is the temperature, K

Fig. 17. Dependence of the rate of steady-state creep of steel AISI310 upon
stresses at test temperature of 650oC. Critical strain rate, below which intergranu-
lar fracture is probable, is shown by the mark in straight line
 Classification and Mechanisms of Cracking 109

As seen from the calculations, the creep rate can be described by the fol-
lowing equation:

46000 mσ

ε = 1.53 ⋅ 1013 ⋅ e 8.31⋅T ⋅ e 8.31⋅T

The above equation of steady-state creep describes the deformation con-

trolled by transverse slip of the dislocations [10].
As shown experimentally, variations in the deformation rate has a strong
effect on the character of fracture. The portion of a brittle intergranular
component of macro fracture grows to 100% with decrease in the deforma-
tion rate.
Appearance of the fracture surface for two characteristic rates is shown
in Fig. 18. Analysis of the fracture surface suggests existence of a brittle
intergranular fracture with traces of micro plastic deformation developing
at the decreased deformation rates (Fig. 18b), in contrast to fracture in the
bulk of grains at a comparatively high deformation rate (Fig. 18a).

Fig. 18. Fractograms of fracture surfaces of steel AISI310 specimens at steady-state

creep rate: (a) –3.6.10–3 s–1; (b) –6.0.10–5 s–1
110 K.A. Yushchenko, V.S. Savchenko

Fractography also showed the presence of the slip planes escaping to the
fracture surface (Fig. 19a), which proves existence of the intragranular
plastic deformation and the dislocation mechanism of creep in the tem-
perature range studied. It also can be concluded on the basis of the data ob-
tained that the decisive role in the fracture process is played by a tough
(liquid) interlayer, the traces of the deformation of which can be seen in
Fig. 19b.

Fig. 19. Microplastic deformation in tough component at grain boundaries:

(a) – an escape of the slip planes to the fracture surface;
(b) – the deformation of a tough (liquid) interlayer.
The traces of deformation are shown arrows
 Classification and Mechanisms of Cracking 111

Auger spectroscopy of element composition of the macro brittle fracture

surface of specimens deformed at 650°C shows that the surface is rich in
sulphur (up to 2.5–3.0 wt.%). Its content stabilises at an ultimate depth
(down to 300 nm) from the fracture surface.
Similar data on considerable enrichment of boundaries can be found in
study [8]. The authors show that the concentration of sodium at a brittle in-
tergranular fracture of aluminium alloy Al-5%Mg grows by several orders
of magnitude (Fig. 20).
This enrichment cannot be realised over a short period of time due to
equilibrium segregation and diffusion. In this case the most probable en-
richment mechanism is an intensive mass transfer of impurity elements
with mobile dislocations [4].
Modelling of thermal-deformation processes in welded joints on
stainless steels (Fig. 21a) and heat-resistant nickel superalloys (Fig. 21b)
shows that there are real prerequisites for the effect of elastic stresses and
localisation of plastic strains on realisation of the processes of redistribu-
tion of impurities at the grain boundaries and their embrittlement.

Fig. 20. Hot ductility and sodium concentration on grain boundaries of Al-5%Mg
alloy as a function of sodium content [8]
112 K.A. Yushchenko, V.S. Savchenko

Fig. 21. Kinetics of variations in stress-strain state in welding of stainless steel (a)
and nickel alloy (b) at distance of 0.5 mm from fusion line [13]

Conclusions

1. Analysis of the character and mechanisms of fracture for a welded joint

metal with an fcc lattice within a temperature range from ambient to so-
lidification ones proved the presence of characteristic temperature
ranges, where cracks were formed, and where ductility dips of the mate-
rial (weld, HAZ) were fixed.
2. According to the cracking classification suggested, these temperature
zones can be characterised by the following fracture mechanisms start-
ing from the TL-TS temperature range (the first ductility dip zone):
The cracks are formed by the mechanism of metal fracture in the inter-
crystalline interlayers and on the fracture surface. So, these zones are
defined as types D, D+F and F [3].
3. The second and third DDC zones, which usually relate to the ductility
dip cracking region, are formed in a temperature range of 0.4–0.7 Tmelt,
or in a range of Tservice, depending upon the degree of metal purity. One
or two mechanisms of intergranular fracture may coexist in this range.
The first mechanism is characterised by formation of impurity at the
grain boundaries. Impurity in the form of metal compounds (S, B, P, O, Na)
 Classification and Mechanisms of Cracking 113

are formed due to diffusion of impurities to the grain boundaries, which

is enhanced by plastic deformation.
The second mechanism is when segregation of elements occurs at the
grain boundaries decreasing their cohesive resistance. Fracture in the
first zone occurs by the liquation mechanism, and in the second zone by
the segregation mechanism.
The third zone shows up at comparatively low temperatures, provided
that metal contains chemical elements that have a diffusion mobility in
this temperature range (H, He).
Therefore, fractures along the grain boundaries of a liquation charac-
ter, segregation character, and creep character may coexist at tempera-
ture of 0.7 Tmelt – Tservice.

References

1. Hemsworth W, Boniszewski T, Eaton NF (1969) Classification and definition

of high temperature welding cracks in alloys. Metal Construction and British
Welding Journal 1, N25:5–16
2. Ramires AI, Lippold JC (2004) High temperature behavior of Ni-base weld
metal, part II – Insight into the mechanism for ductility dip cracking: Materi-
als Science and Engineering, A380, pp 245–258
3. Matsuda F, Nakagawa H, Ogata S, Katayama S (1978) Fractographic Investi-
gation on Solidification Crack in the Varestraint Test of Fully Austenitic
Stainless Steel: Studies on Fractography of Welded Zone (III), Transactions of
JWRT 7, W1, pp 59–70
4. Yushchenko KA, Savchenko VS, Starushchenko TM (2004) The Role of Seg-
regation of Oxygen in Welding Alloys of the INVAR Type. Hot Cracking
Phenomena in Welds, Berlin, pp 59–70
5. Rogberg B (1983) An Investigation on the Hot Ductility of Steels by Perform-
ing Tensile Test on “in Situ Solidified” Samples. Scandinavian Journal of
Metallurgy 12:51–66
6. Radhakrishnan B, Thompson RG (1993) Kinetics of Grain Growth in the
Weld Heat - Affectd Zone of Alloy 718: Metallurgical Trans A 24A, pp
2773–2785
7. Slyško P (1988) Decrease of plasticity in the range of critical recrystallization
temperature of fcc metals (in Czech). 2:129–146
8. Horikava K, Kuramoto S, Kauno M (2000) Sources at a trade amount at so-
dium, and its effect on hot ductility an Al-5 mass%Mg alloy: Light Metals
Review, vol 7, pp 18–23
9. Riecke E (1978) Wasserstoff in Eisen und Staul. Archiv für das Eiseuhüt-
tenwesen 11:509–520
10. Ekobori I (1971) Physics and mechanics of fracture and strength of solid
bodies (in Russian). J Metallurgiya 1071:264
114 K.A. Yushchenko, V.S. Savchenko

11. Okrainets PH, Pishchak VK (1978) Stress dependence of creep rate for metals
with fcc lattice (in Russian). J Fizika Metallov I Metallovedeniye 46, Issue
3:597–602
12. Skleniþka V, Saxl I, öadek I (1983) The role of grain boundaries in deforma-
tion and fracture at high temperatures (in Czech). Kovove materaly 4, vol 21,
pp 364–382
13. Yushchenko KA, Makhnenko VI, Savchenko VS, Velikoivanenko EA,
Chervyakov NO (2006) Investigation Of Thermal-Deformation State Of
Welded Joints In Stable-AusteniticSteels And Nickel Alloys. IIW Doc.IX-
2224-06
Submerged Arc Welding – A Test
for Centreline Cracking

K. Håkansson

Kockums AB, Malmö, Sweden

Abstract

A newly developed test method for centreline cracking or solidification

cracking has been evaluated by adding nickel as cold wire to the weld
metal. Welding was performed by using synergic cold wire submerged arc
welding with various arc wire diameters to achieve different nickel
contents in the weld metal. Welding was done as partial penetration single
bevel groove joints in the PB position. This investigation showed that the
test method picks up the sensitivity to centreline or solidification cracking.
The amount of cracking increased as the nickel content in the weld metal
increased.

Introduction

The present chapter describes briefly the work carried out at Kockums
Laboratory, to design a simple test to examine centreline cracking
behaviour during submerged arc welding of high strength steels. The
welding was done with addition of cold wire containing a high amount of
nickel. It discusses in more detail the results of the addition of cold wire
containing 98% nickel to the weld. The diameter of the arc wire was
changed on purpose to get different amounts of nickel in the weld metal.
An increase in the nickel concentration promotes austenitic solidification
and therefore should promote solidification cracking, the effect being
exaggerated when both the nickel and the sulphur concentrations are large
[1]. It is widely accepted that sulphur and phosphorus are harmful
impurities with respect to the solidification cracking of steel welds.
An excerpt from “The Fabricators and Erectors Guide to welded Steel
Construction” [2] states that centreline cracking is characterized by
separation in the centre of a given weld bead. If the weld bead happens to
be in the centre of the joint, which is normal for a single-pass weld,
116 K. Håkansson

centreline cracks will be in the centre of the joint. In the case of multi-pass
welds, where several runs per layer may be applied, a centreline crack may
not be in the geometric centre of the joint, although it will always be in the
centre of the bead.
Centreline cracking is the result of one of the following phenomena:
-segregation induced cracking
-bead shape induced cracking
-surface profile induced cracking
Unfortunately, all three phenomena reveal themselves in the same type of
crack, and it is often difficult to identify the cause. Moreover, experience
has shown that often two or even all three of the phenomena will interact
and contribute to the cracking problem. Understanding the fundamental
mechanism of each of these types of centre line cracks will help in
determining the corrective solutions [2].
Segregation induced cracking occurs when low melting point
constituents such as phosphorous, zinc, copper and sulphur compounds in
the admixture separate during the weld solidification process. Low melting
point components in the molten metal will be forced to the centre of the
joint during solidification, since they are the last to solidify and the weld
tends to separate as the solidified metal contracts away from the centre
region containing low melting point constituents.
In submerged arc welds, the cracking risk may be assessed by
calculating the Units of Crack Susceptibility (UCS) from the weld metal
chemical composition (weight %) [3, 4, 7]:

UCS = 230C* + 190S + 75P + 45Nb – 12.3Si – 5.4Mn -1

C* = carbon content or 0.08, whichever is higher

Although arbitrary units, a value of <10 indicates high cracking resistance

whereas >30 indicates a low resistance. For fillet welds, runs having a
depth to width ratio of about one, UCS values of 20 and above will
indicate a risk for cracking [3, 4, 7].
When centreline cracking induced by segregation is experienced,
several solutions may be implemented. Since the contaminant usually
comes from the base material, the first consideration is to limit the amount
of contaminant pick-up from the base material. This may be done by
limiting the penetration of the welding process through the use of lower
welding currents.
The second type of centreline cracking is known as bead shape induced
cracking. This type is associated with deep penetration processes like
Submerged Arc Welding (SAW or method 121) and CO2 shielded Flux
 Submerged Arc Welding 117

Cored Arc Welding (FCAW or method 136). When a weld bead is of a

shape where there is more depth than width to the weld cross section, the
solidifying grains growing perpendicular to the steel surface intersect in
the middle, but do not gain fusion across the joint. To correct for this
condition, the individual weld beads must have at least as much width as
depth [2]. The total weld configuration, which may have many individual
weld beads, can have an overall profile that constitutes more depth than
width. If multiple passes are used in this situation, and each bead is wider
than it is deep, a crack free weld can be made. [2].
When centreline cracking due to bead shape is experienced, the obvious
solution is to change the width to depth relationship. This may involve a
change in joint design. Since the depth is a function of penetration, it is
advisable to reduce the amount of penetration. This can be accomplished
by utilizing lower welding current and larger diameter electrodes. All of
these approaches will reduce the current density and limit the amount of
penetration.
The third mechanism that generates centreline cracks is surface profile
conditions. When concave weld surfaces are created, internal shrinkage
stresses will place the weld metal on the surface in tension [2]. Conversely,
when convex weld surfaces are created, the internal shrinkage forces will
pull the surface into compression. Concave weld surfaces frequently are
the result of high arc voltages. A slight decrease in arc voltage will cause
the weld bead to return to a slightly convex profile and eliminate the
cracking tendency. High travel speeds may also result in this
configuration. A reduction in travel speed will increase the amount of fill
and return the surface to a convex profile [2].
Hot cracking tests for weldments are given in the standard EN ISO
17641-part 1. The standard divides the tests into two types where part 2
contains a description of Self-Restraint Tests and part 3 contains a
description of Externally Loaded Tests [5].

Experiment

In order to design a simple test for investigating the sensitivity for

centreline cracking for submerged arc welding of high strength steels, the
following experiments were performed. The joint preparation was a single-
bevel groove T-joint, which is regarded as giving the highest shrinkage
stresses transverse the weld and showing the highest sensitivity to
centreline cracking. The design of the test piece was a 30 mm thick, 300
mm square plate, and a round joint prepared plate with the diameter of 200
118 K. Håkansson

mm and the thickness 35 mm was placed in the centre. On the other side of
the square plate, a round bar of 25 mm diameter was tack welded in the
centre. This bar was used to mount the test piece in a manipulator during
welding. Welding was made using a rectifier type constant voltage with
1400A as maximum output and both wires were fed by a synergic cold
wire kit (SCW), which feed the cold wire in synergy with the arc wire into
the weld pool where it melts. This means that the arc and the cold wire
ratio always remain constant after suitable wire diameter is selected. See
Fig. 1.

Fig. 1. The test set up where the manipulator drives the test piece during the
welding of the fillet weld by submerged arc welding

The chemistry of the weld and the deposition rate was controlled and
pre-selected [6].

Fig. 2. Adjustment of the arc wire and cold wire in the single bevel groove joint. It
is essential that the round test piece is aligned with the driving shaft of the
manipulator
 Submerged Arc Welding 119

The arc wire for all welds was a lean alloyed wire with different
diameters. This wire gives about 0.5% manganese in the weld.
A basic agglomerated welding flux for welding high strength steels and
low temperature steels was used. As cold wire, a wire containing 98%
nickel and 2% titanium with a diameter of 0.8 mm was used. Figure 2
shows the alignment of both wires.
The theoretical chemical composition of the weld metal depends on the
difference in diameter of the arc wire and the cold wire. See Table 1.

Table 1. Area and theoretical composition of nickel of different electrode diameters

Diameter of Area Theoretical % Ni in Theoretical Ni content in
wire mm mm2 the mixed consumable diluted weld metal
containing 0.56% base
metal and 0.44% of the
mixed consumable %
0.8 Cold wire 0.503 – 0.67
2.5 Arc wire 4.909 9.2 4.64
3.0Arc wire 7.069 6.5 3.51
4.0 Arc wire 12.57 3.7 2.32

The weld metal will have a different composition due to dilution from
the base material. In submerged arc welding this dilution is assumed to be
about 50–60%. The chemical composition of the base material, the arc
electrode and the cold wire is given in Table 2.

Table 2. Chemical composition base material and wires

Element Plate material Cold wire Ø 0.8 mm Arc wire
C 0.112 0.031 0.080
Si 0.246 0.27 0.03
Mn 0.884 0.20 0.36
P 0.009 0.000 0.021
S 0.0009 0.003 0.012
Cr 0.424 0.00 0.02
Ni 1.236 96.6 0.03
Mo 0.344 0.00 0.02
Nb 0.013 0.00 0.00
Ti 0.005 2.9 0.00
V 0.007 0.01 0.02
Cu 0.198 0.01 0.08
Fe Balance – Balance
120 K. Håkansson

Welding Procedures

Welding was performed in flat position without any preheat and in one
single pass. The travel speed was set by the rotation performed by the
manipulator. The travel speed was aimed at 40–45 cm/min; the welding
current was aimed at 350 ampere and the voltage at around 28 V. These
figures give a heat input of 1.3–1.5 kJ/mm.

Results

Non-Destructive Testing by Liquid Penetrant

Each of the four welds was non-destructive tested by liquid penetrant to

reveal any cracks extending to the surface of the weld.
The amount of indication in the form of the length of the indication
from the investigation was measured for each of the four tests. The welds
containing high amounts of nickel show the largest length of indication for
cracks, with the length increasing with an increase in the nickel content of
the weld metal. The result from the non-destructive testing is shown in
Table 3.

Table 3. Result from the non-destructive testing

Mark Arc wire Cold Total test Cracked Percentage cracking
diameter wire length length as % of the total weld
mm addition mm mm length
HC1 4.0 Yes 578 10 1.7
HC2 3.0 Yes 578 373 64.5
HC3 3.0 No 581 0 0.0
HC4 2.5 Yes 580 542 93.4

Chemical Composition of Weld Metal

Material from the weld metal was removed from each of the welded
samples by drilling holes in the weld metal. The chips were used for
chemical analysis of the weld metal. The chemical composition was
analysed by Inductively Coupled Plasma emission Spectroscopy ICP. The
result of the chemical analysis of the weld metal is shown in Table 4.
 Submerged Arc Welding 121

Table 4. Chemical composition of the different weld metals (balance Fe)

Mark HC1 HC2 HC3 HC4
Dimension filler wire mm 4.0 3.0 3.0 2.5
Cold wire NG61 addition Yes Yes No Yes
C 0.093 0.094 0.085 0.085
Si 0.21 0.21 0.19 0.22
Mn 0.70 0.75 0.68 0.65
P 0.011 0.012 0.010 0.009
S 0.0075 0.0058 0.0059 0.0068
Cr 0.22 0.21 0.23 0.22
Ni 2.54 3.41 0.70 4.77
Mo 0.18 0.17 0.20 0.18
Ti 0.02 0.02 0.01 0.05
UCS-value 16.28 15.99 14.41 14.30

UCS value was calculated according to EN 1011-2 appendix E [7], where

a value below 10 gives a low risk for cracking and value over 30
represents a high risk.
The nickel content in the weld metal increased as the arc wire
diameter was decreased as expected. Figure 3 shows a piece of the
surface on the test piece with the highest nickel content. Due to the
large penetration into the base material, by the submerged arc welding
method, a large part of the base material is mixed together with the
consumables to form the weld metal. Calculation shows that the weld
metal contains about 50–60% base material.

Fig. 3. The test piece with the highest amount of nickel shows a solidification
crack almost all around the circumference
122 K. Håkansson

Macro Test of Welds

From each weld two macro test pieces were taken transverse to the weld. A
cut divided the test plate into two halves, and the macro tests were removed
from each exposed side. The tests were prepared metallographically and
etched in Nital. The tests were inspected using a stereo microscope at ten
times magnification and the results are shown in Table 5.

Table 5. Results of inspection of macro tests transversing the weld

Mark Nickel Remarks Depth to
content in width
weld metal ratio
HC3-1 0.7% Acceptable. See Fig. 4a 0.74
HC3-2 0.7% Acceptable 0.71
HC1-1 2.54% 1 mm crack in the root. Root gap of 0.85
0.4 mm. See Figs. 4b and 5a.
HC1-2 2.54% No cracks. Undercut 0.5 mm. Root gap 0.70
0.3 mm
HC2-1 3.41% Centreline crack 3 mm. Sideline crack 1.12
2 mm+1 mm. Root crack 4 mm. Root gap
1 mm.
See Fig. 4c.
HC2-2 3.41% Centreline crack 1.5 mm. Root crack 0.88
3 mm.
Root gap 0.5 mm.
HC4-1 4.77% Sideline cracks 1 mm + 1 mm + 2 mm. 0.75
Root crack 1.5 mm. See Fig. 4d.
HC4-2 4.77% Centreline crack 1 mm. Sideline cracks 0.76
2 mm + 1.5 mm. Root crack 1.5 mm.
Root gap 0.2 mm. See Figs. 4e and 5b.

According to Kazutoshi Ichikawa et al. [1] the probability for

solidification cracking is predicted to be dependent on the sulphur and the
nickel content in the weld metal, where the probability increases
considerably if the nickel content exceeds 2.5% and the sulphur content
exceeds 0.010%.
The amount of cracking in the weld increases as the nickel in the weld
metal increases. See Fig. 6. All tests with addition of nickel show cracks
from the root of the weld. The smaller diameter of wire, which gives the
highest amount of nickel, gave deeper penetration. Increased root gap have
a tendency to increase the depth of penetration. This deeper penetration
increases the depth and decreases the width of the weld, which gives a
higher risk for solidification cracking.
 Submerged Arc Welding 123

a b

c d

Fig. 4. Cross-sections of welds discussed in Table 5 with increasing Ni content:

(a) 0.7%, (b) 2.54%, (c) 3.41%, (d) 4.77, (e) 4.77%

a b

Fig. 5. Micrographs of welds (a) HC1-1 and (b) HC4-2 showing root cracks
124 K. Håkansson

Pe rce ntage crack ing as a f unctio n of th e nick el conten t in the w eld

m etal

10 0

90

80

70

60

50

40

30

20

10

0
0 ,7 2,5 4 3,4 1 4,77

P e r c e nt a ge n i c k e l i n t he we l d me t a l

Fig. 6. Percentage of cracking shown as a function of weld metal nickel content

Microstructure

The microstructure of the weld metal with no nickel addition contains a

large amount of acicular ferrite together with proeutectoid ferrite side
plates. As the nickel content increases the microstructure of the more
highly alloyed weld metal shows a change to a microstructure containing
more lath martensitic constituents.

Hardness Test According to Vickers HV5 on Macro Tests

The hardness test was made according to EN 1043-1 with the load of 5 kg,
with the exception of that no hardness readings were made in the base
material [8]. Three indentations were made in the heat affected zone of the
square plate, in the weld metal and in the heat affected zone of the round
plate. The result of the hardness survey is shown in Table 6.
The hardness in the weld metal increases as the nickel content increases
and this may be due to changes in the microstructure in the weld metal to a
more martensitic microstructure. The hardness indicates that the weld
metal may have a tensile strength of about 1200 MPa for the two highest
nickel levels.
 Submerged Arc Welding 125

The hardness of the weld metal with no addition of nickel is also rather
high, which can be due to the high amount of highly alloyed base material
mixed into the weld metal. The fillet weld itself normally experiences a
more rapid cooling, which increases the hardness.

Table 6. Results from hardness tests on macro tests

Mark Nickel content HV5 in square HV5 in weld HV5 in round
in the weld plate metal plate
metal
HC3-1 0.7% 391, 386, 296 241, 244, 244 396, 391, 376
HC3-2 0.7% 362, 376, 293 236, 241, 236 376, 376,396
HC1-1 2.54% 371, 381, 353 286, 299, 289 407, 412, 412
HC1-2 2.54% 401, 386, 283 306, 358, 336 407, 407, 386
HC2-1 3.41% 396, 371, 289 376, 371, 381 367, 429, 423
HC2-2 3.41% 391, 391, 277 321, 381, 367 362, 396, 407
HC4-1 4.77% 396, 396, 299 358, 367, 367 418, 418, 401
HC4-2 4.77% 391, 386, 286 386, 376, 362 412, 396, 329

Conclusions

The developed test method uses ordinary welding equipment, such as a

manipulator and submerged arc welding equipment, to perform the test.
The design of the test piece in form of a T-butt joint shows that it may
be too severe to predict the risk according to the UCS number for
solidification cracking in the weld metal.
The amount of centreline cracking in form of solidification cracking
increases as the nickel content increases.
The root gap may influence the result by affecting the depth to width ratio.
The microstructure of the weld becomes martensitic for the highest levels
of nickel.
The hardness increases as the nickel content increases in the weld metal.

References

1. Kazutoshi Ichikawa, Bhadeshia HKDH, and MacKay DJC (1996) Modell for
solidification cracking in low alloy steel weld metal. Science and Technology
of Welding and Joining Vol. 1 no. 1, pp 43–50
2. The James F. Lincoln Arc Welding Foundation. Weld Cracking. An Excerpt
from the Fabricators’ and Erectors Guide to Welded Steel Construction. pp 1–4
126 K. Håkansson

3. TWI World Centre for Material Joining Technology – Defects- solidification

cracking
4. Bailey N, Jones SB (1978) The Solidification cracking of ferritic Steel during
Submerged Arc Welding. Welding Journal, Aug. pp 217s–231s
5. ISO/FDIS 17641-1 Destructive tests on welds in metallic materials- Hot
cracking tests for weldments- Arc welding processes- part 1: General
6. ESAB Brochure “Fluxes and wires for Submerged Arc Welding”
7. EN 1011-2:2001 issue 1 Welding-Recommendations for welding of metallic
materials-Part 2: Arc welding of ferritic steels
8. EN 1043-1 (1996) Destructive tests on welds in metallic materials-Part 1:
Hardness test on arc welded joints
Influence of Local Weld Deformation
on the Solidification Cracking Susceptibility
of a Fully Austenitic Stainless Steel

A. Kromm, Th. Kannengießer

Federal Institute for Materials Research and Testing, Berlin, Germany

Abstract

For the evaluation of the solidification cracking behaviour of welded struc-

tures, the influence of the external boundary conditions needs to be con-
sidered, in addition to the metallurgical aspects. Against this background,
the solidification crack formation in a fully austenitic stainless steel under
variation of external restraint was examined in this study.
For this purpose, a newly developed hot cracking test (CTW test) was
used for the first time, which allows application of a defined tensile load
transverse to the welding direction during welding. In addition, the strains
and strain rates could be determined with the help of a mechanical-
electrical measuring device in the near field of the weld pool. These values
were examined both under free contraction and varied external load, i.e.
under different constant cross head speeds. The Critical strain and strain
rate required for the propagation of macroscopic surface cracks were de-
termined. By means of high speed recording the authors succeeded in cor-
relating strain and strain rate with the relative position of the weld pool. In
addition, centreline crack initiation and growth were located.

Introduction

A potential problem when welding austenitic steels is their distinct ten-

dency to hot crack. This is due to the solidification behaviour of these ma-
terials. The range between liquidus and solidus temperature, i.e. the mushy
zone, is the place of solidification crack formation. Here, liquid metal sur-
rounds the already solidified material. Volume shrinkage during the liquid
to solid phase transformation must be compensated for the remaining melt.
Otherwise solidification cracks will occur [1, 2].
128 A. Kromm, Th. Kannengießer

For the evaluation of the hot cracking resistance of welded components,

the influence of the external boundary conditions is to be considered in
particular, apart from the metallurgical aspects [3]. According to Prokhorov
[1] and Rappaz et al. [2], local strain rates in the vicinity of the weld pool
can significantly affect hot crack formation and can thus be regarded
as a suitable hot crack criterion. In addition, in the work of several authors
[4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18], it was pointed out that a
hot crack formation is usually connected with the occurrence of tensile
stresses and strains in hot crack-sensitive areas. In the case of an external
restraint, e.g. by the design specific shape of the component itself or by
additional external load, these stresses/strains near the weld pool are af-
fected and are directly associated with hot crack formation. Up to now a
multiplicity of self restraint or externally loaded tests have been developed
in order to evaluate the hot cracking susceptibility of materials. With the
help of such tests mostly just material rankings can be provided. Transfer-
ability to component welds or quantification of hot cracking susceptibility
can be achieved by determining, among other things, the hot crack critical
strain and strain rate in the weld pool vicinity [9, 10].
In the present study, the approach of a deformation measurement at pins
located near the weld was used [9]. The movements of these pins were
transferred to an inductive measuring sensor by means of a mechanical
measuring device. The collected data allowed determination of strains and
strain rates. Under variation of the external load in the Controlled Tensile
Weldability (CTW)-Test, the influence of deformation near the weld pool
on the solidification cracking susceptibility was examined. The hot crack-
ing behaviour in the weld pool vicinity was characterized by high speed
recording.

Design Specific Influences on Local Thermo-Mechanics

Solidification cracking in weld metal results from the competition between

cracking resistance of the material and the mechanical load. If the load ex-
ceeds the cracking resistance, then cracking develops [4]. This means, with
regard to the hot cracking behaviour, that there is a close relationship be-
tween metallurgical and mechanical influences, which is affected by weld-
ing parameters and design boundary conditions. Investigations into this
topic have clarified that metallurgical, thermal and mechanical factors can
have an influence on the thermo-mechanics in the vicinity of the weld
pool. In many investigations the definition of areas is common, in which
due to heterogeneous heating and cooling either compressive or tensile
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 129

stresses are present. Further stresses with opposite algebraic sign develop
in other areas, due to the balance of forces [4-8, 12-18].
It is an undisputed fact that in front of the moving weld pool compres-
sive stresses are directly formed. The magnitude of this stress cell is
strongly affected by the welding parameters, particularly by the travel
speed [4-8]. In reaction to this, tensile stresses in the still cold base mate-
rial appear. The expansion in front of the weld pool continues directly at
its sides. The maximum of the longitudinal extensions is within this range.
At high welding speeds, this strain maximum can extend to the region be-
hind the weld pool. This has direct consequences on the solidified material
already cooling down there. Normally, contraction due to cooling creates a
tensile stress in this region, as with slow welding speeds. However, ac-
cording to Chihoski [17, 18], the expansion of the surrounding material
can thus superimpose the contraction within the cooling range -supported
by longitudinal expansions- resulting in a pressure cell. Thereafter, under
uninfluenced contraction, a tensile cell develops. The investigations of
Feng and Zacharia [4-8] confirm the presence of a pressure cell directly
behind the weld pool. Figure 1 illustrates this circumstance. Thus, solidifi-
cation cracking can be prevented.
The development of compressive and tensile cells around the weld pool
crucially depends on the welding parameters, namely on the heat input.
From this, it becomes clear that the place of a possible crack initiation can
be shifted relative to the weld pool. Whether the mechanisms of crack ini-
tiation are the same in different locations, needs to be clarified, particularly
since any thermal pre-treatment of the component should have an addi-
tional effect [12]. Up to now, the emphasis of most investigations has been
on the influence of welding parameters on hot crack formation. The num-
ber of investigations into the design specific influences is smaller.
It becomes clear that the influence of external load or geometry itself
during the welding procedure acts in a similar way, i.e. with movement of
compressive and tensile cells. First, a dependence of the stress/strain dis-
tributions of the momentary weld pool position on the specimen is to be
emphasized. Consequently, this distribution changes with distance and
proximity to boundary regions. Without the supporting effect of these ar-
eas, the material behind the weld pool seems clearly higher stressed and
more rapidly loaded under tensile stress, which should favour cracking
[4, 6].
130 A. Kromm, Th. Kannengießer

Fig. 1. Compressive stresses behind the weld pool [8]

The application of an external, static load has similar effects, as used with
the Sigmajig – test [19]. Applied tensile stresses directly affect the level of
the locally evolving deformations. These can be increased or decreased [4,
5, 7, 8]. The influence of external restraint was examined recently also by
Kannengiesser [20]. Results show that higher restraint intensities can lead
to a decrease of the solidification cracking susceptibility.
Hence it follows that potential cracking can either be increased or de-
creased by the design specific or mechanical boundary conditions. In [17,
18] this statement is also confirmed. It seems that the causes leading di-
rectly to cracking (local strains and strain rates, respectively) can alter with
the geometry of the material (e.g. round vs. rectangular specimens) [13,
14]. That would mean that crack-critical local values for the same material
vary with the geometry (restraint intensity). In the present study a contri-
bution is made to the question of how local strains and strain rates are in-
fluenced with respect to external load and how solidification cracking is
affected.
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 131

Experimental

Test Material

The material used in the investigation is the high-alloyed, fully austenitic

stainless steel UNS NO. S 34565. The chemical composition is shown in
Table 1. The mechanical characteristics determined in the tensile test can
be seen from Table 2.

Table 1. Chemical composition of S 34565 (wt.%)

C Si Mn Cr Ni Mo N
0.03 0.111 6.639 24.1 19.04 4.972 0.551

Table 2. Mechanical characteristics of S 34565

Yield Strength, Tensile Strength, Ultimate Strain, Reduction of
Rp0,2 Rm A50 Area, Z
548 MPa 932 MPa 50 % 48 %

Hot Cracking Behaviour in the MVT-Test

For the characterisation of the hot cracking susceptibility, preliminary in-

vestigations using the MVT-Test [21] were performed. The investigations
were accomplished in the Varestraint – mode. The crack evaluation
showed predominantly solidification cracks in the weld metal, marginal li-
quation cracks and no ductility dip cracks. The material was therefore clas-
sified by the expression “increasing hot cracking susceptibility”.

Controlled Tensile Weldability Test

The solidification cracking behaviour was examined both under free con-
traction and under varying external load. For this purpose, a newly devel-
oped, externally loaded hot cracking test (CTW-Test) was used for the first
time. In this test a defined tensile load can be applied during welding,
transverse to the welding direction [11]. The CTW-Test setup is shown in
Fig. 2.
132 A. Kromm, Th. Kannengießer

Fig. 2. CTW – Test setup

In order to quantify the susceptibility to solidification cracking, a mini-

mum critical cross head speed va,crit was determined for producing solidifi-
cation cracks at the specimen surface. This value was determined to be
va,crit = 4 mm/min. Based on this value the parameters for the following in-
vestigations were chosen and are indicated in Table 3.

Table 3. External load parameters

External Restraint Free Constant Cross Head Speed
Load [mm/min]
Parameter – 3 4 5
Variation
Solidification no no yes yes
Crack

The specimen geometry and dimensions are shown in Fig. 3. The welding
parameters are summarized in Table 4. In order to avoid additional notch
effects and cracking at the weld start, special clamping adapters were used,
and the weld start and end were placed 5 mm away from the edge of the
sample resulting in a weld length of 45 mm.
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 133

Table 4. GTA bead-on-plate welding parameters

Basic parameter Value
Welding Current 100 A
Welding Voltage 13 V
Travel Speed 5 mm/s
Heat Input 0,26 kJ/mm

Fig. 3. Specimen with weld and centreline crack position

Strain Monitoring

In order to capture local strains, a mechanical-electrical measuring system

was developed. The system is based on direct displacement measurement
at defined points on the specimen surface. Via a mechanical leverage sys-
tem the specimen movements were transferred to an inductive measuring
sensor. Its electrical signals were online computed to obtain the respective
strains. Figure 4 show the measuring system on a specimen. For the link-
ing of measuring instrument and specimen surface tungsten pins were
placed on the surface using induction welding. Reproducible positioning of
the pins was accomplished with the help of a positioning device.
Distances from the fusion line of 1.5 mm (near field) and 3.5 mm (far
field) were selected. From the measured pin movements over the outside
pin distance (see Fig. 3) the transverse strain over the weld seam was de-
termined. The strain rate was calculated as temporal derivative of the
strain.
134 A. Kromm, Th. Kannengießer

Fig. 4. Mechanical leverage positioned on specimen

High Speed Recording

In order to relate the weld pool dimensions and position to the strain meas-
urements, the weld pool movement was recorded optically. This was real-
ized using a high-speed camera. Furthermore, the records served to deter-
mine of the following parameters:
• Time of Solidification
• Time of crack initiation
• Location of crack initiation
• Characteristics of crack initiation
Before the investigations were started, the samples were provided with a
mark perpendicular to the welding direction, which was used as point of
reference in order to determine the weld pool dimensions. Furthermore, the
mark represents the place of the hot crack initiation determined in prelimi-
nary tests.

Results and Discussion

Weld Pool Dimensions and Crack Initiation

The high speed records helped to determine the dimensions of the weld
pool and to observe crack initiation and crack growth. The following Figs.
5–7 illustrate the formation and progress of a solidification crack at a con-
stant cross head speed of va,crit = 4 mm/min in the CTW-Test. As shown in
Fig. 5a, the crack arose at the solidification line. This is at the same point
of time as the weld pool begins to narrow.
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 135

The hot crack grows perpendicular to the solidification line toward weld
pool centre. The growth direction corresponds to the solidification direc-
tion. This is represented in Fig. 5b. At the same time, a second hot crack
develops on the opposite flank of the travelling weld pool (Fig. 6a). Both
cracks grow toward the weld pool centre, until they unify at a certain point
in time at the trailing edge of the weld pool, as shown in Fig. 6b.
(a) (b)

Crack Growing Path

Solidification
Line

Weld Pool
Crack initiation
Crack

Solidification Path

Fig. 5. Solidification crack initiation and growth

(a) (b)

Crack Unification
Opposite
Crack

Fig. 6. Initiation of opposite crack and crack unification on trailing edge of the
weld pool

The described hot cracks obviously unify to become a centreline crack

parallel to the welding direction. The centreline crack is always located
right behind the weld pool at the trailing edge. Fig. 7 show growth and ex-
pansion of the developed centreline crack.
136 A. Kromm, Th. Kannengießer

(a) (b)

Trailing Edge of Expanding Centreline

Weld Pool Crack

Centreline
Crack

Fig. 7. Centreline crack following trailing edge of weld pool

Note that the formation of solidification cracks in the present study is

bound to an external load. Without an external load no centreline cracking
appeared. Likewise, at an external load below the determined critical cross
head speed no centreline cracking occurred. This permits the conclusion
that centreline cracking is caused by exceeding a critical external load
(strain or strain rate, respectively).
The discussed crack features were observed also by [5] in a similar way.
There, transverse cracks arose at the fusion line when welding a nickel
base alloy under external preloading. Cracks grew during progressive
welding towards the centreline. However, they did not form a centreline
crack. By SEM photographs, the observed transverse cracks were identi-
fied as solidification cracks. This phenomenon was observed particularly
when high welding speeds were applied and prompted the authors to as-
sume strains, mainly longitudinal strains, as the cause of cracking.

Strain and Strain Rate during Welding

Restraint Free

In the following diagrams, strain and strain rate in the near (Fig. 8) and
far field (Fig. 9) of the weld pool are represented. Solidification cracking
did not occur with these welding conditions. The weld pool position is
highlighted. In the curve of the near field, first a rise in strain rate with
approaching weld pool is observed due to the forward-moving heat. The
consequence is a slight strain of approx. 0.3%. With the arrival of the weld
pool, the strain rate decreases and changes its algebraic sign. The pin
movement towards the melt is now accelerated. With narrowing of the
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 137

weld pool, this movement is slowed and goes finally towards zero during
cooling. The cooling-conditioned compression takes a value of approxi-
mately –4%.

Fig. 8. Strain and strain rate in near field of restraint free specimen

A qualitatively similar curve shape is obtained for the far field (Fig. 9).
However, the absolute values here are obviously smaller. In [15], similar
strain processes were measured and simulated when welding an aluminium
alloy.

Fig. 9. Strain and strain rate in far field of restraint free specimen

It remains to be stated that the material in the near field of the weld pool is
exposed to both positive and negative strains. These strains are mainly
thermally conditioned, since the material was welded under free contrac-
tion. In the near field, the amounts of the arising strains and strain rates are
138 A. Kromm, Th. Kannengießer

significantly higher than in the far field. The observed strain behaviour
agrees in principle to the discussed model conceptions of Zacharia [7, 8]
and Chihoski [17, 18].

Effect of External Load

In Figs. 10 and 11, strain and strain rate curves are represented when weld-
ing in the CTW-Test under external load at a cross head speed of
va = 3 mm/min for near field and far field, respectively. A hot crack was
not observed at this load. The results of measurement show that, in com-
parison to Figs. 8 and 9, an external load directly affects the weld pool
near strains and strain rates. In contrast to welding under free contraction,
near and far field curves show higher strain caused by the external load.
In addition, significantly different strain behaviour is observed as a
function of the weld distance, while the weld pool passes the measuring
point. Two strain maxima are found in the near field. Strains and strain
rates are smaller in the far field. It should be emphasized that the already
discussed typical strain behaviour in the weld pool vicinity, under free
contraction, was registered qualitatively here, although the thermally
caused strains were superimposed by an additional external load. Indeed,
additional strains acting on the solidifying weld metal are present, but the
strain rate does not seem to be high enough to initiate cracking.

Fig. 10. Strain and strain rate in near field applying an external load of va = 3 mm/min
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 139

Fig. 11. Strain and strain rate in far field applying an external load of va = 3 mm/min

In Figs. 12 and 13, strain and strain rate are represented when welding in
the CTW-Test applying cross head speeds of va,crit = 4 mm/min and
va = 5 mm/min, respectively. These loads triggered solidification crack
formation in the form of a centreline crack as mentioned above. Passing of
the crack at the measuring point is marked in the Figures.

Fig. 12. Strain and strain rate in near field applying an external load of
va,crit = 4 mm/min (critical cross head speed to induce a centreline crack)

Note that the already mentioned typical strain behaviour in both cases was
observed here again. However, the absolute values rise at increased load.
The curves of strains and strain rates in the far field show lower values and
resemble qualitatively the far field measurements in Fig. 11.
140 A. Kromm, Th. Kannengießer

Fig. 13. Strain and strain rate in near field applying an external load of
va = 5 mm/min

Solidification Crack accompanying Strain and Strain Rate

Assuming that when a centreline crack follows directly the trailing edge of
the weld pool, the time when the centreline crack propagates through the
measuring range could be determined with the help of the high speed video
records as follows (see Fig. 3).
Pcrack − Pstart + lpool (1)
tcrack =
vs
(2)
lpool = (tedge − tarc) × vs

tcrack point in time of crack propagation at measuring point

Pcrack location of crack tip
Pstart location of weld start
vs travel speed
lpool weld pool length
tedge trailing edge of weld pool at measuring point
tarc arc / electrode at measuring point
Thus, the strains and strain rates available during crack propagation could
be determined from the measuring curves. These values are displayed in
the diagram represented in Fig. 14. Additionally, strains and strain rates of
welds without cracking are also shown in this diagram.
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 141

Fig. 14. Strain and strain rate for cracked and crack free welds

Although the results are subject to relatively large scatter, it becomes clear
that solidification crack propagation requires high critical strain rates. It is
obvious from the ranges A and B in Fig. 14 that quantification of the ex-
ternal restraint depending hot crack initiation should be possible based on
the local strain rates in the weld pool vicinity.

Influence of External Load on Local Thermo-Mechanics

Following the established relationship between hot crack formation and

tensile stresses present within hot crack-sensitive ranges, as reported in [4-
8, 16-18], the effect of an external load on the strain in the direct weld pool
vicinity is discussed. First, only thermally conditioned effects are shown
schematically on this basis. In Fig. 15, the near field range around the trail-
ing weld pool is represented schematically. Besides heated and cooled ar-
eas, arising strains are indicated by arrows.
The temperature gradient in front of the weld pool is extremely high.
This entails a sudden material expansion which can run only in transverse
or longitudinal direction toward the weld pool edge. In [16-18], such a
longitudinal flow of material is mentioned several times. In the transverse
direction, the expansion takes place outward, since in front of the weld
pool there is also material which must expand. Therefore, a tensile strain is
registered in the near field range. In the vicinity of the weld pool the ther-
mal expansion can run also toward the liquid area. The weld pool is able to
absorb a certain portion of the expansion [18].
142 A. Kromm, Th. Kannengießer

Fig. 15. Strains (indicated by arrows) caused by pure thermal load in vicinity of
the weld pool

Just behind the arc, when the temperature at the fusion line begins to
fall, the contraction by cooling starts in the direct vicinity of the weld pool.
The areas outside this region, however, are still heated up, i.e. they still
expand. With progressive cooling and contraction of adjacent areas, the
heated material here is more and more able to let its expansion proceed
inward towards the liquid area. The strain registered changes into a com-
pressive one, although heating (expansion) is still present. Due to progres-
sive cooling, the compressive strain will remain. Figure 15 clarifies this re-
lation. In the literature [17, 18], it is reported that the expansion of areas
still heated up can superimpose the contraction of already solidified areas
in such a manner that even compressive stresses develop there. According
to [4-8], this compressive cell can reach behind the weld pool and may
prevent cracking (see Fig. 1).
When an external load is applied, the deformation behaviour (Fig. 16)
changes. The described thermo-mechanics are superimposed by external
load. In the initial state not yet warmed up material strains, due to the load.
With heating of the material these strains become higher, since the rigidity
of the material decreases. Additionally, thermal expansion begins. Thermal
mechanics and load mechanics interact here in the same direction.
As previously mentioned, it is possible for the heated material to pro-
ceed with its expansion towards the weld pool. This thermo-mechanical in-
fluence interacts with the external load and at first prevails. With progres-
sive external load; its influence continues to increase until it finally
exceeds the thermal mechanics. Now again; a tensile strain is registered.
Thermo-mechanics and external load act against each other.
The presence of this tensile cell enables cracking in agreement with the
discussed model conceptions (see Fig. 1).
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 143

Cracking became possible from the moment when a hot crack-sensitive

area (mushy zone) is exposed to a tensile strain and to a corresponding
strain rate of a certain level. In this study, this was already possible with
narrowing of the weld pool, as shown in the high speed records (Figs. 5 and 6).

Fig. 16. Strains (indicated by arrows) caused by interacting thermal and external
load in the vicinity of the weld pool

Conclusion

1. According to [1, 2], local strain rates in the vicinity of the weld pool can
significantly affect hot crack formation, and can be regarded as a suit-
able hot crack criterion.
2. By means of high speed video technology, the weld pool dimensions as
well as the location and point in time of crack initiation and growth
could be reconstructed. This permitted a direct correlation with the de-
termined weld pool near strains and strain rates.
3. The observed typical strain behaviour in the weld pool vicinity is in
principle in agreement with the model conceptions developed and dis-
cussed by Zacharia [7, 8] and Chihoski [17, 18]. An external load causes
an additional significant increase in the strains and strain rates. In the
near field, the magnitudes of the arising strains and strain rates are
clearly higher than in the far field.
4. Contrary to samples under free contraction, additional tensile strains
evolved in appropriate areas of the solidifying weld pool. Mechanical
and thermal loads superimpose and interact differently in the near and
far field.
5. The hot crack formation in the weld pool vicinity is directly affected by
the local strain rate, which seems suitable for the quantification of
144 A. Kromm, Th. Kannengießer

restraint specific hot crack susceptibility. It could be shown that for hot
crack propagation high critical strain rates are necessary.
6. By application of a constant external load, solidification cracks can be
initiated with narrowing weld pool on both sides of the solidification
line. In the present case, the cracks grow perpendicular to the solidifica-
tion line towards the weld centreline where they unify to become a cen-
treline crack.

Acknowledgements

The authors would like to thank A. Hannemann und M. Lammers for the
technical support during this work.

References

1. Prochorow NN, Jakuschin BF, Prochorow N Nikol (1968) Theorie und

Verfahren zum Bestimmen der technologischen Festigkeit von Metallen
während des Kristallisationsprozesses beim Schweißen (in German).
Schweißen und Schneiden 18 1:8–11
2. Rappaz M, Drezet, JM, Gremaud M (1999) A New Hot-Tearing Criterion.
Metallurgical and Materials Transactions A vol 30A February:449–455
3. Cross CE (2005) On the Origin of Weld Solidification Cracking. In: Bölling-
haus Th, Herold H (eds) Hot Cracking Phenomena in Welds. Springer, Berlin
Heidelberg, pp 3–18
4. Feng Z, Zacharia T, David SA (1997) On the Thermomechanical Conditions
for Weld Metal Solidification Cracking. In: Cerjak H (ed) Mathematical Mod-
elling of Weld Phenomena 3. Maney Publishing, London, pp 114–148
5. Feng Z, Zacharia T, David SA (1997) Thermal Stress Development in a
Nickel Based Superalloy During Weldability Test. Welding Research Sup-
plement November:470–483
6. Feng Z, Tsai CL (1993) Modelling the Thermomechanical Conditions at Weld
Pool. In: Zacharia T (ed) International Conference Proceedings on Modelling
and Control of Joining Processes, American Welding Society, Florida, U.S.A.,
pp 525–533
7. Zacharia T (1994) Dynamic Stresses in Weld Metal Hot Cracking. Welding
Research Supplement July:164–172
8. Zacharia T, Aramayo GA (1993) Modelling of Thermal Stresses in Welds. In:
Zacharia T (ed) International Conference Proceedings on Modelling and Con-
trol of Joining Processes, American Welding Society, Miami, Florida, U.S.A.,
pp 533–540
Influence of Local Weld Deformation on Solidification Cracking Susceptibility 145

9. Kannengiesser Th, McInerny T, Florian W, Boellinghaus Th, Cross, CE

(2002) The Influence of local weld deformation on hot cracking susceptibility.
In: Cerjak H (ed) Mathematical Modelling of Weld Phenomena 6. Maney
Publishing, London, pp 803–818
10. Wolf M, Kannengießer Th (2005) Moderne Prüfverfahren zur Bestimmung
der Heißrisssicherheit beim Schweißen von hochlegierten Stählen und
Nickelbasislegierungen (in German). In: Grellmann W (ed) Tagungsband
Werkstoffprüfung. Verlag Stahleisen GmbH, Düsseldorf, pp 419–426
11. Kromm A, Kannengießer Th (2006) Anwendung des neuen Controlled
Tensile Weldability (CTW) Tests zur Untersuchung der Heißrissneigung (in
German). In: Borsutzki M, Geisler S (eds) Tagungsband Werkstoffprüfung.
Verlag Stahleisen GmbH, Düsseldorf, pp 451–456
12. Brooks JA, Dike JJ, Krafcik JS (1993) On Modelling Weld Solidification
Cracking. In: Zacharia T (ed) International Conference Proceedings on Mod-
elling and Control of Joining Processes, American Welding Society, Miami,
Florida, U.S.A., pp 174–185
13. Dike JJ, Brooks JA, Li M (1998) Comparison of Failure Criteria in Weld So-
lidification Cracking Simulations. In: Cerjak H (ed) Mathematical Modelling
of Weld Phenomena 4. Maney Publishing, London, pp 199–222
14. Brooks JA, Dike JJ (1998) Modelling Weld Solidification Cracking Behavior
in Aluminum Alloys-Analysis of Fracture Initiaton. In: Vitek JM, David SA,
Johnson JA, Smartt HB, DebRoy T (eds) Trends in Welding Research, Pro-
ceedings of the 5th International Conference. Materials Park, Ohio, U.S.A., pp
695–699
15. Dike JJ, Brooks JA, Krafcik JS (1995) Finite Element Modelling and Verifi-
cation of Thermal-Mechanical Behavior in the Weld Pool Region. In: Smartt
HB, Johnson JA, David SA (eds) Trends in Welding Research, Proceedings of
the 4th International Conference. ASM International, Materials Park, Ohio,
U.S.A., pp 159–164
16. Johnson L (1973) Formation of Plastic Strains During Welding of Aluminum
Alloys. Welding Research Supplement July:298–305
17. Chihoski RA (1979) Expansion and Stress Around Aluminum Weld Puddles.
Welding Research Supplement September:263–276
18. Chihosky RA (1972) The Character of Stress Fields Around a Weld Arc Mov-
ing on Aluminum Sheet. Welding Research Supplement January:9–18
19. Goodwin GM (1987) Development of a New Hot-Cracking Test – The Sigma-
jig. Welding Research Supplement February:33–38
20. Kannengießer Th, Kromm A (2007) Design-Specific Influences on Local
Weld Displacement and Hot Cracking. Proceedings to II. International Con-
ference in Welding and Joining of Materials-ICWJM 2007, Cusco, Peru, 16.-
18.4.2007, pp 1–9
21. Wilken K, Kleistner H (1982) Der MVT-Test – ein neues universelles
Verfahren zur Prüfung der Heißrissanfälligkeit beim Schweißen (in German).
Material und Technik 1:3–10
Weld Solidification Cracking in Solid-Solution
Strengthened Ni-Base Filler Metals

J.C. Lippold1, J.W. Sowards1, G.M. Murray1, B.T. Alexandrov1, A.J. Ramirez2
1
The Ohio State University, Columbus, Ohio, USA
2
Brazilian Synchrotron Light Laboratory, Campinas-SP, Brazil

Abstract

The weld solidification cracking susceptibility of several solid-solution

strengthened Ni-base filler metals was evaluated using the transverse
Varestraint test. The alloys tested included Inconel 617, Inconel 625,
Hastelloy X, Hastelloy W, and Haynes 230W.* Susceptibility was
quantified by determining the solidification cracking temperature range
(SCTR) which is a direct measurement of the range over which cracking
occurs. This temperature range was then compared to the equilibrium
solidification temperature range derived from Calphad-based
ThermoCalc™ calculations, Scheil-Gulliver solidification simulations, and
in-situ measurements using the single sensor differential thermal analysis
(SS-DTA) technique.
Good correlation among the simulated and measured solidification
temperature ranges, and SCTR values were found for the 617 and 230W
filler metals. These two filler metals exhibited the best resistance to weld
solidification cracking. Correlation among measured and simulated
temperature ranges, and SCTR was poor for Hastelloy alloys X and W.
Alloy 625 was found to be the most susceptible to solidification cracking,
but this result is in conflict with fabrication experience. This appears to be
the result of the inability of the Varestraint test to account for crack
“healing” during the final stages of solidification.

*
Inconel® is a registered trademark of Special Metals Company, a PCC company.
Hastelloy® is a registered trademark of Haynes International. In this paper, a
shortened version of these alloys will be used that does not include the trademark.
148 J.C. Lippold et al.

Introduction

Solid-solution strengthened Ni-base alloys are used in applications where

moderate strength and excellent corrosion resistance are required. Nickel-
based alloys are widely used in a variety of industries, including pulp and
paper, power generation, chemical handling and processing, coal
gasification and liquefaction, and turbine engines. In all these industries,
the combination of mechanical properties and corrosion resistance of these
alloys are required to meet demanding service conditions.
In welded fabrication, one problem that is often encountered with these
alloys is weld solidification cracking. Since the Ni-base alloys are
austenitic, and solidify as a face centered cubic (FCC) gamma phase, they
are inherently susceptible to weld solidification cracking. This form of
cracking occurs along solidification grain boundaries in the weld metal due
to the presence of liquid films at the boundary at the end of solidification.
In this investigation, the susceptibility to weld solidification cracking was
quantified using the transverse Varestraint test. The solidification behavior
was evaluated using a combination of computational and experimental
techniques. A comparison between solidification temperature range and
cracking susceptibility measured using the Varestraint test was then
conducted.
The Varestraint (VAriable RESTRAINT) test was originally developed
by Savage and Lundin in 1965 to measure the susceptibility of an alloy to
solidification cracking [1]. The Varestraint test was originally developed to
be a direct and relatively simple technique for isolating the metallurgical
and compositional factors responsible for cracking from the welding
process variables. The test is widely applicable to most materials and the
results are quantitative, allowing relative susceptibility to solidification
cracking to be easily determined.
The concept of the Varestraint is represented schematically as a beam
supported by rollers, as shown in Fig. 1. An autogenous weld is made
across the sample as a downward bending motion is initiated by applying
force at the loading rollers, which forces the sample to conform to the
radius of the die block.
By using die blocks of different radii and by varying the length of the
stroke of the hydraulic ram to a pre-calculated distance, the augmented
strain in the outer fibers of the test sample can be altered according to the
relationship below.
 Weld Solidification Cracking of Ni-Base Alloys 149

t
ε= (1)
2R + t

where: İ = strain
t = sample thickness
R = radius of curvature of die block

By testing over a range of strain, a relationship between strain and

cracking susceptibility can be determined. In this investigation, the
susceptibility to cracking was determined by measuring the maximum
crack distance (MCD) on the surface of the sample tested at a given strain.
Duplicate samples are normally run at the same strain under identical
testing conditions.
A representative plot of strain versus MCD is shown in Fig. 2. The
lowest strain at which solidification cracking is observed is termed the
“threshold” strain. The strain level above which the maximum crack
distance does not increase is called the “saturated” strain, since at this level
sufficient strain has been applied to force the solidification crack to fully
propagate within the solidification temperature range.

Fig. 1. Schematic illustration of the transverse Varestraint test

150 J.C. Lippold et al.

Maximum Crack Distance (MCD)

MCD at Saturated Strain

Threshold Saturated
Strain Strain

Applied Strain
Fig. 2. Transverse Varestraint test results plotted as maximum crack distance
versus strain

From the MCD versus strain data, the solidification cracking

temperature range (SCTR) can then be determined. The SCTR is useful
because it quantifies solidification cracking susceptibility in terms of
temperature rather than simply a crack length. As a result, it is directly
related to the brittle temperature range (BTR) which is commonly used to
describe the temperature range over which solidification cracking occurs.
The SCTR can be calculated by the following equation:

MCD (2)
SCTR = ∗ CR
VW

where: MCD = maximum crack distance at saturated strain (Fig. 2),

Vw = weld travel speed,
CR = cooling rate in the solidification temperature range.

By using the SCTR approach both an order ranking and quantitative

measure of weld solidification cracking can be obtained. Additional details
and SCTR approach using the transverse Varestraint test can be found
elsewhere [2].
 Weld Solidification Cracking of Ni-Base Alloys 151

Experimental Procedures

Materials

The Ni-base alloys tested in this investigation were in the form of filler
metals. The compositions of the filler metals, as provided by the
consumable manufacturers, are shown in Table 1.

Table 1. Chemical composition (wt %) of Ni-base filler metals

Element 617 625 Hast W Hast X Haynes 230W

Ni 56.48 63.9 64.57 47.99 60.58
Cr 21.4 22.1 5.52 20.99 21.9
Fe 0.38 0.9 5.35 19.54 0.8
Mo 8.91 8.98 23.62 8.44 1.58
Co 11.03 0.04 0.6 1.5 0.5
W – – – 0.44 13.78
Nb – 3.21 – – –
Ti 0.3 0.24 – – 0.01
Al 1.2 0.2 – – 0.41
Si 0.03 0.06 0.32 0.16 0.42
C 0.06 0.01 0.02 0.07 0.07

Sample Preparation

Samples for Varestraint testing were prepared by depositing the filler

metals into grooves machined in Alloy 600 base plates. The base plate was
0.375-inch (~9.5 mm) thick. The groove depth was 0.156-in. (~4 mm) and
was 0.75 in. (19 mm) wide at the surface and 0.5 in. (13 mm) wide at the
bottom of the groove. Using the gas tungsten arc welding (GTAW)
process, a minimum of four layers of weld metal was deposited to
minimize dilution between the filler metal and the base plate. In the four
plus weld layers, there were approximately 15 weld passes per groove.
Dilution of the filler metals by the Alloy 600 base metal in the final layer
was determined to be less than 10%.
Upon completion of the groove welds, the weld deposit was machined
flush with the base plate. The plates were then sectioned into 3.5 inch wide
by 8 inch long samples in preparation for Varestraint testing. A schematic
of the sample preparation process is shown in Fig. 3. The compositions of
the actual weld metal deposits are provided in Table 2.
152 J.C. Lippold et al.

1 – Initial Fill-in 2 – Machine Flat 3 – Varestraint

Fig. 3. Sample preparation and testing sequence. Grooves in Alloy 600 plate were
filled in four layers with approximately 15 total weld passes. The weld
reinforcement was then removed and Varestraint testing conducted within the
deposited weld metal

Table 2. Composition (wt %) of actual weld deposits

Element 617 625 Hast W Hast X Haynes 230W
Ni 60.1 65.4 63.2 52.5 63.1
Cr 19.9 20.8 8.53 20.06 20.82
Fe 2.58 1.38 4.84 17.53 1.8
Mo 7.07 8.35 16.84 7.53 1.31
Co 9.08 0.1 2.42 1.25 0.45
W 0.1 0.1 0.07 0.43 11.2
Nb 0.1 3.38 0.02 0.1 0.02
Ti 0.27 0.28 0.12 0.1 0.06
Al 1.07 0.32 0.38 0.6 0.36
Si 0.1 0.1 0.3 0.1 0.33
Mn 0.14 0.1 0.41 0.39 0.47
C 0.034 0.017 0.016 0.056 0.038

Varestraint Testing

The transverse Varestraint test was used to determine the solidification

cracking susceptibility of the Ni-base filler metals deposited on Alloy 600.
The gas tungsten arc welding (GTAW) parameters were established to
develop a weld bead that was approximately 10 mm wide on the surface.
The test conditions are listed in Table 3. Prior to testing, the samples were
thoroughly cleaned with methanol.
Augmented strains in the range from 0.5 to 7% strain were used to
evaluate the cracking susceptibility of the weld deposits. For each sample
both the maximum crack distance (MCD) and total number of cracks were
determined using a binocular microscope at magnifications up to 50X.
Three independent measurements were made for each sample.
 Weld Solidification Cracking of Ni-Base Alloys 153

The solidification cracking temperature range (SCTR) was determined

using the approach described in the Introduction by plunging a Type-C
thermocouple into the molten weld metal of a weld made using identical
parameters to those listed in Table 3. The cooling rate data collected by
this technique was then used to calculate the SCTR.

Table 3. Transverse Varestraint Test Conditions

Var iable Settings

Current 180 amps, DCRP
Voltage ~ 8 V (2.4 mm arc gap)
Electrode W-2ThO2, 2.4 mm diameter
Shielding Gas 100% argon
Shielding Gas Flow Rate 14.2 L/min
Gas Preflow Time 4s
Gas Postflow Time 10 s
Weld Length 50 mm
Bend Position 44 mm from weld start
Bend Orientation Transverse
Bend Rate 250 mm/s
Bend Hold Time 2 s after arc shut off
Weld Bead Width ~10 mm on surface

Characterization

Weld metal samples of 617, Hast X, and Haynes 230W were prepared for
metallographic analysis by electrolytic etching with 10% chromic acid at
2.5 V for 30–60 s. The 625 and Hast W samples were etched
electrolytically with 10% oxalic acid at 4 V for 30 s. Scanning electron
microscopy (SEM) and EDS analysis was performed on a Sirion
SEM/FEG and Phillips XL-30 ESEM/FEG at 15 kV.

Calphad Calculations

Equilibrium phase diagrams and Scheil-Gulliver solidification simulations

were calculated using Calphad-based software Thermocalc£ and Ni-base
alloys database developed by Thermotec£. The measured chemical
154 J.C. Lippold et al.

compositions of the actual weld deposits (Table 2) were used as input for
the calculations. The phases used for the calculations were liquid, δ, γ, γ´,
Laves, Ș, μ, σ, R-phase, MC, M6C, M7C3, and M23C6.

Results and Discussion

Calculated Phase Diagrams

Calphad-based calculated equilibrium phase diagrams for Ni-base alloys

based on Inconel 625 and Hastelloy X systems as a function of Mo content
are shown in Fig. 4. The vertical lines in these diagrams represent the
specific Mo content of Inconel 625 and Hastelloy X. As can be seen from
these diagrams, Ni-base systems are very complex and contain multiple
phases.
Since cooling from the solidification temperature in arc welds is
relatively rapid, these alloys do not reach equilibrium conditions.
Therefore, the phase diagrams and phase molar fraction evolution
presented as examples in Figs. 4 and 5 are just a first approximation to the
actual phase transformation behavior under non-equilibrium cooling
conditions. Despite this limitation, these diagrams are extremely important
when evaluating the material behavior under slower cooling conditions,
postweld heat treatment, and material service at elevated temperature.
According to the equilibrium calculations, all the filler metals in this
study undergo fully austenitic solidification with the exception of
Hastelloy X, which forms M6C at the end of the solidification. Table 4
summarizes the calculated solidus (TS) and liquidus (TL) temperatures,
equilibrium solidification temperature ranges and sequences for the weld
deposit compositions listed in Table 2.
During solidification there is a strong segregation that results in the liquid
enrichment in some elements and impoverishment of others. As a result of
this segregation, there are significant changes in the solidification
temperatures and path relative to the equilibrium diagrams. There is a simple
model that approximates the solidification path under non-equilibrium
conditions, the so-called Scheil-Gulliver solidification simulation. This
model is based on local equilibrium at the solid/liquid interface, complete
mixing of solute within the liquid, and no solid diffusion following
solidification (once formed the solid composition does not change). Scheil-
Gulliver solidification simulations were performed using ThermocalcTM
software and a summary of the results is presented in Table 4.
 Weld Solidification Cracking of Ni-Base Alloys 155

Alloy 625 L

L+γ
Temperature [°C]

γ + M6C
γ + M6C + ı
γ + MC
γ + M6C + ı + ȝ

γ + M6C + į + ȝ

γ + M23C6 + į γ + M23C6 + į + ȝ

γ + M23C6 +
γ + M23C6 + į + ȝ + γ’
į + γ’

Weight Percent Mo

Hastelloy X L + BCC

L+ı
L
L+γ + BCC
L+ı
L+ı
Temperature [°C]

+ M6C
γ
γ + M6C

γ + M6C + ı
γ + M23C6 γ + M6C + ı + γ’

γ + M6C + ı + ȝ

γ + M6C + ı + ȝ + γ’

Weight Percent Mo

Fig. 4. Calculated phase diagrams for Alloy 625 (top) and Hastelloy X (bottom)
systems as a function of Mo content. The vertical line represents the nominal Mo
content of each alloy
156 J.C. Lippold et al.

L
γ γ
Mole Fraction

Alloy 625

M23C6
ȝ M 6C
γ’ į

Temperature [oC]

L
γ
Mole Fraction

Hastelloy X

M23C6
ȝ
γ
γ’ σ M 6C

Temperature [oC]

Fig. 5. Calculated equilibrium evolution of molar fraction of phases of Alloy 625

and Hastelloy X at the Mo concentrations indicated in Fig. 4

The comparison of the equilibrium and non-equilibrium solidification

behavior in Tables 4 and 5, respectively, shows the effect of solidification
segregation on the both the solidification temperature range and solidification
path of the studied alloys. There are a number of eutectic reactions that occur
during the solidification of these alloys, causing a significant reduction in the
 Weld Solidification Cracking of Ni-Base Alloys 157

solidus temperature relative to equilibrium solidification. This expansion of

the solidification temperature range has an important impact on solidification
cracking susceptibility.

Table 4. Equilibrium solidification temperature ranges and solidification sequence

determined using Calphad-based ThermoCalc™ calculations
Equilibrium Solidification
Alloy
TS (°C) TL (°C) ΔT (°C) Predicted Solidification Sequence

Alloy 617 1339 1386 47 L → L+ γ → γ

Alloy 625 1275 1355 80 L → L+ γ → γ

Hastelloy W 1344 1385 41 L → L+ γ → γ

Hastelloy X 1333 1396 64 L → L+ γ → L+ γ + M6C → γ + M6C

Haynes
1365 1402 37 L → L+ γ → γ
230W

Table 5. Solidification Temperature Ranges and Solidification Sequence determined

using ThermoCalc™ Scheil-Gulliver Prediction
Scheil-Gulliver Simulation
Alloy
TS (°C) TL (°C) ΔT (°C) Predicted Solidification Sequence

L → L+ γ → L+ γ + M6C → L+ γ +
Inconel 617 1226 1386 160
M6C + δ → γ + M6C + δ
L → L+ γ → L+ γ + MC → L+ γ +
Inconel 625 1112 1355 243 MC + δ → L + γ + MC + δ +σ → γ +
MC + δ +σ
L → L+ γ → L+ γ + M6C → L+ γ +
Hastelloy W 1060 1385 325
M6C + μ → γ + M6C + μ
L → L+ γ → L+ γ + M6C → L+ γ +
Hastelloy X 1236 1396 160 MC + M6C + σ → γ + MC + M6C +
σ
L → L+ γ → L+ γ + M6C → L+ γ +
Haynes
1277 1402 125 M6C + M23C6 → L+ γ + M23C6 → L+ γ
230W
+ M6C + M23C6 → + γ + M6C + M23C6
158 J.C. Lippold et al.

Weld Metal Characterization

The alloys examined in this study solidify as austenite (face centered cubic
crystal structure). Austenite solidification promotes strong segregation
resulting in local variations in composition in the weld metal. Due to this
segregation, many of these weld metals form a terminal eutectic
constituent (or constituents), as predicted by the Scheil-Gulliver
simulations. Figs. 6–8 show the solidification structure of the alloys
examined in this study. Included with the micrographs are composition
spectra (measured with SEM/XEDS) showing the composition of both the
austenite matrix and interdendritic eutectic constituent.

Fig. 6. Micrographs showing Alloy 617 (top) and Alloy 625 (bottom)
solidification structure and XEDS spectra of matrix and eutectic compositions
 Weld Solidification Cracking of Ni-Base Alloys 159

Fig. 7. Optical micrograph showing Hastelloy W (top) and Hastelloy X (bottom)

solidification structure and XEDS spectra of the matrix and interdendritic regions

Fig. 8. Optical micrograph showing Haynes 230W solidification structure and

XEDS spectra of matrix composition
160 J.C. Lippold et al.

An example of multiple phases forming during solidification can be

seen in Fig. 9, where MC (NbC) forms by a eutectic reaction during the
terminal solidification of Alloy 625. The MC eutectic phase is the first to
form along the interdendritic boundaries as predicted by Scheil-Gulliver
simulations. A second phase can be seen surrounding the MC particles.
This is the result of “Z-contrast” in the SEM due to the difference in
atomic number (composition). This phase is most likely the δ phase
predicted by Scheil-Gulliver, which is rich in the elements Nb and Mo.
The MC reaction resulted from the presence of C and segregation of Nb
during solidification of Alloy 625 (Nb = 3.4 wt%).

Fig. 9. Back-scattered electron SEM image (1000X) and XEDS spectra of NbC
and matrix composition of Alloy 625

Image analysis of photomicrographs taken at 400X magnification was

performed on each of the alloys to rank the amount of second phase
present in the weld metal microstructure. The area fraction of secondary
phases measured in each micrograph was assumed to be proportional to
the volume fraction. It is presumed that most of this second phase
comprises the eutectic constituents that form at the end of solidification.
Table 6 summarizes the characteristics of eutectic formation for each alloy
including the amount of eutectic formed during solidification and the
elements that were enriched in the eutectic constituents as measured by
XEDS.
Weld metal produced during Varestraint testing of all five filler metals
solidified with a cellular dendritic solidification structure. Primary dendrite
arm spacing (DAS) and secondary dendrite arm spacing (SDAS) were
measured at locations near the weld centerline of samples tested at 3%
strain. Solidification rate along the weld centerline can be approximated by
 Weld Solidification Cracking of Ni-Base Alloys 161

the travel speed. The measured values of DAS and SDAS (shown in Table
6) were comparable for each of the alloys indicating similar solidification
behavior.

Transverse Varestraint Results

Both the maximum crack distance (MCD) and total number of cracks
(TNC) were determined from the tranverse Varestraint test samples for the
five filler metals. The results from this analysis are shown in Figs. 10–12.
Except for Hastelloy W, cracking was observed at even the lowest applied
strain (0.5%). Based on these results, the 617 and Haynes 230W filler
metals exhibited the lowest cracking susceptibility. The MCD results for
all five filler metals are compiled in Fig. 13.

Table 6. Summary of weld metal microstructure characteristics

Area Elements
SDAS
Alloy Fraction of Enriched in DAS (ȝm)
(ȝm)
2nd Phase Eutectic
617 0.045 Al, Si, Mo, Ti, Cr 17.9 7.4
625 0.13 C, Si, Nb, Mo 16.8 6.4
Hast X 0.082 Al, W, P, Mo, Cr 17.6 6.3
Hast W 0.011 C, Si, P, Mo 17.8 6.7
230W 0.002 Not evaluated 17.6 5.9

Using the technique described in the Introduction, the solidification

cracking temperature range (SCTR) was determined based on weld metal
cooling rate data collected by plunging thermocouples into the molten
weld metal. Based on this data, the SCTR values for the five filler metals
are plotted in Fig. 14. The SCTR data from an autogenous GTA weld in
Type 304L stainless steel with Ferrite Number (FN) 6 is included for
comparison. This weld metal is known to be extremely resistant to weld
solidification cracking. In general, SCTR values below 50°C are consistent
with weld metals that are very resistant to weld solidification cracking.
Based on SCTR values below 100°C, Alloy 617 and Haynes 230W
would be the most resistant to weld solidification cracking. The SCTR
results for Alloy 625 and Hastelloy X would predict that these filler metals
have the highest susceptibility. The results for Alloy 625 are not consistent
with actual fabrication experience with this filler metal, since it is
generally considered to have good resistance to cracking. This
inconsistency will be discussed later.
162 J.C. Lippold et al.

IN617 Weld Deposit

3 50
MCD 45
Maximum crack distance (mm)

2.5 TNC
40

Total number of cracks

35
2
30
1.5 25
20
1
15

10
0.5
5

0 0
0 1 2 3 4 5 6 7 8
Applied strain (%)

IN625 Weld Deposit

3 50
MCD 45
Maximum crack distance (mm)

2.5 TNC
40
Total number of cracks

35
2
30

1.5 25
20
1
15

10
0.5
5
0 0
0 1 2 3 4 5 6 7 8
Applied strain (%)

Fig. 10. Transverse Varestraint results for Alloys 617 and 625 filler metals
 Weld Solidification Cracking of Ni-Base Alloys 163

Hastelloy W Weld Deposit

3 50
MCD 45
Maximum crack distance (mm)

2.5 TNC
40

Total number of cracks

35
2
30

1.5 25

20
1
15
10
0.5
5

0 0
0 1 2 3 4 5 6 7 8
Applied strain (%)

Hastelloy X Weld Deposit

3 50
MCD 45
Maximum crack distance (mm)

2.5 TNC
40
Total number of cracks

35
2
30

1.5 25

20
1
15

10
0.5
5
0 0
0 1 2 3 4 5 6 7 8
Applied strain (%)

Fig. 11. Transverse Varestraint results for Hastelloy W and Hastelloy X filler
metals
 164 J.C. Lippold et al.

Haynes 230W Weld Deposit

3 50
MCD 45
Maximum crack distance (mm)

2.5 TNC
40

Total number of cracks

35
2
30

1.5 25
20
1
15
10
0.5
5
0 0
0 1 2 3 4 5 6 7 8
Applied strain (%)

Fig. 12. Transverse Varestraint results for Haynes 230W filler metal

3.5
IN617
IN625
3
Hastelly X
Hastelloy W
Maximum crack distance (mm)

2.5 Haynes 230W

1.5

0.5

0
0 1 2 3 4 5 6 7 8
Applied strain (%)

Fig. 13. Summary of maximum crack distance (MCD) versus strain for all the
filler metals tested
 Weld Solidification Cracking of Ni-Base Alloys 165

250

Hast X

625
Hast W
200 210
190

Haynes 230
SCTR, °C

150
145
Type 304L, FN6

617

100
95
85

50

31
0

Fig. 14. Solidification cracking temperature range for the filler metals tested. Type
304L with Ferrite Number (FN) 6 is listed for comparison

Comparison of Non-equilibrium Solidification and SCTR

Actual weld solidification temperature ranges were measured using the

SS-DTA technique [3, 4, 5]. This technique uses thermocouples plunged
directly into the weld pool to measure the cooling rate from above the
liquidus to room temperature. The thermal history is then analyzed to
determine the liquidus and solidus temperatures, and the solidification
temperature range is calculated.
The comparison of the Scheil-Gulliver simulations, SS-DTA
measurements, and Varestraint SCTR are shown in Table 7. The Scheil-
Gulliver and SS-DTA values are in reasonably good agreement for Alloys
617, 625, and Haynes 230W. For Hastelloy alloys X and W, the simulation
overestimated the measured solidification temperature range.
Based on solidification cracking theory, the SCTR should be a fraction
of the solidification temperature range, since cracking will occur only
when substantial solid has formed. Thus, the SCTR values for Alloy 625,
and Haynes 230W seem most appropriate when compared to both SS-DTA
166 J.C. Lippold et al.

and Scheil-Gulliver values. For Alloy 617 and Hastelloy W, the SCTR and
SS-DTA values are nearly equivalent, while the SCTR for Hastelloy X is
greater than both the simulated and measured solidification temperature
range. Since these materials are also susceptible to ductility dip cracking
(DDC), it is possible that some portion of the crack length measured in the
Varestraint test is actually DDC extension off the solidification crack. This
would result in an artificially high value of the SCTR. Additional analysis
is ongoing to rationalize these differences.

Table 7. Comparison of non-equilibrium solidification temperature range determined

by Scheil-Gulliver and single sensor differential thermal analysis (SS-DTA) with the
SCTR measured by the Varestraint test

Solidification Range (°°C) Var estr aint

Weld Metal
Scheil-Gulliver SS-DTA SCTR (°°C)
617 160 93 88
625 243 97 (306)* 205
Hastelloy W 325 162 142
Hastelloy X 160 108 191
Haynes 230W 125 139 98
* Value in parentheses includes end of eutectic solidification

Effect of Backfilling on Cracking Susceptibility

The SCTR determined using the Varestraint test is based on the MCD at
saturated strain. In the case of the weld metals tested here, the saturated
strain at which SCTR was determined was 5%. At such high applied
strains, the effect of crack backfilling by excess liquid at the end of
solidification can be reduced or eliminated. Under normal welding
conditions, where actual strains in the solidifying weld metal may be much
lower, backfilling may in fact be effective in reducing cracking
susceptibility.
Photomicrographs taken near the crack tips of Varestraint samples from
Alloy 625 and Haynes 230W are shown in Fig. 15. Note that in Alloy 625
considerable residual liquid (arrows in Fig. 15A) is observed near the
crack tips. This represents liquid that was along the solidification grain
boundaries at the end of solidification and is drawn into the tips of the
cracks by capillary attraction during testing at high strain levels. In
contrast, virtually no residual liquid is observed near the crack tips of the
Haynes 230W weld metal (Fig. 15B).
 Weld Solidification Cracking of Ni-Base Alloys 167

This residual liquid effect is also observed when examining the fracture
surface of the solidification cracks in these weld metals, as shown in Fig. 16.
In both cases, the characteristic dendritic nature of solidification cracking
is evident. However, in the case of Haynes 230W, the dendritic fracture
morphology is much more distinct. This occurs since there is little residual
liquid at the end of solidification and the solidifying dendrites are covered
by only a thin layer of liquid. In contrast, the large amount of terminal
liquid during solidification of Alloy 625 acts to coat the dendritic structure
and make it less distinct.

A)

B)

Fig. 15. Microstructure near the crack tip of a solidification crack in (A) Alloy
625, and (B) Haynes 230W. Note difference in magnification
168 J.C. Lippold et al.

(A)

(B)

Fig. 16. Solidification crack surface of (A) Alloy 625, and (B) Haynes 230W.
Note difference in magnification
 Weld Solidification Cracking of Ni-Base Alloys 169

Under lower applied strain conditions, the terminal liquid in the Alloy
625 weld metal would be available to heal any solidification cracks that
would form, improving the weld solidification cracking resistance relative
to what the SCTR ranking predicts. This may explain why Alloy 625 is
generally considered to have good resistance to weld solidification
cracking in actual practice, particularly at low to moderate restraint levels
where backfilling can be effective in healing cracks that form..
For Alloy 625, differences between the solidification cracking
susceptibility determined using the Varestraint test and actual fabrication
experience has been reported by other investigators [6, 7]. This suggests
that for weld metals that achieve resistance to solidification cracking
through a crack backfilling mechanism, the Varestraint test may not
accurately reflect their susceptibility to cracking. The high applied strains
used during this test force cracking to occur even in cases where
considerable terminal liquid is present. The use of a threshold strain for
cracking rather than MCD at saturated strain may be a more appropriate
measure of susceptibility for these weld metals.

Conclusions

1. Solidification cracking susceptibility of the five Ni-base filler metals

as determined by the solidification cracking temperature range (SCTR)
is as follows (lowest to highest): Alloy 617, Haynes 230W, Hastelloy
W, Hastelloy X, and Alloy 625.
2. Good correlation was found among the SCTR, simulated Scheil-
Gulliver solidification range, and SS-DTA values for Alloy 617 and
Haynes 230W. These filler metals exhibited the highest susceptibility
to solidification cracking.
3. Correlation between the simulated solidification range and the range
measured by SS-DTA was poor for both Hastelloy alloys W and X. In
both cases the simulation overestimated the actual range.
4. Alloy 625 weld metal was determined to be the most susceptible to
solidification cracking, even though fabrication practice has shown
that this filler metal is quite resistant to cracking. It appears that the
crack “healing” mechanism may be responsible for this contradiction.
5. Additional research is needed to relate calculated and measured
solidification temperature ranges, and to effectively use these to
predict solidification cracking susceptibility.
170 J.C. Lippold et al.

Acknowledgements

The authors wish to thank Siemens-Westinghouse for providing the filler

metals used in this investigation. The Varestraint testing and
characterization was carried out as part of a senior project at OSU. We
extend our thanks to Jeff Bernath and Matthew Slate (along with Jeff
Sowards) who participated in that project.

References

1. Savage, W.F. and C.D. Lundin, The Varestraint Test, Welding Journal,
October 1965, pp. 433s–442s.
2. J.C. Lippold, 2005. Recent Developments in Weldability Testing, 1st
International Workshop, Hot Cracking Phenomena in Welds, Berlin, March
2004, publ. by Springer-Verlag, pp. 271–290.
3. Alexandrov B.T. and J.C. Lippold, Single Sensor Differential Thermal
Analysis of Phase Transformations and Structural Changes during Welding
and Postweld Heat Treatment, IIW Doc. IX-2199-06, 59th Annual Assembly
of IIW, Quebec City, Canada, August 2006, accepted for publishing in
Welding in the World.
4. Alexandrov B.T. and J.C. Lippold, Relationship Between the Solidification
Temperature Range and Weld Solidification Cracking Susceptibility of
Stainless Steels and Ni-base Alloys, IIW Doc. IX-2163-05, 58th Annual
Assembly of IIW, Prague, July 2005, www.iiw-iss.org.
5. Alexandrov B.T. and J.C. Lippold, In-Situ Weld Metal Continuous Cooling
Transformation Diagrams, Welding in the World, Vol. 50, No. 9/10, 2006, pp.
65–74 (Doc. IIW 1744-06, ex-doc. IX-2162-05).
6. Lingenfelter, A.C., Varestraint Testing of Nickel Alloys, Welding Journal,
September 1972, pp. 430s–436s.
7. Karlsson, L., E.L. Berquist, S. Rigdal, and N. Thalberg, Evaluating Hot
Cracking Susceptibility of Ni-base SAW Consumables for Welding of 9% Ni
Steel. Proceeding of 2nd International Workshop on Hot Cracking, Berlin,
March 2007.
Hot Cracking Susceptibility of Ni-Base Alloy
Dissimilar Metal Welds

H. Hänninen, A. Brederholm, T. Saukkonen

Helsinki University of Technology, Espoo, Finland

Abstract

Hot cracking susceptibility of Ni-base alloy dissimilar metal welds in nu-

clear power plant (NPP) applications was studied by preparing mock-ups
of various relevant dissimilar metal welds of the NPPs. The motivation for
the study was the need for repair welding of Alloy 182 welds of safe ends
in BWR plants and possible welding problems with new Alloy 152/52
weld metals for new reactors. Weldability of the Ni-base alloys (Alloys
182/82 and Alloys 152/52) was evaluated based on their composition and
weld metal microstructures. Susceptibility to hot cracking was examined
by Varestraint tests. The studied dissimilar metal welds showed clear seg-
regation of Nb, Si, P and Mn to the last liquid to solidify at dendrite
boundaries. Hot cracking occurred along the dendrite boundaries and the
susceptibility to hot cracking in the weld mock-up samples was observed
to follow the order: Alloy 152 > Alloy 52 > Alloy 182 > Alloy 82, while in
pure (undiluted) weld metal hot cracking tests the susceptibility followed
the order: Alloy 182 • Alloy 152 > Alloy 52 • Alloy 82. The differences
are thought to be related to dilution effects in mock-up welds because Fe,
Si, and C enhance the eutectic phases and expand the solidification tem-
perature range.

Introduction

The Ni-base alloy dissimilar metal welds are typically made using Alloy
182 and Alloy 82. Recently, Alloy 52 has been used both in new construc-
tions as well as in repair welding. The trend towards alloys with higher
contents of chromium is driven by the observed environment-assisted
cracking (EAC) in Alloy 182, and recently also in Alloy 82. One driving
force towards the more EAC resistant alloys is also the challenges and
172 H. Hänninen et al.

costs related to non-destructive examination of dissimilar metal welds in

nuclear power plant applications.
Several studies have been made to investigate hot cracking susceptibil-
ity of filler metal Alloys 82, 182, 52 and 152 and base material Alloy 690
[1, 2, 3]. Hood and Lin [1] studied hot cracking susceptibility of Ni-base
alloy filler metals (Alloys 82, 182, 52 and 152) using the Varestraint and
spot-Varestraint tests. The results showed that the hot cracking resistance
of the four filler metals deposited on Alloy 690 and carbon steel A285 is
similar or better than that for 1¼Cr-½Mo steel and Alloy 690-AISI 316L
base metal combinations. Alloy 52 exhibited the best resistance to both
weld solidification cracking and weld metal liquation cracking, followed
by Alloys 82, 152 and 182. The AISI 316LN steel/Alloy 52 combination
exhibited better resistance than Alloy 690/Alloy 52. Wu and Tsai [2] in-
vestigated the hot cracking susceptibility of Alloys 82 and 52 with Alloy
690 by Varestraint tests. Alloy 82 showed greater total crack length (TCL)
than Alloy 52. Liquation cracks and ductility dip cracks were found in the
HAZ, but an accurate distinction of the crack type was difficult. These re-
sults indicate that hot cracking susceptibility of Alloy 82 is greater than
that of Alloy 52 in Varestraint testing.
Nishimoto et al. [4] investigated the microcracking behaviour in TIG-
welded multi-pass Alloy 690 weld metal by spot- and trans-Varestraint
tests using three different filler metals of Alloy 52 by varying the impurity
contents of P and S. The results of the study showed that the microcrack-
ing susceptibility decreased with decreasing impurity element content. The
microcracking susceptibilities of ductility-dip, liquation and solidification
cracking were much improved by adding 0.01–0.02 mass% La to the weld
metal. However, excessive La addition to the weld metal led to liquation
and solidification cracking. The ductility-dip cracking susceptibility was
improved by the amelioration of hot ductility of reheated weld metal. The
excessive La addition resulted in the formation of Ni-La intermetallic
compound, and therefore, liquation and solidification cracks can occur in
the weld metal attributed to the enlargement of brittle temperature range
(BTR) and/or the occurrence of local liquation during the reheating. It was
concluded that the microcracking susceptibility in multi-pass welding of
Inconel 690 was improved drastically by La addition to the filler metal [4].
Recently a Gleeble-based test technique, termed strain-to-fracture
(STF), has been developed by Nissley et al. [5], and employed for evalua-
tion strain-to-fracture of ductility-dip cracking (STF DDC) susceptibility
curves for filler metals of Alloys 82 and 52 [3, 5]. The filler metal Alloy
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 173

52 was more susceptible to DDC than Alloy 82, which also exhibited heat-
to-heat variation in susceptibility. Thus, because of varying observations in
various tests there is a major need to rank the Ni-base alloy weld metals
for their hot cracking susceptibility and to understand the mechanisms of
solidification cracking properly.

Materials and Experimental Methods

In this study the hot cracking tendency of six different dissimilar metal
weld joints was tested by Varestraint testing [6]. The Varestraint equip-
ment, Fig. 1, was built and the obtained hot cracking and weld microstruc-
tures were studied by optical and scanning electron microscopy (FEG-
SEM/EDS). The studied materials and welds are typical for nuclear power
plant and oil and gas industry applications. In Table 1 the chemical com-
positions of the base materials and filler metals used in the manufacturing
of the studied dissimilar metal welds are presented.

Linear drive

Programming unit
Video recorder
TIG torch

Power source
Automatic voltage
control (AVC)

Display
Pneumatic sylinder (18 kN)

Fig. 1. General view of Varestraint testing equipment built for hot cracking testing
 174

Table 1. Chemical compositions of the base materials and filler metals used in the dissimilar metal welds (wt.-%)
C Si Mn P S Cr Mo Ni Nb Ti Fe Al Cu
H. Hänninen et al.

Base materials
42CrMo4 0,42 0,7 0,4 1,05 0,225
AISI 321 < 0,08 17-19 9-12 5*C-0,7
AISI 316 NG < 0,03 16,5-18,5 2,5-3 11-14
SA508 (Grade 2) 0,24 0,21 1,38 0,009 0,16 0,54 0,58 0,015 0,12
Inconel 600 0,07 0,19 0,21 0,007 0,001 16,3 72,86 Nb+Ta 0,10 0,28 9,44 0,229 0,13
Inconel 690 0,02 0,04 0,16 0,001 29,46 59,82 0,33 9,96 0,2 <0,01
Filler metals
E 347 (OK 61.81) 0,07 0,7 1,7 0,021 0,006 19,8 0,1 10 0,61 0,14
AISI 347 (OK Tigrod 16.11) 0,04 0,85 1,4 0,02 0,01 19 0,02 9,5 0,55 0,03
E 347 (OK Band 11.62) < 0,025 0,4 1,8 20 10 0,7
AISI 308L (OK Band 11.61) < 0,025 0,4 1,8 20 10
AISI 309L (OK Band 11.65) < 0,025 0,4 1,8 24 13
Inconel 182 (OK 92.26) 0,03 0,8 6,5 0,01 0,003 15,7 68 1,8 0,1 6,7 < 0,01
Inconel 82 (OK Tigrod 19.85) 0,039 0,03 2,98 0,001 0,004 19,94 72,6 Nb+Ta 2,47 0,34 1 0,01
Inconel 52 0,03 0,13 0,24 <0,001 <0,001 29,2 0,03 59,28 Nb+Ta <0,02 0,51 9,8 0,72 0,04
Inconel 52M 0,02 0,09 0,8 0,003 0,001 30,06 0,01 59,54 Nb+Ta 0,83 0,224 8,22 0,11 0,02
Inconel 52MS 0,014 0,12 0,68 0,004 0,0007 29,53 0,02 60,14 0,78 0,19 8,33 0,13 0,03
Inconel 152 0,048 0,41 3,48 0,003 0,003 28,74 0,01 55,2 Nb+Ta 1,54 0,09 10,39 0,06 <0,01
Inconel 152M 0,0293 0,3276 3,247 <0,001 0,0042 28,98 <0,005 57 Nb+Ta 1,533 0,039 8,75 <0,005 <0,005
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 175

Figure 2 presents the groove geometry, materials and cross-section of

Neste Oil mock-up dissimilar metal weld. Inconel 182 buttering side of the
test specimen was heat treated at 700°C for 7 h. Figures 3–6 present the
groove geometries, materials and cross-sections of mock-up TV, 1, 2 and 3
dissimilar metal welds for nuclear applications. The clad pressure vessel
steel SA 508 side of specimen TV (with Alloy 182 buttering) was heat
treated at 610°C for 6 h and mock- up 3 (with Inconel 52 and Inconel 152
buttering) was heat treated at 610°C for 4 h before welding. Whole welded
test specimens of mock-ups 1 and 2 were heat treated at 610°C for 16 h af-
ter welding.
50°
4 4
Buttering
Inconel 182
8 mm

80 mm Inconel 182
42 CrMo4

Inconel 42CrMo4
Inconel
82 Inconel
E 347 182
82
E 347
E 347
E 347

Inconel 82

AISI 321
E 347 E 347

Fig. 2. Groove geometry, materials and cross-section of Neste Oil mock-up dis-
similar metal weld. Length of the specimen is 300 mm

50 °

8 mm

2,5
80 mm

SA 508 Inconel 600

Inconel 182

AISI 309 Inconel 182

AISI 308

Fig. 3. Groove geometry, materials and cross-section of mock-up TV dissimilar

metal weld. Length of the specimen is 300 mm
176 H. Hänninen et al.

50 °

8 mm

2,5

80 mm

SA 508 AISI 316 NG

Inconel 152

AISI 309
AISI 308

Fig. 4. Groove geometry, materials and cross-section of mock-up 1 dissimilar

metal weld. Length of the specimen is 300 mm

16 °

8 mm
1,5

80 mm

SA 508 AISI 316 NG

Inconel 52

AISI 309
AISI 308

Fig. 5. Groove geometry, materials and cross-section of mock-up 2 dissimilar

metal weld. Length of the specimen is 300 mm

4 4 45 °

8 mm
1,5

2,5
80 mm

Inconel
Inconel 690
SA 508 152

Inconel
52 Inconel
152

Fig. 6. Groove geometry, materials and cross-section of mock-up 3 dissimilar

metal weld. Length of the specimen is 300 mm
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 177

Figure 7 presents the test piece for Varestraint testing of pure weld met-
als and weld pass sequence of the test specimen welded with Alloy 82.
Base material was AISI 304 plate into which a 20 mm wide, 60 mm long
and 4 mm deep groove was machined and then welded with different filler
metals. The surface of the weld was ground to the level of the base metal
so that the surface quality was equal in all Varestraint tests. Filler metals
(supplier in brackets) used in the pure weld metal tests were: Alloy 182
(Esab OK 92.26), Alloy 82 (Esab Tigrod 19.85), and Alloys 152, 152M,
52, 52M, and 52MS (Special Metals). Table 2 presents the welding pa-
rameters used in the Varestraint testing.

20
Welding
filling with
with filler filler
material metal

8 AISI 304
4

60

300

60

Fig. 7. Test piece for Varestraint testing of pure weld metals and weld pass se-
quence of the test specimen welded with filler metal Alloy 82
178 H. Hänninen et al.

Table 2. Summary of the welding parameters used in the Varestraint tests

Current Voltage Welding Heat input
[A] [V] speed [kJ/cm]
[cm/min]
Neste Oil mock-up (Inconel 182) 250 12,5 12 9,4
Mock-up TV (Inconel 182) 150 10,5 12 4,7
Mock-up 1 (Inconel 152) 150 10,5 12 4,7
Mock-up 2 (Inconel 52) 150 10,5 12 4,7
Mock-up 3 (Inconel 52/152) 150 10,5 12 4,7

Experimental Results

Figure 8 (a) presents the surface (Alloy 182 weld, augmented strain 4%) of
Neste Oil mock-up with hot cracks after Varestraint test. Optical microscopy
of a hot crack (dark areas) in the cross-section of the weld (b) and scanning

6 mm

a) b)

c) d)
Fig. 8. (a) Surface (Alloy 182 weld, augmented strain 4%) of Neste Oil mock-up
with hot cracks after Varestraint test. (b) Optical micrograph of the hot cracks
(dark areas) in a cross-section of the weld (note the white phase ahead of the hot
crack tip). (c), (d) Scanning electron micrographs (SEM) of weld hot crack tips
showing details of the irregular shaped white phase
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 179

electron microscopy (FEG-SEM) of weld hot crack tips (c) and (d) show
details of the irregular shaped white phase connected to hot cracks of this
alloy. The composition of the white phase areas was examined by EDS
analysis. Figure 9 presents X-ray element maps of the white phase area in a
dendrite boundary region and Fig. 10 shows a SEM image and EDS
analyses of the white phase showing marked Si, Nb, Mn and P enrichment,
which is compensated by depletion of Cr, Fe and Ni. Based on the high
magnification BSE-images and the EDS analyses the observed white phase
consists of two phases, white and grey, with varying Nb, Si, Mn and Ti
contents. Inside and along the dendrite/grain boundaries white Nb-rich
particles, Nb(C, N), are present.

Fig. 9. X-ray element maps of the white phase area in a dendrite boundary region.
Note marked Nb enrichment in the white phase and less enrichment in a wider
area along the dendrite boundary. Marked Si, Mn and P enrichment compensated
by depletion of Cr, Fe and Ni can also be seen
180 H. Hänninen et al.

Fig. 10. SEM image and EDS analyses of the white and grey phase showing Nb,
Si, Mn and P enrichment and Nb rich particles, Nb(C, N). The spot size in EDS
analysis has an interaction diameter about 1 μm, and therefore the surrounding
matrix affects also the result

The hot crack fracture surfaces were opened for fractography and phase
analysis with an FEG-SEM/EDS system. Figure 11 shows an opened fracture
surface of Alloy 182. With a low acceleration voltage of 7 kV the informa-
tion comes from close to the surface. The wavy morphology of the fracture

a) b)
Fig. 11. Fracture surface of Alloy 182 hot crack showing the white phase in the
valleys between dendrites
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 181

surface can be clearly seen, which is an evidence of the molten phase along
the dendrite boundaries in which the separation has occurred.
Study of the fracture surfaces of the Neste Oil mock-up weld (Alloy
182) hot cracks revealed marked segregation of the same elements on frac-
ture surfaces as seen in the white phase in the cross-sections of the hot
cracks. On the fracture surface Nb-rich particles, Nb(C, N), and the lamel-
lar white phase are present intermixed in the final microstructure. The
white phase exhibits similar enrichment of Nb, Si, Mn, and P as observed
in the cross-sections of the weld metal.
Pure Alloy 82 weld metal hot cracking took place only at higher heat
input (Q = 9,4 kJ/cm, augmented strain 4%). The formed hot cracks were
markedly oxidized, probably because of the high heat input and also be-
cause of element segregation to the final melt. Figure 12 presents a thick ox-
ide layer on the fracture surface of Alloy 82 weld metal hot crack. Table 3
presents EDS analyses of three areas of the oxide layer on the hot crack
surface presented in Fig. 12. The oxide layer is enriched in Mn, Nb, Fe,
and Ti.

Fig. 12. Oxide layer on the hot crack fracture surface of Alloy 82 weld metal

Table 3. EDS analyses of three areas of the oxide layer on the hot crack surface
presented in Fig. 12 (wt.-%)
Element O Al Si Ti Cr Mn Fe Ni Nb
Area 1 23,4 0,02 0,18 1,2 8,6 13,4 11,7 30,1 11,6
Area 2 1,4 0,03 0,17 0,3 8,9 2,5 5,9 77,2 3,6
Area 3 1 0,04 0,15 0,3 10 2,2 5,2 79,2 1,9

Figure 13 (a) presents SEM image and (b) X-ray element maps of a
cross-section of hot cracks exhibiting a strong Nb segregation and less
marked Si, P and Ti segregation to the dendrite boundary at the hot crack
182 H. Hänninen et al.

tip. Figure 14 shows SEM image and X-ray element map of fracture surface
of Alloy 82 exhibiting distribution of Nb and Ti.

a)

O Si P

Ti Cr Mn

Fe Ni Nb
b)
Fig. 13. (a) SEM image and (b) X-ray element maps of a cross-section of hot
crack exhibiting strong Nb segregation and additional segregation of Si, P and Ti
to the dendrite boundary along and ahead of the hot crack
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 183

a) b)
Fig. 14. (a) SEM image and (b) X-ray element map of fracture surface of Alloy 82
showing distribution of Nb and Ti

The chemical composition of the oxide layer on the hot crack fracture
surface of Alloy 82 shows marked increase of Nb, Mn and Fe contents,
which are the elements of a high tendency for oxidation and are available
on the fracture surface of a hot crack. Under the oxide layer the white
phase rich in Nb, Mn, Si and P as well as the Nb- and Ti-rich (Ti content
of Alloy 82 0,34 wt.-%) particles, Nb, Ti(C, N), are also clearly visible. In
the pure weld metal tests it was also observed that cracking took place
under the weld bead made in the Varestraint test by a liquation cracking
mechanism following the boundaries of lower weld passes (see Fig. 7).
Figure 15 presents SEM images of a cross-section of hot cracks in mock-
up 1 weld metal (Alloy 152 weld, augmented strain 4%) after Varestraint
test and Fig. 16 shows SEM images of fracture surface, where the presence
of the white phase in the valleys between the dendrites is clearly visible.
SEM image and X-ray element map of the hot crack fracture surface of
Alloy 152 exhibits high enrichment of Nb to the white phase (see Fig. 17).

a) b)
Fig. 15. SEM images of a cross-section of hot cracks in Alloy 152 weld metal
exhibit the connection of the white phase to hot cracks
184 H. Hänninen et al.

a) b)
Fig. 16. SEM images of hot crack fracture surface of Alloy 152 weld metal. Note
the brittle nature of the lamellar white phase

a) b)
Fig. 17. (a) SEM image and (b) X-ray element map of the fracture surface of Al-
loy 152 exhibit high enrichment of Nb to the white phase

Alloy 152 weld metal shows similar hot cracking tendency as compared
to Alloy 182. The Nb-rich white phase is connected to the hot cracks and
the phase can be easily found in the cross-sections of the hot cracks as well
as on the fracture surfaces (local coverage can be as high as 30%, see Fig.
16). The dendrite boundaries containing the white phase show in addition
to Nb a marked increase of Si, Mn and P contents both in the white phase,
but also in the wider zone along the dendrite boundaries. Simultaneously
depletion of Cr, Ni and Fe is observed in these zones. In addition to the
white phase both in the cross-section as well as on the fracture surfaces
plenty of small particles, Nb, Ti(C, N), are present.
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 185

Filler metal Alloy 52 contains Ti (0,51 wt.-%), but does not contain Nb
as Alloys 182, 82 and 152 and, thus, hot cracks form along the dendrite
boundaries without the presence of the Nb-rich white phase. Figure 18 pre-
sents SEM images of mock-up 2 weld metal (Alloy 52, augmented strain
4%) hot cracks after Varestraint test. A lot of small particles (TiN) are pre-
sent along the dendrite boundaries. Figure 19 presents a SEM image and
X-ray element maps of the hot crack fracture surface.

Fig. 18. SEM images of mock-up 2 (Alloy 52, augmented strain 4%) weld hot
crack cross-sections after Varestraint test

Hot cracking of Alloy 52 does not show any indications of the Nb-
rich white phase observed in the earlier cases with Alloys 182, 82 and
152, since Alloy 52 does not contain Nb. Alloy 52 contains Ti and
therefore a lot of precipitation of TiN(C) was observed both in the
cross-sections along the dendrite boundaries as well as on the fracture
surfaces of the hot cracks, where the coverage of the hot crack fracture
surface by TiN(C) can be quite extensive. The dendritic large TiN(C)
phase particles are presumed to have formed in the melt prior to final
solidification of the weld. On the fracture surface also MnS phase
particles are observed as separate phases. Thus, in Alloy 52 the only
marked phase on the hot crack surfaces is TiN(C).
186 H. Hänninen et al.

a)

b)
Fig. 19. (a) SEM image and (b) X-ray element maps of fracture surface of Alloy
52 revealing the dendritic phase of TiN(C), and separate MnS particles are also
present

Results of Varestraint Tests

Total crack lengths (TCL) of Neste Oil mock-up, mock-up TV and mock-ups
1, 2 and 3 are shown in Fig. 20 (a) and maximum crack lengths (MCL) in
Fig. 20 (b), respectively, with augmented strains of 0%, 0.5%, 1%, 2%, 3%
and 4%. In the legend box markings are as follows: case, tension side and
filler metal. Table 4 presents summary of Varestraint tests (augmented strain
4%) made for pure Ni-base alloy weld metals (see, Table 1 and Fig. 7).
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 187

50
Neste Oil, surface, 182
45 Mock-up TV, root, 182

40 Mock-up 1, root, 152

Total crack length [mm]

35 Mock-up 2, root, 52

Mock-up 3, root, 152

30

25

20

15

10

0
0 1 2 3 4 5 6 7 8
Augmented strain [%]

a)
8
Neste Oil, surface,182
7,5
7 Mock-up TV, root,182
6,5 Mock-up 1, root, 152
Maximum crack length [mm]

6
Mock-up 2, root, 52
5,5
Mock-up 3, root, 152
5
4,5
4
3,5
3
2,5
2
1,5
1
0,5
0
0 1 2 3 4 5 6 7
Augmented strain [%]

b)
Fig. 20. Total solidification crack lengths (TCL) (a) and maximum crack lengths
(MCL) (b) as a function of augmented strain of Neste Oil mock-up, mock-up TV
and mock-up 1, 2 and 3 samples in Varestraint testing
188 H. Hänninen et al.

Table 4. Summary of Varestraint tests made for pure Ni-base alloy weld metals
(augmented strain 4%, heat input 4,7 kJ/cm, except for Alloy 82)
Inconel 182 (Q=4,7 kJ/cm) TCL = 19,3 mm / 2nd test -> cracks
Inconel 82 (Q=4,7 kJ/cm) no cracks / 2nd test -> no cracks
Inconel 82 (Q=9,4 kJ/cm) TCL = 32,9 mm
Inconel 152 (Q=4,7 kJ/cm) TCL = 19,7 mm / 2nd test -> cracks
Inconel 152M (Q=4,7 kJ/cm) TCL = 8,1 mm
Inconel 52 (Q=4,7 kJ/cm) no cracks / 2nd test -> no cracks
Inconel 52M (Q=4,7 kJ/cm) TCL = 3,1 mm / 2nd test -> no cracks
Inconel 52MS (Q=4,7 kJ/cm) no cracks / 2nd test -> no cracks

Discussion

Weldability of the studied Ni-base materials was evaluated based on the

results obtained with various mock-up welds of real components or antici-
pated weld designs of future plants and weld metals of different chromium
contents. The susceptibility to hot cracking was examined based on the
Varestraint test results by ranking the materials. The mechanistic under-
standing of the hot cracking susceptibility of Ni-base weld metals is still
largely missing and careful metallurgical studies of the solidification
mechanisms were carried out by a modern FEG-SEM/EDS instrument.
Marked segregation of alloying elements such as Nb, Si, P and Mn to the
last liquid to solidify at dendrite boundaries was observed and in Nb-
bearing alloys an eutectic Nb-rich Laves phase (white phase), (Ni, Cr,
Fe)2(Nb, Si, Mn, P, Ti), formed in the interdendritic regions containing Nb
as much as 15–30 wt.-%. The formation of the Laves phase is probably
enhanced by the presence of Si, Mn and P, which are also markedly segre-
gated to the white phase and/or interdendritic areas. Solidification cracking
occurs during the terminal stages of solidification, when a liquid film is
distributed along the interdendritic regions as a continuous film and the
shrinkage strains across these boundaries can not be accommodated. The
solidification temperature range and morphology of the interfacial liquid
are the primary factors controlling the solidification cracking susceptibil-
ity. Solute redistribution affects the solidification temperature range and
amount of terminal liquid (small amount of terminal liquid is most detri-
mental).
Minor variations in Nb, Si, and C content in Ni-base alloys have a
strong influence on the solidification temperature range, type and amount
of secondary phases, which form during terminal stages of solidification
and affect the solidification cracking susceptibility [10]. These alloy addi-
tions decrease the liquidus and solidus temperatures and increase the
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 189

solidification temperature range, Nb being most effective in this. Of the two

types of the eutectic-type constituents, Ȗ/NbC (or TiN(C)) and Ȗ/Laves,
known to form as the Ni-base alloy welds solidify, the Ȗ/Laves constituent
is more deleterious in terms of hot cracking as it forms at a lower tempera-
ture and extends thus the solidification temperature range. In dissimilar
metal welds the studied alloys are used to join low alloy steels to stainless
steels and due to dilution the weld metal can become significantly enriched
in Fe, Si and C, which may significantly change the solidification behavior
and associated cracking tendency. It is known from Fe-base alloys and
based on Fe additions to Nb-bearing Ni-base alloys that Fe promotes the
formation of Ȗ/Laves constituent. Silicon addition has similar effect. Nio-
bium addition promotes higher amounts of total eutectic and both eutectic-
type structures (Laves and NbC) are highly enriched in Nb. [10]
Solidification temperature ranges were measured by differential thermal
analysis (DTA) by Wu and Tsai [2] for Alloy 82 (44°C) and Alloy 52
(13°C). However, in the studied Ni-base alloys the Ȗ/Laves constituent
forms in very small amounts, which probably precludes the measurement
of the L => (Ȗ + Laves) reaction temperature by DTA. For Alloys 718 and
625 weld overlay a wide solidification temperature range (around 160°C
and 170°C, respectively) and presence of interdendritic Laves phase is
typical. The Ȗ within the Ȗ/Laves mixture in Alloy 625 weld overlay has
composition about 9,3 wt.-% Nb for the eutectic composition of 18,9 wt.-
% Nb and Laves phase composition of 22.1 wt.-% Nb and respectively for
Alloy 718 9,3, 19,1 and 22,4 wt.-% Nb were found [7, 8, 9, 10]. In this
study the composition of Ȗ close to the white (Laves) phase in the inter-
dendritic region was 5,5 wt.-% Nb and in the Laves phase even 30,8 wt.-%
Nb content (Alloy 152) was measured.
Hot cracking occurred along the dendrite boundaries and the susceptibil-
ity to hot cracking in the weld mock-up samples was observed to follow
the order based on TCL: Alloy 152 > Alloy 52 > Alloy 182 > Alloy 82. In
pure weld metal hot cracking tests the susceptibility was observed to fol-
low the order: Alloy 182 • Alloy 152 > Alloy 52 • Alloy 82. The differ-
ences between these two test types are thought to be related to dilution ef-
fects because Fe, Si, and C enhance the eutectic phases and expand the
solidification temperature range.
In order to avoid solidification cracking in Nb-bearing Ni-base alloys
the solidification temperature range has to decrease and the terminal sec-
ondary constituents have to form in small quantities. Carbon addition may
improve the resistance to hot cracking by decreasing the amount of Laves
phase (promotes formation of Ȗ/NbC), but especially Nb and Si seem to be
detrimental. Niobium segregation is governed by the prevailing solidifica-
tion conditions: the low weld metal cooling rate and high heat input increase
190 H. Hänninen et al.

Nb segregation and the amount of Laves phase [11]. Thus, for controlling
the Laves phase in the Nb-bearing Ni-base weld metals lowest possible
heat input with high cooling rates have to be used. Additionally, control of
dilution by Fe, Si, and C is very important, since these elements lower the
liquidus and solidus temperatures, increase the solidification temperature
range and favour eutectic-type solidifications.

Conclusions

The following conclusions can be drawn based on the results of this study:

− Dissimilar metal weld joints of Ni-base weld metals show in their mi-
crostructure marked segregation of Nb, Si, P and Mn to the last liquid
to solidify wetting the dendrite boundaries. In the Nb-bearing alloys
eutectic Laves and Nb(C, N) phases form along the dendrite bounda-
ries and in Ti-alloyed Alloy 52 TiN(C) phase forms, respectively.
− Hot cracking occurred along the dendrite boundaries and the suscepti-
bility to hot cracking in the weld mock-up samples was observed to
follow the order: Alloy 152 > Alloy 52 > Alloy 182 > Alloy 82.
− In pure weld metal hot cracking tests the hot cracking susceptibility
was observed to follow the order: Alloy 182 • Alloy 152 > Alloy 52 •
Alloy 82.
− The differences are probably related to the dilution effects because Fe,
Si, and C enhance the eutectic phases and expand the solidification
temperature range.

References

1. Hood B, Lin W (1995) Weldability Testing of Inconel Filler Materials. In:

Seventh International Symposium on Environmental Degradation of Materials
in Nuclear Power Systems – Water Reactors. Breckenridge, Colorado, USA,
7–10 August 1995. pp. 69–79.
2. Wu W, Tsai C (1999) Hot Cracking Susceptibility of Fillers 52 and 82 in
Alloy 690. Met. Mat. Trans., 30A, pp. 417–426.
3. Collins M, Lippold J (2003) An Investigation of Ductility Dip Cracking in
Nickel-Based Filler Materials – Parts I-III. Welding Research, pp. 288–295.
4. Nishimoto K, Saida K, Okauchi, H (2004) Microcracking Susceptibility in
Multi-Pass Weld Metal of Inconel 690 Alloy. IIW Doc. No. IX-2097-04. 15 p.
 Hot Cracking Susceptibility of Ni-Base Alloy Dissimilar Metal Welds 191

5. Nissley N, Collins M, Guaytima G, Lippold J (2002) Development of the

Strain to Fracture Test for Evaluating Ductility Dip Cracking in Austenitic
Stainless Steels and Ni-base Alloys. Welding in the World, 46, 7/8, pp. 32–40.
6. CEN ISO/TR 17641-3 (2005) Destructive Tests on Welds in Metallic Materi-
als. Hot Cracking Tests for Weldments. Arc Welding Processes. Part 3: Ex-
ternally Loaded Test. 17 p.
7. Knorovsky GA, Cieslak MJ, Headley TJ, Romig Jr AD, Hammetter WF
(1989) Inconel 718: A Solidification Diagram. Met. Trans., 20A, 10, pp.
2149–2158.
8. Cieslak MJ, Headley TJ, Kollie T, Romig Jr AD (1988) A Melting and So-
lidification Study of Alloy 625. Met. Trans., 19A, 9, pp. 2319–2331.
9. Dupont JN (1996) Solidification of an Alloy 625 Weld Overlay. Met. Trans.,
27A, 11, pp. 3612–3620.
10. Dupont JN, Robino CV, Marder AR (1998) Solidification and Weldability of
Nb-Bearing Superalloys. Welding J., 77, pp. 417s–431s.
11. Radhakrishna CH, Prasad Rao K (1997) The Formation and Control of Laves
Phase in Superalloy 718 Welds, J. Mater. Sci., 32, pp. 1977–1984.
Evaluation of Weld Solidification Cracking
in Ni-Base Superalloys Using the Cast
Pin Tear Test

B.T. Alexandrov1, J.C. Lippold1, N.E. Nissley2

1
The Ohio State University, Columbus, Ohio, USA
2
Exxon Mobil Upstream Research, Houston, Texas, USA

Abstract

A second-generation cast pin tear test (CPTT) that is capable of ranking

the weldability of Ni-base superalloys has been developed at the Ohio
State University. The CPTT utilizes an optimized testing procedure and
apparatus design that provide controllable and repeatable testing
conditions, and yield reproducible and reliable test results.
The CPTT provided a weldability ranking of four highly alloyed
stainless steels that is in very good correlation to the results of a round
robin study on six externally restrained weld hot-cracking tests. The
solidification microstructure of the test samples (cast pins) closely
simulates the microstructure of low to medium heat input welds, such as
those made by the GTAW process. The CPTT fracture surfaces exhibit
dendritic “eggcrate” fracture morphology, which is a characteristic of
solidification cracking. This fracture morphology is similar to the high and
medium temperature regime of solidification cracks generated using the
Varestraint test. The nature of crack nucleation and propagation, and the
crack healing phenomenon in CPTT closely resemble weld solidification
cracking.
The CPTT technique was applied for evaluating the solidification
cracking susceptibility of a series of high performance turbine engine
alloys. These alloys were ranked in order of decreasing susceptibility to
solidification cracking as follows: René alloys 142, 125, and 77, alloy 718,
René 80, Waspaloy, and alloy 600. These rankings are in good correlation
to field experience with the solidification cracking susceptibility of the
tested alloys.
The second generation CPTT proved to be capable of ranking the
solidification cracking susceptibility of both “difficult-to-weld” alloys and
“standard” alloys, and to differentiate their solidification cracking
194 B.T. Alexandrov et al.

behavior. It also provides an efficient and inexpensive tool for weldability

testing in the process of alloy and consumable development.

Introduction

The cast pin tear test was originally developed by Hull [1] and used in the
1970’s as a method for evaluating the susceptibility of alloys to hot
cracking during welding and casting. It consisted of levitation melting and
casting of small charges of material in copper molds to produce conical
cast pins with varying geometry. The charge mass in this test was constant
(19 g). The test sensitivity was controlled by the pin length and geometry,
and by the mold diameter. The pins were examined for circumferential
cracks and the total crack length was plotted versus the mold number.
Based on the percent of circumferential cracking and mold number, the
CPTT generated arbitrary ranking of the susceptibility to hot cracking.
Compared to the other hot-cracking tests, it required only several hundred
grams of material to develop a cracking susceptibility curve and was
suitable for alloy development purposes.
The CPTT of Hull was modified by the Welding and Joining
Metallurgy Group at The Ohio State University [2]. The costly and
complex levitation melting technique was replaced by an arc melting
technique. A standard gas tungsten arc welding (GTAW) torch and power
supply are used to melt a small charge of material on a water-cooled
copper hearth under an inert (argon) atmosphere. The hearth was equipped
with a simple delivery system that transferred the molten charge into a
copper mold positioned below the hearth. Although a successful system
was initially developed using this approach, problems with complete mold
filling and reproducibility of the cracking results were encountered.
Because of the problems with mold filling, a second-generation cast
pin tear test that is more robust and capable of ranking the weldability of
Ni-base superalloys has been recently developed at the Ohio State
University [3]. The reproducibility and reliability of test results was
improved by further development of the CPTT apparatus and by
optimizing the melting and mold filling procedures. The test sensitivity
was improved by introducing smaller increments in cast pin length. The
capacity of testing various alloys was expanded by using mold materials
with different thermal conductivity and by extending the range of mold
lengths.
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 195

This chapter presents the development and verification of the second-

generation CPTT, and its application for ranking the susceptibility to
solidification cracking of a series on Ni-base superalloys for turbine engine
application.

Second-Generation Cast Pin Tear Test

The general design of the second-generation CPTT apparatus is shown on

Fig. 1. A charge of the tested alloy is melted in a gas-shielded chamber
over water cooled copper hearth by a GTAW torch. The hearth has a
central opening that connects the gas-shielded chamber to a copper mold.
The molten charge is transferred through the hearth opening into the mold
where it solidifies as a cast pin. The parameters of the CPTT procedure are
provided in Table 1.

Fig. 1. Cast Pin Tear Test apparatus

The CPTT apparatus is equipped with vacuum and overpressure

systems that are controlled by electromagnetic valves and pressure and
vacuum gauges. These are used to both provide an inert atmosphere in the
melting chamber and mold before testing, and facilitate the transfer of
molten charge to the mold.
196 B.T. Alexandrov et al.

Table 1. Cast Pin Tear Test procedure

Parameter Value
Mold Material Cu; Cu-Be
Shielding Gas Ar
Gas Flow Rate, ml/s 70–80 (9–10 cf h–1 )
Arc Current, A 250
Arc Length, mm 15 (0.6 in)
Arc Time, s 5–7
Over-Pressure, MPa 0.007–0.014 (1–2 psi)
Cast Pin Diameter, mm 9.525 (0.375 in)
Cast Pin Length, mm 12.7–50.8 (0.5–2 in)
Length Increments, mm 3.175 (0.125 in.)

The hearth design is presented in Fig. 2. The hearth opening limits the
heat extraction from the charge during its melting, thus allowing for
sufficient superheating of the latter and facilitating its quick transfer into
the mold. The mold cavity is designed to produce a cylindrical cast pin
with restraining head and foot, Fig. 3. The CPTT test utilizes molds and
cast pins with constant diameters and varying lengths. The charge mass
corresponds to the cast pin length, Table 2.

Fig. 2. Open hearth

 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 197

Head

Foot

Fig. 3. Cast pin mold

Table 2. Range of mold (cast pin) lengths and correspondent pin/charge volume
and mass (for pure Ni samples)

Mold Length, Volume, Mass,

mm (in.) mm3 g
12.700 (0.500) 1122.6 10.0
15.875 (0.625) 1178.7 10.5
19.050 (0.750) 1234.8 11.0
22.225 (0.875) 1291.0 11.5
25.400 (1.000) 1347.1 12.0
28.575 (1.125) 1403.2 12.5
31.750 (1.250) 1459.4 13.0
34.925 (1.375) 1515.5 13.5
38.100 (1.500) 1571.6 14.0
44.450 (1.750) 1683.9 15.0
50.800 (2.000) 2806.5 16.0
198 B.T. Alexandrov et al.

The rate of heat extraction from the molten metal is controlled by the
thermal conductivity of mold material. In this study, two mold materials
were used: high purity copper (C10100) and a Cu-Be-Co alloy (C17000)
with thermal conductivities of 391 W/m-K and 118 W/m-K respectively.
The lower rate of heat extraction of the Cu-Be-Co molds made it possible
to test some Ni-base alloys that had been previously problematic with
respect to filling of copper C10100 molds.
Longitudinal tensile stress is generated in the cast pin as it solidifies
and cools down to room temperature. The stress level increases with the
pin length. At the critical stress level (pin length) for particular alloy, hot
cracking occurs at the pin surface usually just below the pin head, as
indicated on Fig. 4. With further increase in the cast pin length, the total
surface crack length increases to 100% circumferential cracking and
eventually results in complete pin separation, Fig. 5.

Fig. 4. Cast pin samples of alloy 625. Fig. 5. Complete separation in 1.25 in.
Arrows show the typical location of long pin of alloy 800H
solidification cracks

The cast pins are examined for cracks using a binocular microscope at
magnifications from 10x to 70x. During this examination the cast pins are
rotated around their longitudinal axes by a specialized device. The
projected crack length in a plane that is perpendicular to the pin axis is
measured in degrees. The percentage of cracking is calculated as follows:
LT
% Cracking = ×100 , %,
360
where LT is the total length in degrees of all cracks measured on the pin
surface.
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 199

The fracture surface of each completely cracked sample is examined to

ensure that no casting defects contributed to cracking. The partly cracked
cast pins are longitudinally sectioned and also checked for internal casting
defects that might have affected the test results. Based on the valid test
results, a response curve of maximum circumferential cracking is plotted
as a function of pin length, Fig. 6. The maximum pin length of no cracking
and the minimum pin length of 100% circumferential cracking are used as
quantitative criteria for ranking the susceptibility to solidification cracking.
The sensitivity of the CPTT depends on the increments in cast pin
length. Pin lengths between 12.7 mm and 25.4 mm with increments of
3.175 mm proved to be sufficient to differentiate and rank the
susceptibility to solidification cracking in highly susceptible Ni-base
alloys. The reproducibility of the CPTT results and the reliability in
determination of the maximum pin length of no cracking and the minimum
pin length of 100% cracking have been demonstrated by extensive testing
of alloy René 77 (Fig. 6).
100%
1 2 2
90%
Minimum pin length
80% of 100% cracking
Percent Circumferential Cracking

Maximum circumferential
70% cracking response curve

60%

50%
0% to 100%
40% cracking range

30% Maximum pin length

of no cracing
20%

10%
1
2 2 3 3 1
0%
0.5 0.625 0.75 0.875 1 1.125 1.25 1.375 1.5 1.625
Pin Length (in)

Fig. 6. Quantitative criteria for evaluation of susceptibility to solidification

cracking by the Cast Pin Tear Test (CPTT). The circumferential cracking
response curve for René 77 is shown here. The numbers on the plot indicate the
number of samples tested for each mold length
200 B.T. Alexandrov et al.

Comparison of CPTT to Other Hot Cracking Tests

In order to compare the CPTT to other “hot” cracking test techniques, a

series of high-alloy austenitic stainless steels were tested by the CPTT,
including alloys 800H, 926, AC66, and 825. The chemical composition of
these alloys is given in Table 3.

Table 3. Composition of the IIW round robin study alloys, wt.%

Alloy 800H 926 AC66 825

Ni 30.55 24.80 31.45 39.15

Cr 20.40 20.85 27.35 22.25

Fe Bal Bal Bal 31.30
C 0.068 0.010 0.072 0.006
Si 0.38 0.32 0.21 0.32
Mn 0.70 0.82 0.50 0.69
P 0.011 0.017 – 0.015
S 0.002 0.003 – 0.003
Mo – 6.38 – 3.16
Ti 0.33 – – 0.77
Al 0.280 – 0.014 0.090
Other – 0.91 Cu 0.83 Nb 1.90 Cu
0.196 N 0.085 Ce

These alloys had previously been the subject of an IIW round robin
study of externally restrained weld hot cracking tests [4, 5]. The study was
conducted by ten participating organizations and included the Varestraint,
Trans-Varestraint, PVR, MVT, MIS-1, and LTP-1-6 tests [4]. Each
participating laboratory applied its own testing procedures.
That study produced a total of twelve separate rankings of the hot
cracking susceptibility of the tested alloys. Most of the rankings, generated
by the different hot cracking tests and by different laboratories using same
tests, were conflicting. This poor correlation was related to the different
aspects of hot cracking evaluated by the different tests, and to the different
testing procedures used with the same hot cracking tests [4, 5]. A summary
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 201

ranking of the tested alloys that is based on the results from that study is
presented in Table 4.
The CPTT results for these alloys, in terms of maximum pin length for
no cracking and response curve for circumferential cracking, are
summarized in Fig. 7. The CPTT ranked alloy 800H as the most
susceptible, followed by alloys 926, AC66 and 825. These results are in
fairly good correlation with the general results of the IIW round robin test.
A very good correlation was found between the CPTT and PVR test
rankings. This is not surprising since both the CPTT and the PVR test
restrain the test samples in the direction of solidification.

Table 4. Hot cracking susceptibility ranking of the IIW round robin study alloys

Test Type and Laboratory 800H 926 AC66 825

CPTT 1* 2 3 4
PVR Laboratory 1 1 2 4 3
Laboratory 2 1 2 3 4
Laboratory 3 1 2 4 3
Laboratory 4 1 3 2 4
IIW ranking (out of 12 labs) 1 (8) 2 or 3 (7) 2 or 3 (9) 3 or 4 (7)
* 1 = highest susceptibility

100%
% Circumferential cracking

90%

80%

70%
Alloy 800H Alloy 926 Alloy 825
60%

50%
Alloy AC66
40% .

30%
AC66 Max
20% 800H Max
825 Max
10% 926 Max

0%
0.5 0.625 0.75 0.875 1 1.125 1.25 1.375 1.5
Pin length, in.

Fig. 7. Susceptibility to solidification cracking of the IIW round robin study

alloys evaluated by CPTT
202 B.T. Alexandrov et al.

The fracture morphology of the solidification cracks produced by the

CPTT and the transverse Varestraint test was examined using scanning
electron microscopy (SEM), as shown in Fig. 8. The fracture surface in
transverse Varestraint samples has a dendritic appearance that varies from
flat at the low temperature region of the crack to the dendritic “eggcrate”
morphology at medium and high temperatures (Fig. 8a). The cracks
generated by CPTT have a dendritic fracture morphology that is similar to
the high and medium temperature region in the Varestraint cracks.
The cracking in the CPTT samples initiates at the pin surface and
propagates along the solidification grain boundaries and solidification
subgrain boundaries, as shown in Fig. 9.

a) b)
Fig. 8. Alloy 926 fracture surface tested with Varestraint (a) and CPTT (b)

Fig. 9. Solidification cracking at the cast pin surface of alloy A66

 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 203

Testing of High Performance Ni-base Superalloys

A solidification cracking susceptibility of seven high performance turbine

engine alloys was also determined by the CPTT. These alloys included
René alloys 77, 80, 125, and 142, and alloys 600, 718, and Waspaloy. The
chemical compositions for these materials are provided in Table 5.

Table 5. Nominal chemical composition of the tested Ni-base superalloys in wt.%

Alloy Inconel Inconel Wasp- René René René René

600 718 aloy 77 80 125 142

Ni 76 53 58 57 60 59 57

Co – – 13.5 15.0 9.5 10.0 12.0

Fe 8.0 18.5 – 0.5 – – –

Cr 15.5 19.0 19.5 14.6 14.0 8.9 6.8

Al – 0.50 1.40 4.30 3.00 4.80 6.15

Ti – 0.90 3.00 3.35 5.00 2.50 –

Ta – – – – – 3.80 6.35

Nb – 5.10 – – – 0.10 –

W – – – – 4.00 7.00 4.90

Mo – 3.00 4.30 4.20 4.00 2.00 1.50

C 0.08 0.08 0.07 0.07 0.17 0.11 0.12

Hf – – – – – 1.55 1.50

B – – 0.006 0.015 0.015 0.015 0.015

Zr – – 0.09 0.04 0.03 0.05 0.02

Cu 0.15 0.15 – – – – –

Other – – – 0.20 Si – – 2.80 Re

204 B.T. Alexandrov et al.

Solidification Cracking Susceptibility

The solidification cracking susceptibility of the tested alloys was evaluated

by the CPTT procedure given in Table 1 using Cu-Be-Co molds. At least
three samples were tested at each pin length in the transition range from no
cracking to 100% cracking. The results of this testing is summarized in
Table 6 and plotted as function of percent circumferential cracking versus
pin length in Fig. 10.

Table 6. Solidification cracking parameters in Ni-base superalloys tested by

CPTT

Alloy Max. pin length Min. pin length Zero to 100%

for no cracking, 100% cracking, cracking range,
mm (in) mm (in) mm (in)
Inconel 600 38.100 1.500 50.800 2.000 12.700 0.500
Waspaloy 38.100 1.500 50.800 2.000 12.700 0.500
René 80 34.925 1.375 38.100 1.500 3.175 0.125
Inconel 718 25.400 1.000 38.100 1.500 12.700 0.500
René 77 25.400 1.000 28.575 1.125 3.175 0.125
René 125 15.875 0.625 19.050 0.750 3.175 0.125
René 142 12.700 0.500 15.875 0.625 3.175 0.125
% Circumferential cracking

100
90
80
70 R'80
R'125
60
R'142 R'125 R'77 718 R'80 Waspalloy 600 R'142
50 R'77

40 600
718
30
Waspalloy
20
10
0
0.50 0.75 1.00 1.25 1.50 1.75 2.00
10 15 20 25 30 35 40 Pin
45length, in50 mm
Pin Length, in (mm)

Fig. 10. Relative susceptibility to solidification cracking of high performance

Ni-base superalloys tested by the CPTT
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 205

The alloys were ranked in order of decreasing susceptibility to

solidification cracking as follows: R’142, R’125, R’77, alloy 718, R’80,
Waspaloy, and alloy 600. Alloys 718 and René 77 had equal maximum pin
lengths of no cracking, but the minimal pin length of 100% in alloy 718
was three increments (0.375 in.) higher, thus showing lower susceptibility.
Alloys 600 and Waspaloy had coinciding cracking ranges with somewhat
lower extent of cracking in the former at the 1.750 in. pin length (Fig. 10).
The zero to 100% cracking ranges of all the René alloys were one
increment of pin length (0.125 in.) wide. The cracking ranges of alloys
718, Waspaloy and 600 were four increments wide (0.5 in.). These
rankings are in generally good correlation to the field experience with the
solidification cracking in the tested alloys.

Solidification Microstructure

Longitudinal sections of tested cast pins were mounted, polished, and

etched electrolytically in a chromic acid solution at 2.5 V for 30 s. The cast
microstructure was investigated by optical microscope at magnifications
up to 400X. Although some variations were observed among the tested
alloys, the general solidification macrostructure of the cast pins consisted
of a thin layer of equiaxed grains at the mold surface, followed by an area
of extensive growth of columnar and cellular dendrites (Fig. 11a), and a
central area of equiaxed dendritic growth (Fig. 11b). Alloys R’142, R’125
and R’80 exhibited all of these areas, with comparatively finer equiaxed
dendrites in alloy R’142, Fig. 11c. In alloys R’77 and 718 narrow
equiaxed areas with large equiaxed dendrites were present along the cast
pin axis (Fig. 11d) mixed with areas of center contact of columnar
dendrites, Fig. 11e. The macrostructure of alloy 600 and Waspaloy
consisted of columnar dendrites with small isolated areas of equiaxed
dendrites in the cast pin central areas, Fig. 11f.
A series of measurements were conducted to evaluate the primary
dendritic arm spacing (DAS) of the investigated alloys under the
solidification conditions of the CPTT. The measurements were performed
on grains of columnar dendrites that were located close to the fracture
surface and whose primary arm axes were normal to the surface of
measurement. The results have shown that the DAS of the investigated
alloys under the specific conditions of CPTT is in the range from 13 to 30
μm, as summarized in Table 7. This range of DAS corresponds to what is
observed during solidification of low to medium heat input welds using the
gas-tungsten arc welding (GTAW) process.
206 B.T. Alexandrov et al.

a) Crack initiation area. Fine grained b) Transition of columnar to equiaxed

surface region followed by dendritic dendritic growth in cast pin of alloy
growth in alloy R’80 R’125

c) Central area with finer equiaxed d) Isolated central area with large
dendrites in cast pin of alloy R’142 equiaxed dendrites in cast pin of alloy
R’77

e) Central contact of opposite growing f) Isolated equiaxed dendrites in the

dendrites at cast pin center line, alloy central of area cast pin of Waspaloy
718
Fig. 11. Solidification structure in cast pins of the tested Ni-base superalloys
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 207

Table 7. Primary dendritic arm spacing in the tested Ni-base alloys in the
conditions of CPTT (average of 30 measurements)

Alloy Average DAS, [μm] Standard Deviation

René 142 18.2 3.1
René 125 26.0 5.0
René 80 16.0 2.6
René 77 16.9 2.2
Inconel 718 13.3 2.0
Inconel 600 30.1 5.6
Waspalloy 14.7 2.2

Shrinkage Porosity

Some shrinkage porosity was found along the cast pin center line in the
alloys forming large equiaxed dendrites in the last region of the pin to
solidify. Alloys René 125 and René 77 (Fig. 11d) had more significant
center-line shrinkage porosity, followed by alloy 718 (Fig. 11e) and René
80 with some moderate porosity. The shrinkage porosity was typically
found in the longer pins and decreased to an insignificant level with
decreasing the pin length. In general, the shrinkage prosity was not related
to solidification cracking but instead increased the percent of invalid
samples at the higher mold lengths.
Alloys 600, Waspaloy, and René 142 exhibited very little dispersed
shrinkage porosity that was not related to solidification cracking. The
dispersed shrinkage porosity typically forms in the cast pin central area, it
is evenly distributed, and the size of the separate pores is smaller then the
dendrite arm spacing. Therefore, this dispersed porosity is not expected to
affect the general stress distribution in the cast pin, and the processes of
crack nucleation and propagation, which occur in the pin surface area. This
kind of porosity was considered acceptable and not affecting the test
results.
From the total of 27 sectioned cast pin samples, only three had large
shrinkage pores that caused cracking. The test results from these samples
were not included in the results. Most of the tested samples had either
insignificant or no shrinkage porosity and, in general, porosity did not
affect the test results.
208 B.T. Alexandrov, J.C. Lippold, N.E. Nissley

Fracture Morphology

Optical microscopy of longitudinal cast pin sections revealed that

solidification cracking occurred along solidification grain boundaries
(SGBs) and solidification subgrain boundaries (SSGBs), as shown in
Fig. 12. Figures 12a,b represent a clear example of the process of crack
nucleation and propagation under the conditions of CPTT. The solidify-
cation cracks appear to form initially in the fine grained surface area that
solidifies at the surface of the pin and then propagate along the SGBs (Fig.
12b). Propagation along the SSGB’s is also possible during the final stage
of fracture (Fig. 12c,e,f). Cracking often occurred in multiple planes within
the samples (Figs. 12c,e,f) and sometimes joined to form the final fracture.
Clear evidence of backfilling (crack “healing”) caused by an interdendritic
liquid at the end of solidification was found in alloys R’142, R’125 and
R’80 (Fig. 12a,b,c). No such evidence was found in any of the other alloys
(Fig. 12d,e,f).
For each tested material, the fracture surface of the shortest pin that
cracked 100% was evaluated by SEM. All fracture surfaces exhibited a
dendritic “eggcrate” morphology that is characteristic of solidification
cracking, Fig. 13. This fracture morphology is similar to the high and
medium temperature region observed in Varestraint samples, Fig. 8.
Alloys René 142 and René 125 had a mixed mode failure with both the
dendritic morphology and small regions of flat fracture (Fig. 13b). The
latter occurred due to bridging between solidification fracture surfaces due
to tensile overload. This flat fracture bridging was not observed in the
remainder of the examined samples, which exhibited dendritic fracture
morphology, Fig. 13d to 13h.
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 209

a) Crack initiation, propagation, and b) Crack initiation, propagation along

backfilling in René 142 the SGBs, and backfilling in René 142

c) Multiple cracking, propagation d) Cracking along the SSGBs in René

along the SSGBs, and backfilling in 77
René 125

e) Multiple cracking and propagation f) Multiple cracking and propagation

along the SSGBs in alloy 718 along the SSGBs in alloy 600
Fig. 12. Solidification cracks in cast pins of the tested Ni-base superalloys
210 B.T. Alexandrov, J.C. Lippold, N.E. Nissley

20 um 40 um
a) René 142 at 0.625 in. b) René 142 at 0.625 in.

20 um 20 um
c) René 125 at 0.75 in. d) René 77 at 1.25 in.

20 um 20 um
e) Inconel 718 at 1.75 in. f) René 80 at 1.5 in.

20 um 20 um
g) Inconel 600 at 2 in. h) Waspalloy at 2 in.

Fig. 13. Solidification cracking fracture morphology in cast pins of the tested
Ni-base superalloys
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 211

Discussion

The second-generation cast pin tear test has been developed in an effort to
create a robust weldability test that is able to rank the solidification
cracking susceptibility of both “difficult-to-weld” alloys and “standard”
alloys. The cast pin apparatus and testing procedure were designed to
provide controllable and reproducible testing conditions, and to produce
cast pins of good integrity. The cases of unacceptable shrinkage porosity
in difficult-to-cast alloys were reduced to less than 10% of the tested
samples. As a result, the reproducibility and reliability of the CPTT results
were significantly improved.
The solidification conditions in the CPTT closely simulate the
solidification conditions of low to medium heat input welds, such as those
made by the GTAW process. These include the morphology of the
solidification microstructure and the sequence of solidification, starting
with a fine-grained “chill zone” at the mold wall, followed by cellular and
columnar dendritic growth, and finishing with equiaxed dendritic growth
at the center of the pin. The primary dendritic arm spacing was between 13
and 30 μm. Even higher weld heat input conditions can be simulated by
using lower cooling capacity (lower conductivity) molds.
Metallographic and fractographic examination of cast pins have
confirmed that the CPTT also simulates the weld solidification cracking in
actual welds relative to the nature of crack nucleation and propagation,
crack morphology, and crack healing. For example, cracking was primarily
along solidification grain boundaries, just as it is in actual welds in Ni-base
alloys and stainless steels.
The reliability of the CPTT results has been confirmed by comparison
to six externally restrained hot cracking tests. Very good correlation was
found between the weldability rankings of four high alloy stainless steels
generated by CPTT and by PVR test.
The solidification cracking susceptibility ranking of the seven tested
Ni-base superalloys is in good correlation with the field weldability
experience with these alloys. This indicates that the improved, second
generation CPTT technique is a reliable tool for evaluating the weldability
of materials with wide variations in solidification cracking susceptibility,
particularly those with high to moderate susceptibility.
The solidification cracking behavior of the Ni-base superalloys is a
function of their complicated alloying systems and of the solidification
conditions provided by the CPTT. The combination of chemical
composition and cooling rate determine the solidification morphology,
212 B.T. Alexandrov, J.C. Lippold, N.E. Nissley

tendency for formation of eutectic phases, the properties of the latter and
their capability to filling interdendritic shrinkage, and the general liquid-
solid and solid-state shrinkage behavior.
A series of experimental and modeling investigations will be
conducted in the future, aimed at quantifying the parameters that control
the solidification cracking phenomenon under the conditions of the CPTT.
The mold and cast pin design facilitates the application of relatively simple
finite element modeling to study the processes of heat transfer,
solidification, and development of stresses during solidification. The
technique of single sensor differential thermal analysis – SS-DTA [6] will
be utilized in conjunction with the CPTT to determine the solidification
temperature ranges and the possible liquid-state precipitation and eutectic
reactions. Such combined experimental- modeling approaches will allow
the solidification morphology and solidification cracking behavior to be
related to quantifiable parameters as cooling rate, solidification
temperature range, liquid- and solid-state volume shrinkage, and stresses
associated with solidification.

Conclusions

1. A second-generation cast pin tear test (CPTT) was developed that is

capable of testing and ranking the susceptibility of Ni-base superalloys
to solidification cracking.
2. The improved design of the CPTT apparatus and the new CPTT
procedure provide repeatable and controllable testing conditions,
produce cast pins with acceptable integrity, yield reproducible and
reliable test results, and are highly sensitive for ranking the weldability
of difficult-to-weld alloys.
3. The CPTT has provided very good correlation to a variety of
externally restrained solidification cracking tests in ranking the
weldability of four high-alloy, fully austenitic stainless steels. The best
correlation was obtained with the PVR test, which similar to the CPTT,
restrains the samples in the direction of solidification.
4. A ranking of solidification cracking susceptibility has been generated
that includes seven Ni-base high performance alloys. The weldability
of these alloys as ranked by the CPTT is in good correlation with their
field weldability.
 Weld Solidification Cracking in Ni-Base Superalloys Using CPTT 213

5. The CPTT closely resembles the solidification conditions in actual

fusion welds. Based on measurements of the primary dendrite arm
spacing in the cast pins, the solidification conditions approximate those
in low to medium heat input welds. These conditions can potentially be
extended to simulate higher heat input welding by altering the mold
cooling capacity.
6. The CPTT has a tremendous application potential in testing the
weldability of a wide range of non-ferrous alloys and stainless steels.
In combination with numerical modeling and a technique for single
sensor differential thermal analysis, it will provide an efficient and
inexpensive tool for alloy and consumable development.

References
1. Hull F. C., Cast-Pin Tear Test for Susceptibility to Hot Cracking, Welding
Journal, Vol. 38 (4), 1959, pp. 176s–181s.
2. Ryan D. P., Development of Modified Cast-Pin Tear Test to Evaluate the
Solidification Cracking Susceptibility of Fully Austenitic Materials, MS
Thesis, The Ohio State University, 2003.
3. Alexandrov B., Nissley N., Norton S., and Lippold J., Development of a
Weldability Test for High Performance Base and Filler Materials, Report No.
MR0606, Edison Welding Institute, July 2006.
4. Wilken K., Investigation to compare Hot Cracking Tests – Externally Loaded
Specimen, IIW Doc. IX-1945-99 (II-C-168-99).
5. Finton T., Lippold J., and Bowers R., Comparison of Weld Hot Cracking
Tests, Summary of an IIW Round Robin Study, IIW Doc. IX-H-459-99 (II-C-
175-99).
6. Alexandrov B. T. and Lippold J. C., Relationship Between the Solidification
Temperature Range and Weld Solidification Cracking Susceptibility of
Stainless Steels and Ni-base Alloys, IIW Doc. IX-2163-05.
SAW Cold Wire Technology – Economic
Alternative for Joining Hot Crack Sensitive
Nickel-Base Alloys

U. Reisgen, U. Dilthey, I. Aretov

RWTH Aachen University, Welding and Joining Institute, Aachen, Germany

Abstract

The construction of industrial plants and gas turbines makes increasingly

higher demands to the used materials. Consequently, austenitic CrNi-steels
and nickel-based alloys are increasingly applied in power station
construction, crude oil and petrochemistry and, moreover, in industrial
furnace and turbine construction. Besides the good mechanical properties
of these materials also a good process ability is required. This applies
particularly to welding. Due to their low heat input, TIG and MIG (pulsed)
welding are the most frequently used welding methods in the case of heat-
resistant nickel-base super alloys and also in the case of stainless CrNi-
steels. Due to the high heat input, submerged-arc welding is, as a rule, not
applied for the welding of these steels. Submerged arc welding could,
however, be used as efficient alternative if process modifications are
available, which ensure higher hot cracking resistance by reduced heat
input. Apart from the high weld quality which is a result of submerged-arc
welding, the economic efficiency which is particularly marked by the high
deposition rates is also interesting.
Submerged-arc welding knows different possibilities of reducing the en-
ergy to a lowest possible quantity. The easiest way to obtain this objective
is a higher welding speed which reduces the energy-per-unit-length. A
more sophisticated - but also more effective - approach is the direct with-
drawal of heat from the weld pool by addition of filler material, for exam-
ple, in the form of cold wire. The heat in the weld pool is not completely
entering the base metal but is partly used for melting the cold wire.
The tests which were performed within the scope of this project were
made on the nickel-based alloys 617 and 625. It has been the aim of the
project to make a statement about the welding suitability and the economic
efficiency of the SA cold wire technology; to this effect, the welding pa-
rameters and also the optimal electrode and cold wire parameters had to be
216 U. Reisgen et al.

determined. The specific reduction of the energy-per-unit-length is used to

achieve a maximum deposition rate. Comparative tests were carried out us-
ing the conventional SA method and the SA thin-wire method.
One main point had been the examination of the cold wire influence on
the weld geometry and on the chemical composition of the weld metal and
thus also on the hot-cracking tendency. The hot-crack tests were carried
out by means of the MVT test. Moreover, the influence of the modified SA
cold wire technology on the distortion had been examined. Through the
optimisation of the cold wire addition a high process reliability had been
obtained
The transfer of the achieved results to industrial practice has been
successful.

Introduction

The main problems which occur in fusion welding of fully austenitic

materials are caused by hot cracking. The reasons for the development of
solidification cracks and melt cracks lie in the most complex interaction of
several factors which are summarised into three groups. These factors are
of metallurgical, mechanical and thermal nature. They are reflecting the
properties of the welded joint. For a complete evaluation of the hot
cracking susceptibility of a weld, all three factors with their respective
degree of efficiency must be considered.
If a welding method is subject to modification, changes of several
influential factors are often occurring at the same time. It has been the aim
of this research work to examine the extent of the influence and of the
factors’ interactions.

Testing Equipment
Conventional SA equipment has been used for the weld examinations. The
equipment had been modified with an additional cold wire feeding unit,
which works without applying any additional electrical energy to the cold
wire. In order to adjust the feed speeds for electrode and cold wire
independently of each other, two separate feeding units have been used.
This allowed varying the feeding quantity of the cold wire without
changing the cold wire diameter which resulted in a higher flexibility of
the process.
 SAW Cold Wire Technology 217

Fig. 1. Schematic representation of the modified SA cold wire method

The schematic view of the applied process modification is shown in

Fig. 1. The cold wire is fed into the molten pool behind the electrode,
passing the liquid slag and is then molten in the pool. The decision for a
trailing arrangement of the cold wire feeding was based on the results from
the preliminary tests. These tests showed that this arrangement allowed
obtaining a wider seam with a better transfer to the base metal. The
reinforcement of the weld is thus reduced and the ratio “weld width/weld
height” is improved.
Another advantage of this arrangement is the possibility of melting the
cold wire in the molten pool at significantly lower temperatures than
would be the case if the drop transfer occurred during the melting of the
cold wire in the arc region. If a leading arrangement is used, melting
occurs only in the arc region which, again, results in the increased alloy
burn-off. Using the trailing arrangement it is, thereby, possible to pass
alloying elements into the molten pool which would otherwise have been
burned off.
In order to ensure the melting of the cold wire in the molten pool and
not in the arc region, the exact positioning of the cold wire is necessary to
obtain a reproducible welding process. For the exact positioning of the
cold wire, the following parameters must be considered:
• angle of slope of the cold wire
• distance between electrode and cold wire
• wire stick-out.
The parameters are depicted in Fig. 2.
218 U. Reisgen et al.

Stickout

ĮKD < 45°

1,5 mm

Workpiece
surface

Fig. 2. Positioning of the cold wire feeding

Influence of Cold Wire Addition on the Welded Joint

The following tests show the influence which cold wire addition exerts on
the welded joint. In particular, the weld geometry, the structure, the
chemical composition and the connection between cold wire addition and
hot-cracking susceptibility have been considered.

Weld Geometry

The first tests dealt with the determination of the influence which cold
wire exerts on the weld geometry. For that, welds with different quantities
of cold wire were carried out. The parameter “melting ratio” has been
introduced for these tests. It specifies the ratio between the mass of the
cold wire which has been added during welding and the mass of the
electrode. With the increasing quantity of cold wire, weld reinforcement
and weld width were also increasing, while the penetration depth was
reduced (Fig. 3). The smaller penetration area resulted in a lower degree of
dilution with the base metal (Fig. 4).
Caused by cold wire addition, the weld shows a tendency to reinforce-
ment of the weld beads. In order to compensate the reinforcement, the
 SAW Cold Wire Technology 219

14
12
10
Weld height
8 Weld width
[mm]

6
4
2
0 Penetration depth
0% 20% 40% 60% 80% 100%
deposition ratio [%]

Fig. 3. Influence of cold wire addition on welds reinforcement, weld width and
penetration depth

influence of the welding voltage on the weld geometry has been

investigated more closely. It was determined that a voltage increase
allowed the partial compensation of the developing weld reinforcement
(Fig. 5). It must, however, be considered that the increase of the welding
voltage has its process-technical limitations. The limited electrical and
thermal load capacities of the slag and the dynamics of the welding
process bring about process instabilities if the voltage is too high.

60%

50%
dilution [%]

40%

30%

20%

10%

0%
0% 20% 40% 60% 80% 100%
deposition rate [%]

Fig. 4. Influence of cold wire addition on the dilution with the base metal
220 U. Reisgen et al.

The cooling speed during crystallisation has great influence on the for-
mation of hot cracks. Cold wire addition results in faster cooling of the
molten metal; the crystallisation processes in the molten pool are greatly
influenced.
These tests were performed in order to determine a potential influence
on the temperature gradient in the molten pool in a solid state, provided
that the deposition ratio had been varied.
For that, thermocouples have been dipped into the molten pool, right
behind the cold wire through the slag which was by the time still liquid.
The thermocouples allowed the reproducible measurement of the
temperature gradient only within a temperature range of below 1000°C. It
has been observed that the solidification interval for alloy 617 is between
1380°C and 1330°C. Because of the homogenisation of the temperature
fields in the component it is not possible to establish an influence of the
cold wire addition on the temperature profile below 1000°C. For further
examinations it would be advisable to select a measuring method which
allows the measurement of temperature ranges above the melting
temperature of the filler materials.
The macro-sections (Fig. 5) do show differences with regard to the weld
geometry. It is observed that – with the increase of the cold wire addition -
the penetration area is getting smaller and the reinforcement is increasing
(Fig. 6). This causes the shortening of the fusion line, the heat dissipation

U=32,1V I=348A U=33,7V I=355A U=35,4V I=354A

U=37,3V I=361A U=39,0V I=358A U=40,6V I=363A

Fig. 5. Micro-sections of surface welds with increasing welding voltage and a cold
wire addition of 50%. Used wire electrode: UTP UP 6170 Co Mod., weld flux:
UTP UP FX 6170 Co, base material: austenitic CrNi-steel, welding speed:
50 cm/min
 SAW Cold Wire Technology 221

into the base metal is reduced. Particularly in SA welding, the heat dissipa-
tion into the base material is of special importance and the alteration of the
fusion line may have large effects on the thermal balance of the molten
pool and thus also on the solidification conditions.
Figure 6 shows that, up to the increase of the cold wire addition to 50%,
also an increase of the weld cross-section takes place. This means that an
increased quantity of metal is molten without any variation of the heat
input. The heat in the molten pool is more effectively used through the
cold wire addition than this is the case with standard SA welding methods.
A conclusion is that the average temperature of the molten pool which has
been produced with the SA cold wire method is lower than that of the
molten pool produced with SA standard methods. This also brings about a
changed curve of the molten pool isotherms and thus a lower temperature
gradient during a SA cold wire process which again affects the
crystallisation processes and the welded structure.
If the deposition rate is increased further, the excessive heat of the
molten pool is used for melting the increasing quantities of cold wire ad-
dition. The heat of the molten pool is no longer sufficient for the melting
of equal quantities of base material. The penetration area is reduced and
the weld reinforcement area is, at the same time, increased.

75

60 Reinforcement area
Weld area
area {mm²]

45
Penetration area
30

15

0
0 25 50 75 100
deposition ratio [%]

Fig. 6. Influence of the deposition ratio on the weld area, the penetration area and
the reinforcement area
222 U. Reisgen et al.

The fusion line is shortened. The heat transport from the liquid molten
pool into the solid base material is carried out via the fusion line. That im-
plies that: the longer the fusion line, the higher the excessive heat of the
molten pool (deposition ratio – 0%), and, the shorter the fusion line, the
colder the molten pool (deposition ratio – 100%). The increase of the cold
wire quantities is accompanied by the displacement of the weld cross-
sectional area in the direction of the workpiece surface.
If the deposition ratio is too high (above 100%), strong fluctuations of
the weld geometry occur. This is characterised by changes of the weld
width and the weld height and also by non-parallel weld flanks. The reason
for this is that the cooling of the molten pool has been too strong. These
fluctuations show that the welding process is no longer running stable and
that it will be collapsing if the quantity of the cold wire addition is further
increased.

Structure

The examinations of the structure have been carried out by means of light
and scanning electron microscopy. The macrographs of the surfacing
beads showed mainly columnar crystals in the weld metal
As from a 200x magnification and higher, a finer distribution of the
structure in the individual crystals in the vicinity of the fusion line is
observed. With increasing distance to the fusion line, a coarsening of the
crystal structure is observed, (Fig. 7). The coarsening is a result from the
alteration of the cooling conditions during the solidification. It results from
the decrease of the temperature gradient with increasing distance to the
fusion line.

Fig. 7. Fusion line between two weld beads in multiple layer welding, using the
SA Cold wire method, SEM micrographs
 SAW Cold Wire Technology 223

The separation processes are therefore, between high- and low-melting

phases, facilitated in the direction of the weld surface. For this reason,
dendrites with larger trunks and arms are developing with increasing
distance from the fusion line while the regions with the low-melting phases
between the dendrites are increasing .
In multiple-layer welds, the repetition of the crystal structure is
observed in the individual beads. The structure is finer in the vicinity of
the fusion line and coarser in the upper region. The grain boundaries are in
the vicinity of the fusion line (top weld bead) clearly narrower than in the
upper region of the weld bead (bottom weld bead).
Because of the melting down of these crystals with different structures
(Fig. 8) at the fusion line, inheritance of the formed and deep-melting
regions at the grain boundaries between lower and upper beads does not
occur. This may restrict the crack propagation through several weld beads.
In contrast to the fusion line between two weld beads in multiple-layer
welding, the transition between base metal and weld metal is frequently
marked by epitaxial crystal growth. The relatively large grain boundary of
the base material which is liquified at first during heating up, is
contributing towards that. The grain boundary restricts the incompletely
molten grains which act as nucleators for the weld metal crystals. Due to
the restricted solubility of this eutectic fraction it is further distributing
between the weld metal crystals (Fig. 9).

Fig. 8. Grain boundaries in the bottom and the top weld bead in multiple-layer
welding using the SA cold wire method. SEM micrographs
224 U. Reisgen et al.

Fig. 9. Inheritance of the grain boundary from the base material to the weld metal,
SEM-micrographs

If the critical tensile stresses are exceeded still in the high-temperature

range, widening of the grain boundaries may occur (Fig. 10). At that, the
width and the cross-linking of the grain boundaries, the temperature
dependence and the size of the change of volume during cooling and also
the high-temperature strength of the low-melting regions are decisive
factors.

Fig. 10. Widening of the grain boundaries (formation of micro-cracks) in the weld
metal, SEM micrographs
 SAW Cold Wire Technology 225

For a better clarification of these problems, further research work is re-

quired.
Figure 11 depicts two macro-sections. The macro-section (top) has been
taken from a welding with additional cold wire feeding; the macro-section
(bottom) has been taken from a welding without additional cold wire
feeding. Both welds have been carried out with approximately equal
welding parameters. The left and the right side show a magnification of the
structure from the centre regions of the weld.
Especially the centre regions of the welds depicted in the photographs
show clearly that the crystals (from the SA cold wire process) are directed
and almost vertically positioned into one direction.

SA cold wire
Um= 35,7 V, Im= 355 A

SA standard
Um= 35,7 V, Im= 353 A

Fig. 11. Macro-sections with detailed photographs of a weld, Cold wire deposition
ratio: 50% (top) , without cold wire addition (bottom), welding speed: 50 cm/min
226 U. Reisgen et al.

The deviation from the growth direction is lesser than in conventional

SA welding.
The angle where the crystals which are growing from both sides are
meeting in the centre of the weld is clearly smaller than the angle which
develops when using SA standard methods (Figs. 11 and 12).
The crystal growth direction in a SA standard process which is marked
by a larger angle of contingence and larger growth angles is frequently
enclosing deep-melting phases in the weld centre. Hot crack formation is
supported by this (Fig. 12).
As far as SA cold wire methods are concerned, these eutectics are,
through the smaller angle of contingence of the solidification fronts in the
weld centre, transported to the surface of the weld and are there absorbed
by the slag.
During the solidification, a change of volume takes place. This leads to
the development of shrinking stresses which are acting in the counter
direction of the dendritic growth and which are largest in axial position to
the dendrites. With a larger angle of contingence, for example, higher
tensile forces are acting on the centre region of the weld. Smaller angles of
contingence support the inflow of molten metal between the dendrites
which is capable to fill the micro segregation zones which had been caused
by solidification in front of the crystallisation front.
Figure 13 depicts three micrographs from welds which had been carried
out using the SA standard method and the SA cold wire method and two
different energies per unit-length. Different cooling conditions and, as a
consequence, changed solidification processes lead to different distribution
and sizes of micro defects (micro-shrinkages and/or micro cracks).

(a) (b)
Fig. 12. Centre region of the weld bead in the SA cold wire method (a), contacting
columnar crystals in the centre region of the weld bead in SA standard weld-
ing (b), SEM micrographs
 SAW Cold Wire Technology 227

(a)

(b)

(c)
Fig. 13. SA standard method, lower region, transverse section (a), SA cold wire
method, vs = 50 cm/min, lower region, transverse section (b), SA cold wire
method, vs = 70 cm/min, lower region, transverse section (c)
228 U. Reisgen et al.

While using the SA Cold wire method with a welding speed of 50

cm/min, the appearance of micro defects decreased. Their distribution on
the dendrite boundaries is more homogenous. Micrographs show that these
phenomena are less frequent in SA cold wire welds which had been
produced with a welding speed of 70 cm/min.
Hardness measurements where the regions HAZ and weld metal in the
root and flank region have been examined were carried out on these welds
(Fig. 14). When considering these macro-sections it is noticeable that the
hardness traverse in the base material is subject to higher fluctuation in the
event of welding with cold wire than it is during a conventional SA weld.
A possible reason for this might be the higher cooling speed which is a
result from cold wire addition.
The additional cold wire feeding resulted also in the hardness increase
in the lower region of the weld metal structure, Fig. 15. The influence of
the cold wire addition is hardly noticed in the upper region of the molten
pool since the cooling conditions of both methods are differing only
slightly. The reason is that the melting of the cold wire takes place in the
molten pool.

flank
flank
middle
middle

SA standard SA cold wire

Um= 35.7V, Im= 353A Um= 35.7V, Im= 355A

Fig. 14. SA standard method (left) and SA cold wire method (right)

Chemical Composition

Alloy 617 is a nickel-cobalt-molybdenum alloy with excellent strength and

creep properties through mixed crystal hardening of up to 1100°C. Due to
the balanced chemical composition, the alloy possesses a very good resis-
tance against high-temperature corrosion in the form of oxidation and car-
burization. In several tests, a connection between the aluminium content
and the creep rupture strength values has been established. The aluminium
 SAW Cold Wire Technology 229

content of between 1% and 1,2% resulted, in the case of alloy 617, in a

clear improvement of the creep rupture strength [7].
275

250 Centre cold wire

Hardness [HV1]

225

200

175
HAZ WM Centre standard
150
0 2 4 6 8 10
Distance [mm]

275

250
Flank cold wire
Hardness [HV1]

225

200

175 Flank standard

HAZ WM
150
0 2 4 6 8 10
Distance [mm]

Fig. 15. Comparison of the hardness traverse of a weld made with the SA standard
method and of a weld made with the SA cold wire method and a deposition ratio
of 50% in weld centre and flank (welding speed 50 cm/min)

In this connection, modified welding filler materials with higher

aluminium content were developed.
In standard submerged-arc welding, it is not possible to avoid the
relatively high burn-off of aluminium which leads to low and unsufficient
aluminium contents in the weld metal. It has been tried to find a solution
for this problem by using the SA cold wire technology.
230 U. Reisgen et al.

For the welding tests, a modified wire of the type UTP UP 6170 Co Mod.
has been used. The chemical composition of this wire is specified in Table 1.
For testing the influence of the cold wire addition on the Al content in the
pure weld metal, welds on the austenitic CrNi steel were carried out. The
specimens for the multiple-layer welding tests were built up in 5 layers.
The chemical composition of the material used in these welding tests was
each determined from the top layer and is shown in Table 2.

Table 1. Chemical composition of the welding wire and the base material
C Cr Mo Fe Al Co Ti Ni
[%] [%] [%] [%] [%] [%] [%] [%]
Wire analysis
0,05 21,77 8,38 0,24 1,23 10,44 0,31 57,00
acc. to ISF
Base material
analysis acc. to 0,05 18,47 0,27 <69,6 0,01 0,14 0,01 8,51
ISF

Table 2. Chemical composition of the pure weld metal (built up in 5 layers)

C Cr Mo Fe Al Co Ti Ni
[%] [%] [%] [%] [%] [%] [%] [%]
30V, 250A,
0,04 21,89 8,17 0,33 1,01 10,51 0,22 57,40
As,KD/As,E = 100%
30V, 350A,
0,04 21,78 8,14 0,48 1,09 10,49 0,24 57,40
AS,KD/AS,E =100%
35V, 250A,
0,04 21,87 8,25 0,36 0,99 10,47 0,21 57,40
AS,KD/AS,E =100%
35V, 350A,
0,04 21,90 8,15 0,66 1,00 10,44 0,21 57,20
AS,KD/AS,E =100%
35V, 250A,
0,04 21,63 8,06 1,47 0,77 10,36 0,17 57,00
AS,KD/AS,E =0%

Four welding tests with a cold wire addition of 100% of the electrode
mass have been carried out and the welding parameters voltage and current
have been varied. The table shows that the Al burn-off remains constant
with approximately 1%. The fifth test was carried out without additional
cold wire feeding. The chemical analysis shows a clear decrease of the Al
content down to 0.77%.
In comparison to the conventional SA welding, the tests with cold wire
addition show a clear increase of the Al content in the weld metal. The
reason is that through the trailing arrangement of cold wire addition the
melting of the cold wire takes place at distinctly lower temperatures and is
thus significantly reducing the aluminium burn-off.
 SAW Cold Wire Technology 231

Joint Welding

The wire diameters 1.2 mm, 1.6 mm and 2 mm were examined. For all
welds, electrodes and cold wire with equal diameters have been applied.
During the tests it has been established that with the increase of the
current, the melting potential of the molten pool was also increased.
The application of the electrode diameter of 1,2 mm resulted in a slight
increase of heat input only. This was the reason why only a restricted,
additional quantity of cold wire was fed to the process. Also, very precise
positioning of the cold wire was necessary. Therefore, the process stability
was not always ensured. With the application of the electrode diameters
1.6 mm and 2.0 mm, a very high process stability was achieved.
Figure 16 shows an example of a joint on alloy 617, plate thickness: 16
mm, and an electrode diameter of 2 mm. For the weld preparation, a V-
type weld with a preparation angle of 60° had been chosen. Both plates
were firmly clamped during welding. The root pass was welded manually
using the TIG method. The 2 filler passes and 2 final weld passes were
welded with the SA method, with a cold wire addition of 47% (Table 3).
The macro-section in the picture shows good overlap and reliable fusion to
the weld flanks. With those final weld passes, a good weld interface with
the base material was obtained. A distinct increase of the deposition rate
was achieved.
It has been observed that, also during welding in the groove, most
crystals of each individual bead are continuing uninterruptedly from the
fusion line to the weld surface. This facilitates the transport of the low-
melting phases in the slag which again results in a positive influence on the
hot cracking resistance [5].
The bead reinforcement and the increased deposition rate which are
typical of the SA cold wire method have also been observed during
welding tests made in the groove.
The slightly reinforced beads in the groove lead to the reduction of the
tensile stress peaks which may result in longitudinal solidification cracks
in the centre region of the weld bead [5].
Figure 17a shows a horizontal projection of the joint weld surface where
a dye penetration test had been carried out. Surface cracks have not been
observed. The results from the radiography examinations which are
depicted in Fig. 17b do not show any slag inclusions, cracks or other
discontinuities in the joint weld.
232 U. Reisgen et al.

16mm
root (TIG)

Fig. 16. Transverse section of a joint weld (Alloy 617) with a TIG welded root
pass

(a) (b)
Fig. 17. Weld surface (a) and radiograph (b) of the weld

The next step was the examination of the cold wire influence on the
angular distortion. For that purpose, joint welds with and without
additional cold wire feeding have been carried out where one plate had
been fixed and the other plate was allowed to shrink without restrictions
(Fig. 18).
Through the cold wire addition which was caused by the higher
deposition rate, the number of necessary weld beads was reduced.
Moreover, the angular distortion was clearly reduced if welding was
performed with cold wire addition
For a comparison of the supporting effect of the individual passes in
welding, the welding speed in welding with cold wire addition had been
increased in order to enable welding with the same weld build-up as in
conventional SA welding. In this case also a clear decrease of the angular
 SAW Cold Wire Technology 233

distortion was the result (Fig. 18). The exact influence of heat input and
weld build-up is still subject to more detailed examinations.

Table 3. Welding parameters of the joint weld

Layer 1 2 3 4
I [A] 358 359 358 359
U [V] 35,0 34,9 35,0 34,9
vEL [m/min] 4,19 4,19 4,19 4,19
vKD [m/min] 1,95 1,95 1,95 1,95
vs [cm/min] 50 50 50 50
Es [kJ/cm] 18,80 18,79 18,80 18,79
As,KD/As,El [%] 47% 47% 47% 47%
Lab [kg/h] 10,49 10,49 10,49 10,49

110mm
Clamping force

16mm
30 16

25 26,9 mm
15,3° 12
20 22 mm
Distortion [mm]

11,9°
Distortion [°]

19,4 mm
15 10,4° 8
5 layers 7 layers 7 layers
10
SA cold wire SA Standard SA cold wire 4
A=50% A=0% A=50%
5 Um= 35,5 V Um= 35,7 V Um= 35,8 V
Im= 371 A Im= 359 A Im= 358 A
vS=50cm/min vS=50cm/min vS=70cm/min
0 0

Fig. 18. Comparison of the distortion values using the SA standard method and
the SA cold wire method
234 U. Reisgen et al.

Hot Cracking Tests using MVT-Tests

For a closer examination of the hot cracking tendency, MVT tests were
made on SA welds. This is a sensitive and material-oriented testing
method where the specimen is loaded externally. In the MVT test, the
specimen is bent around a die with a defined radius and a defined speed.
During the bending; the SA weld bead is re-molten by a TIG arc.
The three regions which are depicted in Fig. 19 (weldable, restrictedly
weldable and not weldable) have been determined empirically through a
large number of tests at the BAM, Berlin and have been stored in an expert
data base. These three regions serve orientation purposes only since
material-specific discrepancies may occur.
The diagram shows the results from the longitudinally welded SA
specimens with the dimensions 100 mm × 40 mm × 10 mm. For the SA
welding tests, constant welding current and constant welding voltage have
been applied. The energy-per-unit length had been realised through the
variation of the welding speed, its length is 50 cm/min and 70 cm/min,
respectively.

30,0
Sektorgrade 1
27,5
Sektorgrade 2
25,0 KD 50-50 L2
22,5 St L
Total crack length [mm]

20,0 KD 50-70 L
KD 50-50 L1 Weldable
17,5
with restrictions
15,0
not
12,5 weldable
10,0

7,5
5,0

2,5 weldable

0,0
0 1 2 3 4 5
surface bending strain [%]

Fig. 19. Total crack length after application of the MVT test
 SAW Cold Wire Technology 235

The MVT tests have been performed at the BAM, Berlin. The
Varestraint mode has been applied. All specimens have been tested with a
travelling speed of 2.0 mm/s, 4% - strain and energy per unit length of
7.5 kJ/cm. The entire length of the hot cracks which have been produced
during the test will be evaluated.
It is, however, important to mention that all results from the SA cold
wire welded specimens are all within a close range and are forming a
group which is close to the restrictedly weldable area. The diagram shows
that the results do not show a clear difference with regard to the changed
energy-per-unit length caused by the increase of the welding speed from
50 to 70 cm/min.
As already mentioned before, the addition of the cold wire is the
dominating factor with regard to the deposition of the micro-alloying
elements into the weld. This had been confirmed through the chemical
analysis of the SA weld metal by means of MVT tests where a similar
arrangement with regard to the Al content has been determined (Table 4).
The low Al content of the weld made with the conventional SA method is
correlating with a somewhat lower hot cracking susceptibility.

Table 4. Chemical composition of the SA weld metal of the MVT specimens

C Cr Mo Fe Al Co Ti Ni
[%] [%] [%] [%] [%] [%] [%] [%]
36V, 367A,
AS,KD/AS,E =0%, 0,02 21,52 8,35 0,56 0,90 10,65 0,29 57,30
vs=50 cm/min
36V, 367A,
AS,KD/AS,E =50%, 0,03 21,29 8,59 0,57 0,97 10,63 0,31 57,20
vs=50 cm/min
36V, 368A,
AS,KD/AS,E =50%, 0,04 21,87 8,37 0,45 1,01 10,67 0,27 56,90
vs=50 cm/min
36V, 352A,
AS,KD/AS,E =50%, 0,04 21,30 8,58 0,57 1,01 10,68 0,31 57,10
vs=70 cm/min

Tests, carried out by other institutions, confirmed that the high Al

content was responsible for the increase of the strength and the reduction
of the ductility and the impact strength (internal research by company
UTP). It is likely that the hot cracking susceptibility is also changed on a
material level. This must be examined in more detail.
In the MVT test the remelting by a TIG arc changes the structure and
has influence also on the effects of the cold wire.
236 U. Reisgen et al.

For a closer examination of the process-technical influence of the cold

wire, a PVR test would be reasonable. For this test, the specimens should
be welded with the SA process. Here, not only the material side is
considered, but also the process aspects and their influence on the hot
cracking susceptibility.

Industrial Implementation
Finally, a pipe weld had been carried out at the company Essener
Hochdruck Rohrleitungsbau, using the SA cold wire method.
For this, two tube sections made of the material Alloy 617 with a wall
thickness of 78 mm have been welded. A common circumferential weld
preparation with a weld angle of approximately 24° has been chosen. The
root pass was welded using the TIG method. The groove was then filled by
manual arc welding up to a width of approximately 20 mm. The remaining
passes were welded with the SA method and additional cold wire feeding.
The applied weld current was 290–310 A and the voltage was
approximately 32–33 V. The welding speed was 45 cm/min. The cold wire
was added with a ratio of 50% of the electrode feed speed.

Conclusion

SA welding with cold wire addition exerts a considerable influence on the

molten pool geometry. Among the influenced factors are the width, the
depth, the length and also the shape of the molten pool.
By the addition of cold wire into the liquid molten pool, the cooling
conditions and the solidification behaviour of the molten metal are
strongly influenced. One reason is that the heat potential of the molten
pool is better exploited. A constant quantity of heat input allows the
melting of larger quantities of filler material; the cross-section of the weld
may also be increased. Examinations of the structure and the hardness
measurement confirm these observations.
The trailing cold wire feeding arrangement allows the reduction of the
burn-off of some alloying elements which has a direct influence on the
mechanical properties.
A connection between the alloying elements in the weld metal and the
MVT results was also established.For the completion of the examinations
about the influence of cold wire addition on the hot-cracking susceptibility
with regard to material and process-related aspects, it is inevitable to carry
out further tests.
 SAW Cold Wire Technology 237

The tests which have been carried out at the ISF demonstrated that the
addition of cold wire resulted in a distinct reduction of distortion,
compared with the conventional SA method.
The practicability of the process modification was demonstrated through
the industrial implementation of the test results. It has been possible to
realise the increase of the deposition rate by approx. 50%, referring to the
deposition rate of the conventional SA welding methods.

References

1. Schulze G (2004) Die Metallurgie des Schweißens. Spinger, Berlin

2. Böllinghaus T, Herold H (2005) Hot Cracking Phenomena in Welds. Springer,
Berlin
3. Radaj D (1988) Wärmewirkungen des Schweißens. Springer, Berlin
4. Corporate Publication (1996) Erstarrungsformen von Rissen und Brüchen
metallischer Werkstoffe.Verlag Stahleisen GmbH, Düsseldorf
5. Schuster J (2004) DVS-Bericht 233, Heißrisse in Schweißverbindungen.
DVS, Düsseldorf
6. Karlsson L, Arcini H (2003) Synergic Cold Wire (SCWTM) Submerged
Arc Welding of Highly Stainless Steels. Stainless Steel World 2003, KCI
Publishing BV
7. Corporate Publication (2005) Werkstoffdatenblatt Nr. 4119 Nicrofer 5520 Co,
ThyssenKrupp VDM
8. Corporate Publication (2000) Werkstoffdatenblatt Nr. 4118 Nicrofer 6020 hMo,
ThyssenKrupp VDM
9. ISF-Aachen (2007) Verbesserung der Heißriss-Sicherheit beim UP-Schweißen
von Nickelbasislegierungen unter dem Aspekt gesteigerter Wirtschaftlichkeit
Forschungsvorhaben AiF-Nr.: 13.864N, DVS-Nr.: 1.049
 Part III
Solidification Cracking of Aluminium Alloys
Hot Tearing During Laser Butt Welding
of 6xxx Aluminium Alloys: Process Optimisation
and 2D/3D Characterisation of Hot Tears

D. Fabrègue1, A. Deschamps2, M. Suéry2, H. Proudhon1

1
INSA-Lyon, CNRS, MATEIS, Villeurbanne, France
2
INP-Grenoble, CNRS, SIMAP, Saint-Martin d’Hères, France

Abstract

The influence of process parameters on the occurrence of hot tearing during

laser butt welding of 6xxx aluminium alloys with the presence of a filler
wire is investigated. Lowering the welding speed and the solidification rate
leads to welds with fewer defects. The chemical composition of the weld
pool is shown to have a strong influence on hot tearing as well as the
fastening device. The number of hot tears drops drastically when high
silicon content is attained in the weld pool and when a vacuum fastening
device is used. Hot tears are then characterised. They are shown to be
intergranular, coming from fracture of liquid films without plasticity of the
surrounding grains. The hot tear exhibits a round shape around the equiaxed
grains at the centre of the fusion zone and a planar part between the
columnar grains of the weld bead. By using three dimensional X-Ray
tomography, the exact shape of the hot tears has been visualised.

Introduction

The use of laser welding as a joining method is in rapid development in the

industry, particularly in the aerospace sector, owing to its potential for
weight savings and gains in productivity as compared to riveting.
However, further development of this joining technique requires both an
increase in the weld performance and service life (high joint coefficient,
high strength, low distortions and residual stresses, good fatigue and stress
corrosion properties, …) [1], and an increase in the robustness of the
242 D. Fabrègue et al.

welding process, especially in terms of defect control [2]. In order to

obtain a fusion zone free of defects such as porosity or hot tears, the
welding parameters need to be optimized. Among the important process
parameters one finds the laser power, the welding speed and quantity of
filler wire, which have a major role in controlling the microstructure and
mechanical properties of the fusion zone and the adjacent material, namely
the Heat-Affected Zone.
Very few studies have been carried out, which concentrate on the micro-
structure of laser welds, contrary to more classical welding processes such as
arc welding [3, 4, 5]. Usually these studies deal with high strength alloys for
the aerospace industry such as the 2xxx or the 7xxx series [6]. Most of the
published work on laser welding is dedicated to parameter optimization,
especially to decrease porosity levels in the weld [7], but literature on hot
tearing during laser welding is quite limited. The present paper is concerned
firstly with the study of the influence of process parameters (welding speed,
quantity of filler wire, fastening device, …) on the hot tearing phenomenon in
laser welds of 6xxx Al alloys series and secondly with the observation of the
hot tears. The detailed microstructure of such welds has been already
described in a former paper [8]. This study will concentrate on the observation
of hot tears firstly thanks to usual micrography in different planes of the laser
weld. Characterization by Synchrotron X-Ray tomography has been
conducted to obtain 3D information on the defect.

Experimental Procedure

The base material used in this work is AA6056, the chemical composition of
which is given in Table 1. The material in the shape of 1.6 mm thick sheets
was butt welded in the T4 state (namely after a solution treatment, water
quench and several months of natural ageing). Welding was carried out
along the rolling direction, on 400 mm × 150 mm samples, after degreasing.
Butt joints were produced with a Nd:YAG laser operating at 3 kW as shown
in Fig. 1 [8]. The focal point of the laser beam was set at the surface of the
sheets.

Table 1. Composition of 6056 and AS12 alloys (weight percent)

Alloy Mg Si Cu Mn Fe
6056 0.86 0.92 0.87 0.55 0.19
AS12 0.1 12 0.3 0.15 0.8
(4047)
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 243

Filler wire Laser beam Shielding gas

Sheets
Welding direction to weld

Fig. 1. Welding set up [8]

A 1 mm diameter filler wire of AA4047 (12%Si, see Table 1 for exact

composition) was employed and helium was used as shielding gas at a
flow rate of 20 l.min–1. This particular filler wire has been chosen owing to
its ability to prevent the formation of hot tearing [9]. For the different
welding experiments, the welding speed was varied from 4 m.min–1 to 6
m.min–1 and the filler wire speed from 0 to 6 m.min–1. The observation of
hot tears has been carried out at a welding speed of 6 m.min–1 and a filler
wire speed of 1.5 m.min–1. This choice of speed parameters ensures that
numerous hot tears are obtained in the weld. Two different fastening
systems have been investigated: a mechanical one using screws and a
vacuum one using low pressure to maintain sheets on the welding table. In
order to characterize the influence of the cooling rate, inserts have been
used to change the thermal contact between the sheets and the steel
welding table. Decreasing the cooling rate has been realized by inserting
cork whereas brass has been used to increase it. One configuration with
brass below and on top of the sheet has also been used (sandwich
configuration). The number of hot tears has been characterized by using
X-ray radiography. Optical observations have been carried out in the
rolling plane. Samples for optical observations were polished and etched
for 20 s with a Flick solution (10 ml HF, 20 ml HNO3 and 95 ml water).
Specimens for EBSD characterization were obtained after mechanical
polishing by electropolishing with a solution of 330 ml nitric acid and
244 D. Fabrègue et al.

660 ml methanol. Samples for X-Ray tomography have been machined to

obtain a stick of about 10 cm long, 2 mm wide and 2 mm thickness. It is
constituted by the weld line and some base material on the sides. X-ray
tomography has been carried out at the ESRF (European Synchrotron
Radiation Facility, Grenoble) on the ID19 beam line. The specimen was
imaged at 70 keV using 900 projections at different angular positions.
The 3D volume was then reconstructed via a filtered back projection
algorithm and further analyzed with the commercial software VGStudioMax
(http://www.volumegraphics.com/products/vgstudiomax/). More details
about this technique can be found in [10].

Results

Influence of Welding Parameters on the Occurrence of Hot Tearing

This first part is dedicated to the study of the number of hot tears on a
given length of weld (40 cm) as a function of the process parameters.
Figure 2 shows the influence of the welding speed on the occurrence of hot
tearing for different filler wire speeds. The number of hot tears increases
when the welding speed increases whatever the filler wire speed. These
results are in agreement with those obtained by Cicala [11] on the same
alloy but for different welding and filler wire speeds. This increase of the
sensitivity to hot tearing can be due either to a higher instability of the
weld pool (leading to the closure of the keyhole) or/and to the higher
solidification rate when the welding speed increases.
The influence of the solidification rate on hot tearing has been widely
studied over the past. It has been shown in particular that the presence of
a second beam to decrease the cooling rate decreases the number of hot
tears [7]. Early in the 70’s, some models, predicting the sensitivity to hot
tearing, considered that decreasing the solidification rate lead to less
defects [12, 13]. To study the influence of the cooling rate different
inserts have been used between the sheet to weld and the welding table
(Fig. 3). These experiments confirm that the lower the cooling rate, the
lower the number of hot tears. This could partly explain the higher
number of hot tears observed in Fig. 2 when the welding speed was
increased to 6 m.min–1.
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 245

Figure 2 also shows that the quantity of filler wire has an influence on
the occurrence of hot tearing: for higher filler wire speeds, the number of
hot tears drops whatever the welding speed. The filler wire speed changes
the chemical composition of the weld pool and thus of the weld nugget. In
our case, the silicon concentration will increase as the quantity of filler
wire increases. The silicon concentration has been calculated using optical
observation and dilution consideration (for more details see [14]). Figure 4
shows the number of hot tears as a function of the silicon composition. The
number of hot tears is high for low silicon content and then drops when the
composition in silicon increases. This is in accordance with the literature
[15]: for most of the alloying elements (Mg, Si, Cu), the hot tearing
sensitivity during casting decreases with increasing alloying element
concentration. This study shows that in the case of laser welding, the Si
composition acts in the same way. Further experiments are required to
verify and explain the minimum of hot tears observed for 2% of Si.

60
Nbr. of hot tears for a 40cm weld

50 fv=0
fv=1m/min
fv=2m/min
40
fv=2.5m/min
fv=3m/min
30

20

10

0
3,5 4 4,5 5 5,5 6 6,5
welding speed (m/min)

Fig. 2. Influence of the welding speed on the number of hot tears in a 40 cm long
weld for various filler wire speeds (fv=filler velocity)
246 D. Fabrègue et al.

100
Nbr. of hot tears for a 40cm weld

80

60

40

20

0
- 0 + ++
Thermal contact between sheet and support

Fig. 3. Influence of the thermal contact between the sheets to weld and the
welding table on the number of hot tears. The welding speed is 6 m/min and the
filler wire speed is 1.5 m/min. The sign – corresponds to cork insert whereas the
signs + and ++ correspond to an insert of a single sheet of brass and the sandwich
configuration respectively

At last, the sensitivity to hot tearing could be influenced by the fastening

system. Figure 5 presents the results obtained with the two types of
fastening devices used in this study. The welding speed has been chosen in
order to observe the maximum of hot tears with the classical fastening
device (mechanical one). Figure 5 underlines the large influence of the
fastening system on the occurrence of hot tearing during laser welding. FE
calculations would be necessary to understand the difference on
stresses/strains involved by the different fastening systems but these results
definitely show that the vacuum fastening system must be preferably
employed to obtain defect free welds. The fastening system can then be
considered as a parameter of prime importance.
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 247

60
Nbr. of hot tears for a 40cm weld

50

40

30

20

10

0
0,5 1 1,5 2 2,5 3 3,5
Silicon content of the weld nugget
Fig. 4. Influence of the silicon content on the number of hot tears in a 40 cm long
weld for a welding speed of 6 m/min

60
Nbr. of hot tears for a 40cm weld

mechanical
50 vacuum

40

30

20

10

0
0 0,5 1 1,5 2 2,5 3 3,5
Filler wire speed (m/min)

Fig. 5. Influence of the fastening device on the number of hot tears for various
filler wire speeds. Welding speed is equal to 6 m/min

Characterisation of Hot Tearing Phenomenon

Figure 6 shows the EBSD cartography of the fusion zone perpendicular to the
welding direction. A hot tear is observed in black. On this figure the grain
248 D. Fabrègue et al.

colour is randomly chosen and does not correspond to a preferential crystallo-

graphic direction. Indeed, no determined texture has been observed in the
fusion zone. However, the EBSD picture confirms that the phenomenon of hot
tearing is intergranular as already mentioned in the literature [16].
Figure 7 shows the fracture surface characteristic of the phenomenon of
hot tearing in laser welding. The surface exhibits the dendrites which
originate from the solidification process. Moreover, the dendritic surface is
smooth indicating no plasticity. This leads to the conclusion that hot
tearing is the result of the fracture of liquid films as already observed for
hot tearing during casting [9]. Solidification stresses/strains pull apart the
dendrites until a cavitation pressure is attained in the liquid and the
fracture takes place [17]. Note that the secondary arm spacing of the
dendrites observed in Fig. 7 is about 5 μm. Using the equation [11]:
3
§λ ·
tf = M ¨ 2 ¸
© 5.5 ¹
with λ2 the secondary arm spacing, and M a coefficient dependant of the
material, allows calculating the solidification time t f and thus the cooling
rate. With M equal to about 4.10–8 m3.s–1 as in [18] and considering the
solidification interval of this alloy, the cooling rate is about 400 K.s–1. This
high cooling rate is characteristic of the laser welding process.

Fig. 6. EBSD map of the fusion zone perpendicular to the welding direction.
Black zones correspond to the hot tear
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 249

Fig. 7. SEM observation of the hot tearing surface

Hot tearing takes place at the very end of the solidification process [17],
usually for solid fractions higher than 0.9. During laser welding,
solidification occurs first at the fusion line because the main phenomenon
responsible for cooling is conduction in the solid [19] (heat extraction is
mainly assures by this means). Figure 8 shows the micrograph in the
rolling plane of a laser weld containing hot tears. From this figure, a hot
tear can be seen. The crack exhibits a quite planar shape when it goes
between the columnar grains and a round shape around the equiaxed grains
of the centre of the fusion zone. The hot tear results from the strains
applied on the mushy zone. These stresses come from firstly the
solidification shrinkage which is quite high in aluminium alloys (except
for the 4xxx series) and secondly from thermal contraction of the solid
skeleton. Moreover external loads (for example coming from the
fastening) can apply on the mush zone.
250 D. Fabrègue et al.

Equiaxed
grain

Columnar
grain

Base material
Welding Fusion line
direction

Fig. 8. Optical micrograph of the fusion zone seen from above

Observation of the crack by X-ray tomography gives a more general

picture of the exact shape of the hot tear and of its way of propagation.
Figure 9 shows a 2D image where the hot tear is clearly visible. The
extraction of the 3D shape of the crack can be done by a grey level
segmentation process. The result of this operation is shown in Fig. 10. The
surface of the weld bead is represented here in grey. As it can be seen two
hot tears (highlighted by two different colours) can be observed in this
weld. The 3D tomography permits firstly to conclude that the two parts of
the red tear are actually only one defect. This could not have been seen on
a 2D picture. Moreover, Fig. 10 shows perfectly the linear part of the tears
following the basaltic grains of the fusion zone: the surface is quite planar.
Figure 11 exhibits the same hot tears seen from above. The outer surface
of the fusion line is still outlined in grey. This view permits one to observe
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 251

more accurately the centre zone of the hot tears. The deviation around the
equiaxed grains of the centre, as guessed from the 2D micrograph (Fig. 8),
is clearly visible here. The planar surface of the borders becomes more
round as the crack goes around the grains.

Outer
surface

Hot tear

Fusion
zone

Base
material

Welding
direction

Fig. 9. X-ray picture showing a hot tear. The rings are artefact
252 D. Fabrègue et al.

Welding
direction

Fig. 10. 3D X-ray tomography of the fusion zone containing two hot tears. The
grey surface represents the outer surface of the fusion line
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 253

Welding
direction

Fig. 11. 3D X-ray tomography of the fusion zone seen from upward containing
two hot tears. The grey surface represents the outer surface of the weld bead

Conclusions

This paper firstly allows a better understanding of the key process

parameters controlling the quality of laser welds in term of hot tearing.
Welds with fewer defects can be achieved by decreasing the welding speed
and the solidification rate and by increasing the silicon content in the weld
pool. The vacuum fastening device improves significantly the quality of
the welds compared to the classical mechanical fastening device.
Secondly this study permits a better understanding of the hot tearing
phenomenon. Optical microscopy, SEM and EBSD clearly show the
intergranular character of this defect. Hot tearing occurs by fracture of the
254 D. Fabrègue et al.

liquid films between dendrites without any plasticity. Thanks to various

2D micrographs and to 3D tomography, propagation of hot tearing during
laser welding has been characterised: it partly goes around equiaxed grains
located in the centre of the fusion line and partly follow the columnar
grains on the sides of the weld bead. All these results could be used in a
more general finite element simulation of the laser welding process.
Coupled to the rheological behaviour of the mushy zone (determined
elsewhere [20, 21]) and a hot tearing criterion, it would permit a prediction
of the occurrence of hot tears.

Acknowledgements

The authors want to thank L. Salvo and J.-J. Blandin for their help during
the ESRF experiments. The authors are grateful to the French Ministry of
Industry for funding in the framework of the ASA (Allègement des
Structures dans l’Aéronautique) RNMP project, carried out in
collaboration with EADS and ALCAN, which are also thanked for
providing the materials. L. Kirschner of EADS-CCR and J.C. Ehrström of
ALCAN CRV are acknowledged for fruitful discussions.

References

1. Shinzel C, Hohenbeger B, Dausinger F, Hugel H (2000). Laser welding of

aluminium – extended processing potential by different wire positions.
Proceedings of SPIE, Eds. C. Xiangli, F. Tomoo, A. Matsunawa, Osaka,
Vol. 3888, 380.
2. Seto N, Katayama S, Matsunawa A (2000). High-speed simultaneous
observation of plasma and keyhole behavior during high power CO2 laser
welding: effect of shielding gas on porosity formation. Journal of Laser
Applications 12(6): 245–250.
3. Kuk JM, Jang KC, Lee DG, Kim IS (2004). Effects of temperature and
shielding gas mixture on fatigue life of 5083 aluminum alloy. Journal of
Materials Processing Technology 155–156: 1408–1414.
4. Menzemer C, Lam PC, Srivatsan TS, Wittel CF (1999). An investigation of
fusion zone microstructures of welded aluminium alloy joints. Materials
Letters 41: 192–197.
5. Hou KH, Baeslack WA (1996). Characterization of the heat-affected zone in
gas tungsten arc welded aluminium alloy 2195-T8. Journal of Materials
Science Letter 15: 239.
6. Liu C, Northwood DO, Bhole SD (2004). Tensile fracture behaviour in CO2
laser beam welds of 7075-T6 aluminium alloy. Materials and Design 25:
573–577.
 Hot Tearing During Laser Butt Welding of 6xxx Aluminium Alloys 255

7. Haboudou A, Peyre P, Vannes AB, Peix G (2003). Reduction of porosity

content generated during Nd:YAG laser welding of A356 and AA5083
aluminium alloys. Materials Science and Engineering A 363: 40–52.
8. Fabrègue D, Deschamps A (2002). Microstructural study of laser welds
Al6056-AS12 in relation with hot tearing. Materials Science Forum,
Proceedings of the 8th Int. Conf. on Aluminium Alloys, Eds. P.J. Gregson,
S. Harris, Cambridge 396–402, 1567–1572.
9. Instone S, StJohn D, Grandfield J (2000). New apparatus for characterising
tensile strength development and hot cracking in the mushy zone. International
Journal of Cast Metals Research 12: 441–456.
10. Proudhon H, Buffière JY, Fouvry S (2007). Three-dimensional study of a
fretting crack using synchrotron X-ray micro-tomography. Engineering
Fracture Mechanics 74(5): 782–793.
11. Cicala E (2203). Etude de la fissuration à chaud des alliages d’aluminium
6056T4 soudés par lasers Nd :YAG continu. Post doctoral research report,
LTM, Université de Bourgogne.
12. Clyne TW, Davies GJ (1979). Comparison between experimental data and
theoretical predictions relating to dependence of solidification cracking on
composition. In: Solidification and Castings of Metals, The Metals Soc., 275–278.
13. Feurer U (1976). Mathematisches Modell der Warmriβneigung von binären
Aluminiumlegierungen. Giesserei-Forschung 28: 75–80.
14. Fabrègue D (2004). Microstructure et fissuration à chaud lors du soudage
Laser d’alliages d’aluminium 6056. PhD thesis, INPG.
15. Zao H, Debroy T (2001). Weld metal composition change during conduction
mode laser welding of aluminium alloy 5182. Metallurgical and Materials
Transactions B 32. 163–172.
16. Spittle JA, Brown SGR, James JD, Evans RW (1996). Mechanical properties
of partially molten aluminium alloys. Proceedings of the 9th Symposium on
Physical Simulation of Casting, Hot Rolling and Welding, ed. Suzaki, Sakai
and Matsuda, 81–91.
17. Rappaz M, Drezet JM, Gremaud M (1999). A new hot tearing criterion.
Metallurgical and Materials Transactions A 30(2): 449–456.
18. Kurz W, Fisher DJ (1989). Fundamentals of Solidification. Third Edition,
Trans Tech Publications, Ltd., Aedermannsdorf.
19. El-Ahmar W, Julien JF, Gilles P, Taheri S, Boitout F (2006). La robustesse de
la simulation numérique du soudage TIG, application sur des structures en
acier 316L. Proceeding of the Matériaux 2006 conference.
20. Fabrègue D, Deschamps A, Suéry M, Drezet JM (2006), Non-isothermal
tensile tests during solidification of Al-Mg-Si-Cu alloys: Mechanical
properties in relation to the phenomenon of hot tearing. Acta Materialia
54(19): 5209–5220.
21. Fabrègue D, Deschamps A, Suéry M, Poole WJ (2006). Rheological behavior
of Al-Mg-Si-Cu alloys in the mushy state obtained by partial remelting and
partial solidification at high cooling rate. Metallurgical and Materials
Transactions A 37(5): 1459–1467.
The Integral Approach – a Tailored Method
to Optimize Structural Behavior and Weldability

H. Gruss1, A. Pshennikov2, H. Herold3

1
MBDA/LFK, Unterschleißheim, Germany
2
Otto-von-Guericke University, Institute for Machine Design, Magdeburg,
Germany
3
Otto-von-Guericke University, Institute for Materials and Joining
Technology, Magdeburg, Germany

Abstract

The structural design at current welded skin-stringer-joints is geared to a

riveted structure. Consequently, the geometrical discontinuities between
skin and stringer are responsible for the stress concentration in the weld
seam at stringer ends.
Many projects related to welded fuselage structures investigate prob-
lems concerning either the weldability or the structural behaviour under
flight phase conditions, but not the structural behaviour in terms of the
flight phase and the welding process in its complexity.
Hence, it is important to focus on the development of a complex ap-
proach to improve the weldability and to work over the designing for
welding of the current welded structure. The Integral Approach enables to
combine the structural behavior under welding design criterions with an
optimization of weldability. The stringer geometry is modified by the im-
plementation of slots in the stringer web.
From the technical welding experiences points to the tendency that the
weld seam cracks may be reduced by the so called clamping effect. A tai-
lored welding strategy in combination with slots leads to desired residual
compressive stresses in several particular stringer parts. These stresses
prevent the volume expansion due to the heat input and the hot crack resis-
tance grows up.
On the other hand during the flight phase the stress concentrations at
the bearing weld seam ends decrease due to the implementation of
258 H. Gruss et al.

slots. The influences of the same geometrical parameters and welding

directions on the structural behavior are shown for the flight phase as
well as the weldability.

Introduction

Ten years ago laser beam welding was establish in the commercial aircraft
industry (Fig. 1). At the present time skin and stiffener in form of stringers
are welded in the lower shells of some sections at Airbus types A318,
A340 and A380 (Fig. 2). Against the riveted structures the welded ones
have a completely different fracture mechanical behavior. At riveted joints
the stringer is able to take over loads when at the same time the skin col-
lapses. By contrast skin and stringer fail simultaneously at a welded struc-
ture. The different behavior bases on the differential and integral design.
New aircraft aluminum alloys are designed after criteria like weight, re-
sidual strengths, damage tolerance or corrosion but not under considera-
tions concerning weldability. New materials like AA2098 or 7xxx-alloys
contain elements like Li or Zn. Both are characterized by a high vapor
pressure. This circumstance provokes instabilities of capillary during the
welding. Experiments have shown that these base alloys have a lower so-
lidification cracking resistance compared to the applied alloys in fuselage
like AA6013 or AA6056 [1].
Solidification cracks result from transverse shrinkage. They extend
some millimeters and could be visible on the weld seam surface (Fig. 3) or
limit on the inner weld seam. Hilbinger argues that the reason for the ap-
pearance of hot cracks lies in the magnitude of its dimension [2]. That is
that microscopic cracks based on process instabilities and macroscopic
ones have a global cause like process design and position of weld seam.
Solidification cracks belong to the second group.
The Integral Approach enables the combination of the structural behav-
ior under welding design criterions with an optimization of weldability.
The objective of this approach is to improve the performance of the
welded structure and the transition to the upper shell. There are higher ten-
sile stresses in different to the lower one. A structural compressive stress is
somehow preferred for reduced susceptibility to hot cracking because
compressive stresses work against the thermal expansion due to the local
heat input and heat conduction in the structure.
 The Integral Approach 259

Fig. 1. Joining of stringer and skin by both sides simultaneous laser beam
welding

Fig. 2. skin and stringer are welded in the lower shell of the section no. 16 by
A318 (gray-coloured)

Fig. 3. Position of a solidification crack at a welded T-joint configuration. Base

metal: AA6056, filler: AlSi7
260 H. Gruss et al.

Interaction Between Structural Behavior and Weldability

using Integral Approach

The stringer geometry is modified by the implementation of slots in the

stringer web. The outer zone is called secondary part; the inner one is the
primary part. The weld seam in the primary part consists of the inner and
the outer primary weld seams (Fig. 4).

Fig. 4. Modification of the stringer with the Integral Approach by designed two
slots, each at the stringer end. It is only shown one end to simplify matters

The secondary parts, that are located at outer stringer regions lead to a
decrease of the discontinuous between skin and stringer (eccentricity) and
therefore to a higher stiffness at the ends of the outer primary parts. In this
way the bending stresses that result by the loads during the flight phase are
reduced locally. The notch stresses that depend on the bending stresses de-
crease as well.
From the technical welding point of view the tendency that the weld
seam cracks is reduced by the so called clamping effect. After welding of
the inner primary and the secondary parts both areas which surround the
weld seams shrink. The still not-welded outer primary part is pulled down.
In these areas residual compressive stresses are accumulated. With the
welding of the outer primary part the compressive stresses are induced as
well, but both compressive stresses have an opposite direction to each
other. Therefore the material expansion is impeded by the heat input, re-
sulting in a lower volume shrinkage due to the cooling and the resulted
tensile stresses on the mushy zone are lower as well.
This work is focussed on the structural behaviour during the welding
process, less the material behaviour. The next sections prove the influence
of slot on the structural behavior as well as on the hot cracking behavior.
 The Integral Approach 261

Slot Influence on Structural Behavior

The fuselage consists of several sections which each other are joined with
stringer couplings. Each coupling is composed of one short stringer and a
circumferential butt strap. At all Airbus types couplings are located below
a frame (Fig. 5). The cabin pressure and the self-weight of the fuselage
lead to tensile stresses in the structure.

Fig. 5. The stringer coupling for the connection of several fuselage sections

Analysis of Stress Concentration

An optimized current skin-stringer-configuration (without slots) is the de-

termination base for the structural analysis. The stress concentration factor
(notch factor) at the stringer end results by the ratio between the absolute
stress concentration and the average longitudinal tensile stress in the skin
(Fig. 6). The stress concentration at the stringer ends is a combined effect
of skin bulging at stiffness discontinues and eccentricity by the middle sur-
faces of skin and stringer.
262 H. Gruss et al.

Fig. 6. FEA-Model for the determination of the stress concentrations at the

stringer ends

Evaluation of Stress Concentration

Stringers with and without slots are compared in order to evaluate the in-
fluence of slots on the stress concentration. In order to reduce the stress
concentration at the stringer ends the current skin-stringer-joining is re-
vised by a socket below the stringer (Fig. 7 left).
The socket leads to a reduction of discontinuity between skin and
stringer. The reference parameters are dSTR = 15, lFH=1.4, lFW=6, lGD=0,
lGW=0. These ones also apply for the structure with slot. Figure 7 right
shows the structural model for a stringer with slot. The slot relevant
reference parameters in mm are dStr =15, lSLOT =15, hSLOT =15, lS =30.
The characteristic lines in Fig. 8 show that the maximum stress concen-
trations in the primary weld seam at stringers with slots are 20% lower
than at stringers without slots. But the stresses in the outer regions of the
secondary weld seams are increased about 11%. This fact is based on the
about ls longer lever of the stringer coupling.
 The Integral Approach 263

Fig. 7. Structural models for the calculation of stress concentration. (left) for the
revised current skin-stringer-design without slot. (right) for a stringer with slot

2,0

P
1,49 without SLOT
1,5
S
1,34
P
ıx / ıVM

1,07
1,0

0,5

0,0
0 1 2 3 4
a [mm]

Fig. 8. The comparison of the stress concentrations for stringers with and without
slot at the start-end-positions in primary part (P) and secondary part (S) (Fig. 4).
The geometrical parameter for the stress concentration at the stringer-edge a is
shown in Fig. 6

The edge distance (dSTR) has not any significant influence on the maxi-
mum stresses in the bearing primary weld seam. But stresses in the secon-
dary weld seam (S) grow up about 65% at an edge distance dSTR (see
Fig. 7) by 15 mm compared to 5 mm. The longer lever is the reason for the
increasing of stresses again.
The secondary part length (lS) and the slot length (lSLOT) have not any in-
fluence on the stress concentration at the primary (p) and secondary (S)
weld seam. A modifying of the slot height (hSLOT) has a significant influ-
ence on the stress concentration in P and S. In case of reduction from 15 to
10 mm the stress notch factor in P decreases roughly 11% but the stress
264 H. Gruss et al.

notch factor grows up 6% in S. Hence the difference between P and S does

increase with a shorter slot height.

Influence of Slot on Weldability

In contrast to the most welding applications, the Integral Approach desires

the large volume shrinkage of aluminum due to solidification and cooling.
Pshennikov [3] calculates and measures the speed of cross displacements
vq as well as the changing of cross displacements during the time of exist-
ing Brittleness temperature range (BTR) ǻuYBTR as a criterion to assess the
risk of hot cracking at the weld seam ends. He developed effective meth-
ods to avoid hot cracks at unsteady stages of seam ends by design and as-
sembly for welding as well as considering the welding sequence for as-
semble the run-out-plates. An important effect to avoid hot cracking at the
seam end is the induction of compressive stresses by joining of the two
run-out-plate-halves. In shipbuilding is used the one-sided submerged arc
welding of large steel panels.
Further the run-out plates are provided with slots and are able to shrink
in consideration of an adapted welding and tacking strategy as well as a
sufficient cooling-off time before the real welding process starts. Both
steel plates try to move diametrically opposed transverse to the welding
line due to the induction of thermal stresses. But the shrinkage movement
by the run-out plates causes a diametrical orientation relating to the diver-
gent movement of steel plates. The material is cooling-off and the dis-
placements within the BTR are reduced. The BTR is fixed by the alloy
composition, but the displacements during the time of existing BTR are
determined by the structure.
The fuselage design does not allow using any run-in or run-out plates.
Following the shrinkage effect must appear on the outer stringer regions.
Experimental investigations and finite element analysis have shown a
lower hot crack resistance than at the ends of current stringers.

The Active Principle

The skin is fixed on a vacuum clamping board. Against this background

the focus lies on the residual stress in the stringer as well as the strains re-
lating to hot cracking. The terms primary and secondary part refer to the
stringer. By the use of a suitable welding regime it is welded the inner
primary part (1) first and afterwards the secondary one (2). These welded
 The Integral Approach 265

sections shrink and pull down the still not-welded outer primary part (3) to
skin. A higher hot cracking resistance in P is achieved (Fig. 9).
The residual tensile stresses (1 and 2) obtain its maximum within the
weld seams and decrease with increasing stringer height. Thereby it is im-
portant that both residual tensile stress areas are joined by a so called
“Bridge”. This bridge ensures that after the first two welding steps the
outer primary zone is pulled down to the stringer whereby there are oc-
cured residual compressive stresses in this one (Fig. 10).
At the third welding step the base metal is heated due to the thermal
stress and it tries to expand. The residual compressive stresses around P
and the surrounding cold areas constrain the volume expansion. Depending
on the residual compressive stresses in the outer primary part it results
lower tensile stresses within the BTR. At this principle it is spoken about a
clamping effect, because the third welding section is enclosed by two ten-
sile stress areas via a bridge.

Fig. 9. Active principle bases on tensile and compressive stresses

266 H. Gruss et al.

1. welding of the 2. welding of the 3. welding of the

inner primary part secondary part outer primary part
100

50

0
ı [MPa]

ıD1,P
-50

1. cooling step
-100 ıD2,P

-150
2. cooling step

-200
0 30 60 90 120
Time [s]

Fig. 10. The resulted stresses (y-direction) at the start-end-position of the primary
part (P) after the first two welding steps. The characteristic compressive stresses
lines refer to the reference model

Determinations

There are used two Nd:YAG-Lasers with a focus diameter by 0.3 mm. The
heat source’s calibration for the simulation bases on a laser power by 2.3
kW, on a welding velocity by 4 m/s. The focus point is fL=0. Shielding gas
is He with a flow rate by 30 l/min (Fig. 11).The used base metal is
AA6056 and the filler metal is AlSi7 (Table 1), 82% dilution rate. The
stringer thickness is 2 mm and the thickness of skin is 3 mm. Before the
welding process the stringer is tacked on the skin.

Table 1. Chemical composition of AA6056 and AlSi7 [wt%]

Alloy Al Si Fe Cu Mn Mg Cr Zn

AA6056 96.98 0.89 0.09 0.60 0.59 0.71 0.012 0.13

AlSi7 92 6.8 0.6 – – 0.6 – –

 The Integral Approach 267

Fig. 11. Comparison of the cross section and the simulated geometry of welding
seam

Influencing Parameters

In Fig. 12 are shown several influencing parameters on the hot cracking

tendency, which are described either by geometrical parameters or by a
vector. Five geometrical parameters arising from the stringer configura-
tion, the tack welding regime as well as the order of welding steps ( lSLOT,
hSLOT, lS, d und e). The slot width and height are defined by the parameter
lSLOT und hSLOT. The parameter lS determine the secondary part length
which is equivalent to the primary weld seam length.
The distance from the inner slot edge to the beginning of the tacking
weld seam is defined by the parameter d. The parameter e arises from the
distance from to inner slot edge and the start/end point of the inner primary
welding seam. In order to the welding direction result six degrees of free-
dom for all three weld seams. The welding directions are determined by
the variables A1 / A2 for the inner primary weld seam, B1 / B2 for the
secondary weld seam and C1 / C2 for the outer primary weld seam.
268 H. Gruss et al.

Fig. 12. Influencing parameters controlling the hot cracking behaviour at the
position P (start-end-position in the primary part)

Evaluation of Displacements within the BTR

The Reference Model

The slot width (lSLOT) and height (hSLOT) are fixed on the minimal allow-
able dimension by 7 mm. The secondary part length is defined by 25 mm.
The parameter e is determined by 10 mm (Table 2). The displacements at
start-end-position of the primary part (P) are the reference value (R) for the
following investigations (Fig. 13).

Table 2. Parameters for the reference model

hSLOT
lSLOT [mm] lS [mm] d [mm] e [mm] welding directions
[mm]
7 7 25 0 10 A2 B2 C2
 The Integral Approach 269

8
2 3
[ yBTR(x) / yBTR(x P, Ref) -1 ] x 100%

1
6

2 R

0 P

-2

-4

-6
0 10 20 30 40 50 60
x [mm]

Fig. 13. Characteristic BTR-displacement lines calculated for the reference

model. In order to compare the displacements with each other the graphical pres-
entation of the BTR-displacements happens in relation to the displacements on
position p at the reference model (100%)

Basically the BTR-displacements of the inner primary weld seam have a

lower magnitude in relation to the displacements both of the secondary and
of the outer primary weld seam. This circumstance is related to the higher
structure stiffness. At the beginning of the first weld step (inner primary
weld seam) the location where the beam is coupled into the base metal is
surrounded by tack welds. In this way the stringer is fixed previous and
behind the laser beam.
The outer edge of the secondary weld seam (2) represents a current
stringer edge basically. The largest hot crack tendency is observed on the
outer edge. The stringer is only fixed from one side. As soon as the
stringer start is melted due to the heat input, the structure resistance is to
low in order to counteract the resulted compressive stresses effectively.
According to this the strains achieve their maximum values.
At the outer primary part position (P) the strains are lower due to the
“bridge” in contrast to the secondary one. The bridge connects the primary
and secondary parts and lead to the larger stiffness. Furthermore both ten-
sile stress areas on the left and right of the slot cause a compressive stress
area. This facts counteract to the compressive stresses that based on the
heat input.
270 H. Gruss et al.

Geometric Parameter: hSLOT , Slot Width

The slot length is enlarged from 7 to 20 mm. The modification has no in-
fluence on the displacements of the inner primary and the secondary weld
seams. At the beginning of the third weld step (P) the maximum strains
grow up about 50% in comparison with the reference model.

Geometric Parameter: hSLOT , Slot Height

The enlarged slot height from 7 to 20 mm has an effect on the strain be-
havior at the secondary and the outer primary weld seams. The rising of
strains about 90% in the outer primary weld seam results by the reduction
of the web width above the slot (Fig. 14). That leads to a decrease of the
mechanical section modulus und therefore to a reduction of residual com-
pressive stresses in the outer primary stringer part, which counter the heat
expansion (Fig. 15).

8
[ yBTR(x) / yBTR(xP, Ref) -1 ] x 100%

2 3 1
6

4
1
2
0,9
0 P

-2

-4
TIS fest

-6
0 10 20 30 40 50 60
x [mm]

Fig. 14. Effect of slot height on displacements at P. The enlarged slot height from
7 to 20 mm causes higher displacements about 90%
 The Integral Approach 271

1st welding step nd

2 welding step 3rd welding step
inner primary weld seam secondary wled seam outer primary weld seam
100

50

0
ı [MPa]

-50

-100 hSLOT = 20 -112

-150 hSLOT = 7 (Ref) -145

-200
0 30 60 90 120
Time [s]

Fig. 15. Effect of slot height on residual stresses at the start-end-position of the
primary part (P). The enlarged slot height leads to a lower stringer stiffness and
therewith to a lower level of residual compressive stresses in P.

Geometrical Parameter: lS , Length of Secondary Part

An increase of the secondary part (ls) from 25 to 40 mm leads to a not sig-

nificant rising of the residual compressive stresses in the outer primary
part. Not the secondary part length is the crucial factor but rather the in-
crease of the residual tensile stresses which are caused by the volume
shrinkage of the secondary part is important for the hot crack behavior at
the beginning of the third weld step (Fig. 16).
272 H. Gruss et al.

1st welding step 2nd welding step 3rd welding step

inner primary weld seam secondary wled seam outer primary weld seam
100

50

0
ı [MPa]

-50

-100
ls = 25mm (Ref) -145
-150
ls = 40mm -154

-200
0 30 60 90 120
Time [s]

Fig. 16. Effect of the parameter lS on the residual stresses at the start-end-position
of the primary part (P). The residual compressive stresses in P increase insuffi-
ciently, in spite of a extended secondary part length

Geometrical Parameter: d, Length of Tack-Free Primary Part

If the outer primary part is not tacked (d=10 mm), then this part is not able
to develop residual compressive stresses in P. The resulting low stress
level in the outer primary part (P) leads to higher displacements about
430%.

Geometrical Parameter: e, Distance Between Inner Slot Edge and

Start/End of the Inner Primary Part

The shortening of the distance between the inner slot edge and the
start/end of the inner primary part from 10 to 2 mm leads to a partial melt-
ing of the inner primary weld seam during the welding of the outer primary
part. The tensile stresses which are assigned to the inner primary weld
seam are released. Due to this the resulted tensile stress in the inner pri-
mary part and the compressive stress in the outer primary part at the P-
position are decreased.

Geometrical Parameters: lSLOT , hSLOT , e

In this case it is investigated the combined effect of several parameters.

The slot width lSLOT and height hSLOT are 40 mm and 20 mm respectively.
 The Integral Approach 273

This configuration leads at the P-position to a reduced stiffness, so the sec-

ondary part is not able to act as a hold-down device.
Due to the enlargement of parameter e – it describes the distance from
the inner slot edge to the start of the inner primary weld seam – from 10 to
20 mm the residual tensile stresses that bases on the inner primary weld
seam has not any influence on the still not-welded outer primary part.
Both residual tensile stress areas interfere with each other only insuffi-
cient. The low residual compressive stresses at the P-position lead to an in-
crease of strains about almost 500% compared to the reference model.

Welding Direction: A1 / A2

The compressive stresses at the P-position increase, if the welding direc-

tion of the inner primary weld seam will be changed. This circumstance
bases on two events at the P-position. First, at the border between the inner
and the outer primary part the molten pool is characterised by a steady
state behaviour and therewith extended than at the opposite welding direc-
tion. Second, the heat concentration is full formed. In both cases a larger
volume is either molten or heated. On cooling the material shrinks and re-
sulting larger residual compressive stresses, which causes compression of
the mushy zone within the BTR that means a hot cracking avoiding.

Welding Direction: B1 / B2

A changed welding direction at the secondary part results in lower strains

on the outer secondary weld seam. In relation to the reference model the
displacements on the outer secondary part decrease about 20%. Neverthe-
less this welding direction should be prevent in order to avoid hot cracking
and cracks in the end crater pool in the extreme stressed stringer ends.
The reverse welding direction in the secondary part has not any bearing
on the hot cracking behavior for the third weld step (outer primary weld
seam).

Welding Direction: C1 / C2

The change of the welding direction at the inner primary part results to
higher strains about 200%. Furthermore the end crater cracks are expected
on the start-end-positions at the primary and secondary parts (P and S).
274 H. Gruss et al.

Summary of Applied Integral Approach

The designed slots in the advanced stringer geometry reduce the stress
concentrations and increase the hot crack resistance at the outer primary
part (P) assumedly. In order to the conditions in the start-end-positions P
and S during the flight phase and the welding, there are following conclu-
sions for the structural design and the welding procedure:

1. The width and height of slots (lSlot, hSlot) should be chosen as small as
possible in order to reduce the stress concentrations and to increase the
superposition between both residual tensile stresses areas, which are
generated by several welding steps of the inner primary and the secon-
dary parts.
2. The main factor, limiting the residual comressive stresses at the start-
end-position of the primary part (P) is not the length of the secondary
part (ls), but rather the heat input.
3. A short edge distance (dSTR) – the distance between the edges of the
socket and the start-end-position of the secondary part (S) – reduces the
stress concentration at S.
4. In order to avoid end crater cracks the ends of weld seams should be not
located at any stringer edges but rather in regions within the stringer
with lower stress levels during the flight phase.

A complex welding order results by the existence of two slots (Fig. 17).
The welding direction has to be changed at the several welding steps. The
stringer must be tacked on the skin before welding procedure can start.
The composition of 4 welding steps and their welding directions result
by the welding analysis for the whole stringer geometry:

− Step 1: The first welding step starts in the middle of the stringer (inner pri-
mary part) and ends at the distance e in front of the slot at the stringer end.
− Step 2: The welding direction of the second step is diametrical to the
first one and starts also in the middle of the inner primary stringer part.
This welding steps stops at the distance e in front of the slot at the
stringer start.
− Step 3: The welding direction changes again. Two weld seams are pro-
duced by this step. The first weld seam spans along the secondary part
and starting on the start –end-position SStart. The second one completes the
outer primary part and starts on the slot at the stringer start PStart.
− Step 4: The welding of the secondary slot needs the same welding step
sequence than step 3, but starting at Send and Pend in opposite direction.
 The Integral Approach 275

Fig. 17. Recommendation for the order of the several welding steps and their
direction

References

1. Herold H (2007) Rissminderung beim Schweißen von Al-Legierungen

mittlerer und höherer Festigkeit (Hot crack reduction of middle and high
strength aluminium alloys), AIF-Projekt, Nr. 13.983B / DVS 01.047,
Otto-von-Guericke-Universität Magdeburg
2. Hilbinger R M (2001) Heißrissbildung beim Schweißen von Aluminium in
Blechrandlage (Hot cracking initiation on welding of aluminium due to
criticals plate-edge-positions), Doctor Thesis, Universität Bayreuth
3. Pshennikov A (2005) Entwicklung von Maßnahmen zur Heißrissvermeidung
beim Einseitenschweißen langer Schweißnähte. (Development of methods for
hot cracking prevention on one-side-welding of larger weld seams), Doctor
Thesis, Otto-von-Guericke-Universität Magdeburg
Weld Parameter and Minor Element Effects
on Solidification Crack Initiation in Aluminium

N. Coniglio, C.E. Cross

Federal Institute for Materials Research and Testing, Berlin, Germany

Abstract

Earlier work has established that a critical amount of 4043 filler is required
to avoid solidification cracking when arc-welding 6060 aluminium, de-
pending upon local strain conditions. For example, when the mushy zone
behind the weld pool experiences a tensile strain from combined thermal
and shrinkage stresses, the possibility exists for crack initiation. For a
greater rate of strain, it has been determined that a greater 4043 dilution
(i.e. higher weld metal silicon content) is required to avoid crack initiation.
Making use of the Controlled Tensile Weldability (CTW) test and local
strain extensometer measurements, a boundary has been established be-
tween crack and no-crack conditions for different local strain rates and
filler dilutions, holding all other welding parameters constant. Using this
established boundary as a line of reference, additional parameters have
now been examined and their influence on cracking has been character-
ized. These parameter influences have included studies of weld travel
speed, weld pool contaminants (Fe, O, and H), and grain refiner additions
(TiAl3 + Boron). Each parameter has been independently varied and its ef-
fect on cracking susceptibility quantified in terms of a critical strain rate
required to initiate cracking for a given 4043 filler dilution.

Introduction

Al-Mg-Si extrusion alloys are widely used in fabricated structures because

of their good corrosion resistance, moderate strength achieved through
heat treatment, and good weldability when using an appropriate filler
metal. When welded autogenously, however, these alloys have been found
to be highly susceptible to solidification cracking [1, 2]. Typical welding
278 N. Coniglio, C.E. Cross

filler metals, such as Alloy 4043 (Al-5%Si) or Alloy 5356 (Al-5%Mg),

shift the weld pool composition to a hybrid alloy mixture that is less crack
sensitive [3]. Because the filler and base metal have different composi-
tions, the amount of filler dilution will influence the composition of the
weld pool, determining its metallurgical and mechanical properties and
cracking susceptibility [2, 4].
Although it is well established that use of an appropriate filler metal im-
proves weldability, the amount of filler dilution required to avoid cracking
has remained an undefined quantity. Curiously, this has never been con-
sidered an important issue to industry, perhaps because standard welding
practice routinely results in sufficient dilution to avoid cracking. This
would particularly be true for gas-metal arc welding, where high filler di-
lution (30–60%) is common. This explanation has been further substanti-
ated by recent work examining the solidification cracking of Alloy 6060,
showing that only a small increase in weld metal silicon content (above
base metal levels) is required to avoid cracking [5].
Research work spanning fifty years has lead to the general belief that
tensile stress and strain at the trailing end of the weld pool are instrumental
in initiating solidification cracks [6-14]. Strain accumulated as a result of
solidification shrinkage and thermal contraction serves to pull weld metal
grains apart in the mushy zone, resulting in failure of grain boundary liquid
films. Prokovorov [7, 8, 9] and Senda [14] have established ductility limits
for cracking; i.e. maximum strain that can be accumulated over the solidi-
fication range (Brittle Temperature Range- BTR). Strain rate was realized
as being important by these researchers, but only in so far as it serves to
determine how much strain can be accumulated during the time of solidifi-
cation. Accordingly, numerous weldability tests have been developed spe-
cifically to measure critical strain rate [5, 15-17]. Meanwhile, develop-
ments arising from solidification cracking models applied to castings, have
suggested that strain rate may play a more direct role in the actual fracture
mechanism by controlling the pressure drop in the interdendritic liquid [13,
18, 19].
Building upon the concept that Si improves weldability and that weld-
ability can be defined by a critical strain rate, studies were initiated to ex-
amine this interrelationship. Specifically, the critical strain required for
cracking an aluminum 6060 weldment was measured for different 4043
filler dilutions [5]. This was done using a Controlled Tensile Weldability
(CTW) test, where a controlled transverse strain was applied during weld-
ing. A crack-no crack boundary was defined for variable conditions show-
ing that higher local strain rates require higher 4043 dilutions to avoid
cracking. The objective of the present study was to characterize the behav-
iour of this crack-no crack boundary when varying metallurgical factors
 Weld Parameter and Minor Element Effects 279

(e.g. grain size, iron content), welding parameters (torch travel speed), and
external contamination (hydrogen and oxygen contents).

Background

Each of the factors examined in this study could conceivably have an ef-
fect on weld solidification cracking susceptibility. Details regarding these
possible interactions are discussed below.

Role of Silicon

Jennings et al. measured the influence of composition on cracking suscep-

tibility of high purity Al-Mg-Si ternary alloys by the means of a ring cast-
ing test [20]. They found a ridge of high cracking susceptibility along the
Al-Mg2Si quasi-binary line (Mg/Si = 1.73, weight ratio). A peak in crack-
ing susceptibility is observed at 0.5 wt.%Si and 0.2 wt.%Mg, as shown in
Fig. 1. Alloy 6060 sits close to this peak, and is shifted to less crack sensi-
tive regions when diluted with Alloy 4043, according to the compositions
given in Table 1.

Table 1. Measured chemical analysis for aluminium alloy 6060 and 4043 mix-
tures (wet chemical analysis): (a) 6060-T4, (b) 6060-T4 + 10%4043, (c) 6060-T4
+ 20%4043, and (d) 4043
Composition (wt.%)
Si Mg Fe Mn Cu Cr Ni Zn Ti Ga Zr
(a) 0.42 0.59 0.19 0.02 0.01 0.004 0.004 0.009 0.02 0.01 0.001
(b) 0.90 0.53 0.19 0.02 0.04 0.00 0.00 0.02 0.04 0.01 0.00
(c) 1.39 0.47 0.19 0.02 0.07 0.00 0.00 0.03 0.06 0.01 0.00
(d) 5.25 0.001 0.21 0.02 0.3* --- --- 0.1* 0.2* --- ---
*handbook values
--- no value specified

Considering its relationship to the quasi-binary line of the Al-Mg-Si ter-

nary system, it is expected that when welding with 4043 filler, and hence
increasing the silicon content of the 6060 weld metal, the microstructure
should consist of increasing amounts of silicon phase in addition to the
quasi-binary eutectic phase Mg2Si [21]. However, due to the presence of
the impurity iron (0.2 wt.%Fe, Table 1), the phase reactions that actually
occur are considerably more complex as indicated in Table 2. Even at low
concentrations, iron has a strong influence on solidification structure, due
280 N. Coniglio, C.E. Cross

to its strong tendency to partition (equilibrium partition ratio: k=0.03) and

form compounds with silicon and aluminum.

Fig. 1. Quasi-binary line superimposed on ring casting data of Jennings et al. [20]
showing solidification cracking susceptibility for Al-Mg-Si ternary alloy system

Based on thermal analysis, microstructure observations and available

literature, small increases in Si content have been found to have a major
effect on the solidification path of 6060 castings [22]. Increased Si from
0.42 wt.% to 1.39 wt.% causes the solidus temperature to drop from 577°C
to 509°C, increases the quantity of interdendritic constituent from 2% to
14%, and results in different phase formation. Binary β (Al5FeSi) and
Mg2Si phases are replaced with ternary β (Al5FeSi), π (FeMg3Si6Al8), and
a low melting quaternary eutectic. Higher Si content has also been ob-
served to result in grain refinement and reduced solidification shrinkage.
Whether this trend persists at higher cooling rates (e.g. as experienced in
welding) is a subject of ongoing work.
 Weld Parameter and Minor Element Effects 281

Table 2. Phase reactions in the Al-Mg-Si-Fe quaternary system for composition

range corresponding to aluminum 6060 base metal at low and high 4043 dilutions
Temperature (°C)
No. Phase Reactions
6060 6060+20%4043
1 L+TiAl3→α(Al) 666 666
2 Lĺα(Al) 653-650 650-647
3 Coherency Point 642-637 642-639
4 L→α(Al)+β 609-597 -----
5 L→α(Al)+Mg2Si 597-589 -----
6 L→α(Al)+Si 587-578 584-577
7 L→α(Al)+Si+β ----- 564-556
8 L+β→α(Al)+Si+ʌ ----- 539-534
9 L→Mg2Si+Si+α(Al)+ʌ ----- 524-509

Role of Iron

Iron, even when present at low impurity levels, plays an important role in
determining solidification microstructure as noted above. Although nor-
mally found at around 0.20 wt.% (0.30 wt.% max.) in most wrought alu-
minium alloys, iron has a 0.80 wt.% max. limit in the 4043 filler alloy
[23]. It is normally considered an undesirable element due to its resulting
in reduced mechanical properties and corrosion resistance, where both
problems are tied largely to the formation of coarse, iron-bearing intermet-
allic phases that are cathodic relative to the aluminium matrix [24].
During solidification of castings, high temperature iron bearing inter-
metallic phases are believed to block interdendritic channels, leading to
poor feeding of shrinkage and porosity formation [25-27]. Singer et al.
measured the cracking susceptibility of high purity Al-Fe-Si alloys by
means of restrained welds [28]. Plotting the mean crack length as a func-
tion of iron and silicon content (Fig. 2), they observed no cracking for the
condition Fe/Si>1. Cracking susceptibility increased when increasing sili-
con content or decreasing iron content.
Lu and Dahle [25] studied the influence of iron content on the solidifica-
tion path and cracking susceptibility of cast Al-7Si-0.4Mg alloys. Cast
samples were quenched at around 590°C to observe the microstructure. In-
creasing the iron content from 0.3 to 0.7 wt.% favours the binary β-phase
(forming at a temperature over 600°C) over the ternary β-phase (forming
at temperatures lower than 580°C). Moreover, an increase in iron content
resulted in an increase in porosity, but decreased cracking susceptibility. It
is believed that, because these coarse binary β needles form at high tem-
perature, they block the liquid feeding and promote porosity [25] and serve
282 N. Coniglio, C.E. Cross

as effective pore nucleation sites [27]. The improvement in cracking resis-

tance may also be due to an increase in high temperature strength with
high iron content [26], or because β-phase needles act like solid bridges
spanning the α-dendrites [25].

Fig. 2. Mean crack length (in inches) on restrained welds of Al-Fe-Si ternary al-
loys as a function of iron and silicon content [28]

Grain Refinement

Grain size directly affects the strain distribution over the grain boundaries
in the weld mushy zone. Refining the grain structure increases the number
of grain boundaries, thereby reducing the strain seen by each boundary.
Reduced cracking susceptibility of aluminium alloy 7108 has been related
to grain refinement, adding grain refiners scandium and titanium-boron,
i.e. TiBor [29]. Using the circular patch test, cracking was avoided when
adding a minimum of 0.25 wt.%Sc or (0.02 wt.%Ti + 0.004 wt.%B),
 Weld Parameter and Minor Element Effects 283

corresponding respectively to a grain size 80 μm and 180 μm. Note that

the grain size in a 7108 weld without TiBor and Scandium was 290 μm.

Travel Speed

Strain and stress fields around the weld pool have been experimentally ob-
served and simulated [10, 11, 16, 30-34]. These fields are dependent on
weld processing parameters as well as restraining conditions. The influ-
ence of torch travel speed on solidification cracking during welding has
been extensively studied, but is not well understood [16, 30-34]. Some
studies have observed an improvement in weldability when increasing the
torch travel speed [30-32]. These measurements were made for the GTA
process and travel speeds between 2.5 and 13 mm/s. Chihoski [30, 31] ex-
perimentally observed compressive and tensile cells around a moving weld
pool, and postulated that no cracking will occur if the mushy zone is under
compression. At high travel speeds, a compressive cell is located at the
mushy zone, thereby avoiding cracking. With decreasing travel speed, this
compressive cell diminishes and is replaced with a tensile cell that is
highly susceptible to cracking. However, some simulations [33] and ex-
perimental observations [34] show an increase in cracking susceptibility
with increasing travel speed. But in these situations, the welding speeds are
high; between 50 and 100 mm/s for laser welding [34] and between 16 and
25 mm/s for GTA welding [33].
Weld travel speed also influences the weld pool shape, which is known
to have a strong effect on weldability [35]. Increasing weld travel speed
modifies the weld pool shape from round to teardrop, which favours stray
grain formation and long continuous centreline grain boundaries. The ori-
entation of these long grain boundaries are perpendicular to the thermal
and mechanical strains, leading to a high cracking susceptibility. This
counters Chihoski´s observations where high torch travel speed should im-
prove weldability. Therefore, both weld pool shape and local strain cells
must be considered when determining solidification cracking susceptibility.

Oxygen and Hydrogen Contamination

Oxygen and hydrogen contaminations are typically unavoidable, because

these elements are always present in the vicinity of the molten weld pool.
For example, humid air is often aspirated into the shielding gas. Concern-
ing oxygen, molten aluminium has a strong tendency to form aluminium
oxides. These oxides have a density close to that of molten aluminium, and
284 N. Coniglio, C.E. Cross

hence tend not to separate if mixed. Entrapped oxide films may influence
the cracking susceptibility, providing sites for crack nucleation due to de-
cohesion [36]. Such oxide films may also impair the fluidity and feeding
ability of the alloy [24], or affect the solidification path, serving as sites for
phase nucleation [37]. Above 750°C, the γ-Al2O3 will be present, which
acts as the preferred nuclei for β-Al5FeSi needles. The α-Al2O3 oxide sup-
presses nucleation of the β-phase, which as noted before is the phase be-
lieved to promote porosity.
Hydrogen contamination can arise from moisture or hydro-carbons (e.g.
machine oil). Hydrogen in molten aluminium often results in porosity, be-
cause of the sharp drop in solubility when going from liquid to solid. Dur-
ing solidification of pure aluminium, the solubility of hydrogen drops from
500 to 50 cubic millimetre per 100 grams of aluminium with 1 atm of hy-
drogen over the coupon [24]. The role of porosity on cracking is not well
understood, and it could conceivably have either positive or negative ef-
fects. Dissolved hydrogen should make it easier to cavitate, which is one
of the possible mechanisms for crack formation [36, 38, 39]. But pores
also serve to feed shrinkage, reducing the interdendritic volume that needs
to be fed, and hence lowering the interdendritic pressure drop [40].

Critical Strain Rate – Dilution Map Concept

The influence of 4043 filler dilution on aluminium 6060-T4 weldability

has been previously studied by means of the CTW test [5]. Compositions
of both alloys are given in Table 1. The principle of the CTW test is to ap-
ply a controlled transverse strain rate during welding. The 4043 filler
speed was varied to obtain 4043 dilutions in the weld from 0 to 16%. An
extensometer placed underneath the weld joint measured local strains.
Strain rates were afterwards calculated and the critical strain rate to initiate
cracking was determined for different 4043 dilutions.
Cracking susceptibility, plotted as a function of local strain rate, 4043
dilution, and the corresponding silicon content, is shown in Fig. 3. Crack-
ing has been found dependent on both local strain rate and 4043 dilution,
as revealed by the crack-no crack boundary (shadow region) in Fig. 3.
Higher local strain rates require higher 4043 dilution to avoid cracking.
The position of the crack-no crack boundary along the strain rate axis
and its slope are significant. Its relative position along the x-axis represents
the influence of thermo-mechanical factors. Alloys having this boundary
occurring at higher strain rates are considered more weldable, since critical
strain rate is representative of weldability. The slope of this boundary re-
veals the influence of thermo-metallurgical factors on weldability. The
 Weld Parameter and Minor Element Effects 285

more vertical this boundary (i.e. higher slope), the less influence filler dilu-
tion has on weldability. Accordingly, an alloy is considered more weldable
if less 4043 is needed to avoid cracking.

Fig. 3. Critical strain rate – dilution map for alloy 6060-T4, characterizing crack-
ing susceptibility in terms of local strain rate and 4043 filler dilution using CTW
test [5]

Experimental Approach

Controlled Tensile Weldability (CTW) Test

Solidification cracking susceptibility was studied by means of the CTW

test (Fig. 4a), which consisted of applying a controlled transverse strain
rate during welding, with strain applied normal to the welding direction
and away from the weld. Experimental equipment consisted of a unidirec-
tional tensile machine (load capacity 500 kN). The tensile transverse cross-
head speed was varied from 0 to 0.083 mm/s by increased steps of 0.017
mm/s. This globally applied strain rate resulted in variations in local strain
rate (i.e. region adjacent to weld pool), which was measured with an exten-
someter.
286 N. Coniglio, C.E. Cross

Weld coupons of 120 mm × 40 mm × 4 mm were cut from extruded bars

of 6060-T6. These coupons were welded between two bigger aluminum
plates (300 m × 150 mm × 8 mm) (Fig. 4b), which were clamped into the
CTW machine.

(a) (b)
Fig. 4. (a) Overview of CTW test machine and (b) dimensions of test coupon

Table 3. Etch for aluminium oxide removal at room temperature

Etch Description
E1 869 mL H2O, 125 mL HNO3 65%, 6.25 mL HF 48% : applied 15 min
E2 100 mL H2O, 66 mL HCl, 66 mL HNO3, 16 mL HF : applied 1 min

Table 4. Base welding parameters for GTAW-CWF process

Current : 110 A
Voltage : 17.8 V
Arc Gap : 2 mm
Electrode diameter : 3.2 mm
Electrode Type : Tungsten + 1%LaO2
Electrode Tip Angle : 30°
Torch Gas : Helium
Gas Flow Rate : 0.33 L/s
Polarity : DCEN
Torch Travel Speed : 4 mm/s
Wire Diameter : 0.8 mm
Wire Feed : 0–41.7 mm/s
 Weld Parameter and Minor Element Effects 287

Welding Parameters

Welding was performed using the gas-tungsten arc, cold-wire feed process
(GTAW-CWF). The base welding parameters are listed in Table 3. The arc
voltage was kept constant using an arc voltage control system, maintaining
a 2 mm arc gap corresponding to a 17.8 V arc voltage. The heat input was
chosen to obtain a bead-on-plate, full penetration weld. Filler wire speed
and tensile cross-head speed were experimental variables. Prior to the
CTW test, the oxide layer on the test coupon was chemically removed
(etch E1, Table 4), followed by degreasing with acetone.

Test Sequence

The CTW test sequence is summarized in Table 5. A pre-load of 15 kN

was applied prior to welding to compensate for thermal expansion of the
weld coupon during welding, maintaining it in tensile conditions even at a
0 mm/s transverse cross-head speed. The arc was initiated by touch contact
between the electrode and the weld coupon. In a 100 mm long weld, the
transverse cross-head speed was applied 30 mm after the start of welding.
Local strain was measured at weld mid-length. At the end of the weld, the
arc was abruptly extinguished, providing information as to the shape of the
weld pool.

Table 5. CTW test sequence

Distance from the
Step Number Related Action
Weld Start (mm)
1 0 Start weld
2 30 Start cross-head travel
3 50 Electrode at the top of the extensometer
4 90 Stop cross-head travel
5 100 Stop weld

Local Strain Measurement

An extensometer was placed underneath the weld at mid-length of the

weld (Fig. 4b) to measure local strain across the weld pool during the
welding process. Local strain rates were calculated from these meas-
ured strains and the critical strain rate for cracking was determined. The
extensometer output was recorded at a 100 Hz frequency during the
whole test using CATMAN 4.5 computer software for data acquisition
and a Spider 8 analog-to-digital converter. In addition, a few select
288 N. Coniglio, C.E. Cross

strain rate measurements were made at the observed crack initiation site to
correlate these measurements with those measured at weld mid-length. The
gage length of the extensometer was 10.5 mm which, by definition, influ-
ences local strain measurements.
The strain and strain rate values were taken at the coherency tempera-
ture, which corresponds to the point during solidification where the secon-
dary arms of two adjacent dendrites are beginning to be in contact. Ac-
cording to some theories [13, 18, 41–43], the coherency temperature
corresponds to the critical temperature where interdendritic liquid feeding
ability (healing) is sharply reduced due to the continuity of the solid phase.
The calculated strain rate value was also taken at the coherency tempera-
ture, at the critical stage where cracking begins. The position of coherency
relative to the torch along the centreline was determined by temperature
and weld pool shape measurements.
The arc was suddenly extinguished at the end of each test, freezing in
the weld pool shape, the boundary of which corresponds to the liquidus of
the weld metal. The position of the liquidus relative to the torch along the
weld centreline was measured. Temperature measurements were then per-
formed to determine the position of coherency point relative to the liquidus
along the weld centreline.

Temperature Measurement

The cooling curve for the weld pool was monitored during solidification
for all experimental conditions to determine the position of the coherency
point relative to the liquidus. A hole, 0.6 mm diameter and 1 mm deep,
was drilled from the bottom of the weld coupon, 35 mm from the start of
the weld. A 0.5 mm outer diameter, sheathed, electrically grounded,
nickel/chrome-nickel thermocouple was pre-placed inside the hole before
welding. The thermocouple output was recorded at a 200 Hz frequency
during solidification using CATMAN 4.5 computer software for data acqui-
sition and a Spider 8 analog-to-digital converter.

Material

Extrusions

Extruded aluminum 6060-T6 bar was used as the base material, and was
compared against 6060-T4 bar material studied previously [5]. Both mate-
rials came from different heats and had slightly different compositions.
 Weld Parameter and Minor Element Effects 289

Composition of the 4043 filler and both 6060 heats were measured with
wet chemical analysis and spectrometry (Table 6). 6060-T4 and 6060-T6
extrusions had a different hardness (respectively 40 and 83 HV0.5) and
varied, in particular, in alloying elements Mn, Cu, and Zn. The size of the
weld coupons were 120 mm × 40 mm × 4 mm (Fig. 4b).

Table 6. Measured wet chemical analysis and emission spectrometry for alumi-
num alloys 6060, 4043, and controlled mixtures of 6060 with master alloys Al-
10Fe and Al-5Ti-B (wt.%) : (a) 6060-T4, (b) 6060-T6, (c) 4043, (d) insert
6060+Tibor (6060+15%(Al-5Ti)), (e) insert 6060+Fe (6060+15%(Al-10Fe)),
(f) weld pool 6060+insert (6060+Fe), and (g) weld pool 6060+ insert (6060+Tibor)
Composition (wt.%)
Si Mg Fe Mn Cu Cr Ni Zn Ti Ga Zr
(a) 0.42 0.59 0.19 0.02 0.01 0.004 0.004 0.009 0.02 0.01 0.001
(b) 0.51 0.51 0.21 0.04 0.03 0.003 0.003 0.04 0.02 0.01 0.001
(c) 5.25 0.001 0.21 0.02 0.3* --- --- 0.1* 0.2* --- ---
(d) 0.43 0.43 0.18 0.03 0.02 0.003 0.003 0.04 0.77 0.01 0.001
(e) 0.43 0.43 1.68 0.03 0.02 0.003 0.003 0.04 0.02 0.01 0.001
(f) 0.50 0.50 0.43 0.04 0.03 0.003 0.003 0.04 0.02 0.01 0.001
(g) 0.50 0.50 0.21 0.04 0.03 0.003 0.003 0.04 0.14 0.01 0.001
*handbook
--- value not specified

Inserts

The objective was to vary weld metal composition using inserts having a
controlled composition (Table 6). Inserts were machined from cast ingots
made from controlled mixtures of 6060-T6 with master alloys (e.g. Al-5Ti-
B or Al-10Fe). These inserts were then pre-welded into 6060-T6 CTW
coupons in preparation for CTW wedability testing.
Controlled mixtures of 6060-T6, 6060+15%(Al-5Ti-B), and
6060+15%(Al-10Fe) were cast. The composition of the resulting ingots,
when using master alloys, differs from 6060 respectively for titanium (0.8
wt.% instead of 0.02 wt.%) and iron (1.7 wt.% instead of 0.2 wt.%). The
casting mold is shown in Fig. 5, with inside dimensions 140 × 10 × 30 mm,
and mold thickness 10 mm. All mold materials with direct exposure to
molten aluminum were pre-coated with boron-nitride spray. Each cast heat
weighed approximately 130 grams and was melted in a graphite crucible
placed inside an electric furnace held at 800°C. Oxide dross was skimmed
just prior to casting, and the melt was rigorously stirred in order to ensure
thorough mixing with the master alloy. In the case of Al-5Ti-B, 6060-T6
was thoroughly melted before adding the TiBor master alloy, in order to
290 N. Coniglio, C.E. Cross

reduce time exposure of TiAl3 particles in the melt, and thus minimize
particle dissolution and promote efficient grain refinement. Five inserts
(2 mm × 2 mm × 140 mm) were machined from each cast ingot. For pur-
poses of providing an experimental control, inserts of cast 6060 material
were also prepared, left untreated with any master alloy.

(a) (b)
Fig. 5. Casting mold (a) photograph and (b) schematic, used to obtain ingots for
weld insert production

Fig. 6. Insert placed into groove of 6060-T6 coupon before pre-weld

Square grooves (2 × 2 mm) were machined along the centreline of

6060-T6 CTW test coupons (i.e. extrusions) to receive the inserts (Fig. 6).
Prior to assembly, the oxide layer was chemically removed by etching both
weld coupon and insert (etch E1, Table 4), followed by degreasing with
acetone. The insert was tapped into the groove and pre-welded using the
welding parameters listed in Table 7. Variable polarity current was used
here to help to remove oxides at the weld surface. The pre-weld consisted
of a bead-on plate, partial penetration weld (6 mm wide and 3 mm thick),
which completely melted the insert.
 Weld Parameter and Minor Element Effects 291

Shielding Gas

In the case of controlled oxygen and hydrogen contamination, helium flow

rate was maintained constant (0.33 L/s) and mixed with gases from pre-
mixed bottles of either Ar+2%H2 or Ar+1%O2. Because of the effect of
gas additions in changing the weld heat input, current was varied in order
to maintain a constant weld pool size. Ar+2%H2 flow rate was fixed at
0.27 L/s to obtain a sufficient amount of dissolved hydrogen in aluminum
to promote limited interdendritic pore formation. Regarding oxygen con-
tamination, Ar+1%O2 flow rate was fixed at 0.03 L/s, corresponding to the
maximum gas flow rate possible while maintaining arc stability. At a 0.08
L/s Ar+1%O2 flow rate, the arc became highly erratic, likely due to the
large quantity of aluminium oxides formed on the weld pool.

Table 7. Welding parameters for pre-welding inserts

Positive Current /Duration : +80 A /20%
Negative Current /Duration : –220 A /80%
Frequency : 50 Hz
Arc Gap : 1.5 mm
Electrode diameter : 3.2 mm
Electrode Type : Tungsten + 2% ThO2
Electrode Tip Angle : 30°
Torch Gas : Argon
Gas Flow Rate : 0.3 L/s
Polarity : Variable polarity
Torch Travel Speed : 4 mm/s

Critical Strain Rate - Dilution Maps

The crack- no crack boundary in a critical strain rate – dilution map was
determined for each experimental condition by measuring cracking suscep-
tibility at two levels of dilution: 0% and 15–20%. 4043 filler speed was
chosen to reach a dilution between 15% and 20%. Dilution is an important
concept in determining weld pool composition, approximated using Eq. (1):
B+C
filler dilution = × 100 (%) , (1)
A+B+C
where A is the melted area of 6060 base metal, and B+C is the difference
between the total area of the weld metal and A (Fig. 7). Calculated dilu-
tions are summarized in the Appendix.
For both low and high dilution levels, the transverse cross-head speed
was incremented in steps of 0.017 mm/s, noting the value where cracking
292 N. Coniglio, C.E. Cross

occurred. Because of the fixed incremental step of transverse cross-head

speed (0.017 mm/s), the exact location of the crack-no crack boundary lies
somewhere within this fixed step. This corresponds to the difference be-
tween the highest measured strain rate without cracking and the lowest one
to initiate a crack. Hence, the accuracy of testing is limited by the magni-
tude of this step.

Fig. 7. Illustration for filler dilution calculation from weld metal cross-section

Results and Discussion

Characterization of Weld Pool Shape

Strain measurements were made at the point where the mushy zone (co-
herency point) passes between the extensometer. The weld pool tempera-
ture was recorded during solidification to know where the mushy zone and
coherency was located relative to the torch position. Thermal analyses per-
formed previously measured the liquidus at 660°C and coherency at 624°C
for the 6060 alloy [5]. For all experimental conditions, it has been found
that the weld temperature drops during solidification from 660°C to 624°C
in approximately 0.2 s. Thus, knowing the travel speed, the distance from
the fusion line to the coherency point can be calculated.
The weld pool shape and resulting grain structure was observed and char-
acterized from the top surface. The application of the etch E2 (Table 4) on
the top surface of the weld revealed the grain structure and weld pool shape,
obtained by extinguishing suddenly the arc. Figure 8 illustrates the measured
distances to characterize the weld pool shape. For all the welding conditions,
the dimensional characteristics of the weld pool were: top width 8 mm, and
distance behind electrode 6 mm, with a deviation of ± 1 mm. The measure-
ments of the weld pool shape are summarized in the Appendix.
 Weld Parameter and Minor Element Effects 293

(a) (b)
Fig. 8. Weld pool measurements : (a) top width (A) and (b) distance behind elec-
trode (B) and in front of electrode (C)

Metallographic transverse cross-sections were cut from the mid-length

of the weld. These sections were ground and polished to 1 μm and then
chemically polished using a slightly basic solution of colloidal silicon di-
oxide. The cross-sections were observed with optical microscopy. Weld
bead dimensions were measured in all experimental conditions (see Ap-
pendix) and remained almost constant. An example of weld cross-sections
made on 6060-T6 coupons with 0% and 17% 4043 filler dilution coupons
are shown in Fig. 9. Cross-sections for all experimental conditions re-
vealed similar weld pool shape. More detailed weld metal dimensions for
0% 4043 dilution are: 8 mm overbead width, 6 mm underbead width, 4
mm centreline thickness, and 24–27 mm² cross sectional area. For 17%
4043 dilution these dimensions are: 8 mm overbead width, 6 mm under-
bead width, 5 mm centreline thickness, and 28–32 mm² cross sectional
area.

(a) (b)
Fig. 9. Weld cross sections of 6060-T6 coupons made with (a) 0% and (b) 17%
4043 dilution
294 N. Coniglio, C.E. Cross

Local Strain Measurement

An example of the strain measured in a CTW test is plotted in Fig. 10a,

shown as a function of time, with the CTW test sequence (Table 5) super-
posed. Figure. 10b shows the calculated strain rate from the point (3) to the
point (4), i.e. during cooling. With the objective being the study of strain
rate conditions during weld solidification, time has been redefined in
Fig. 10b, with 0 seconds corresponding to step 3 (Table 5).

(a) (b)
Fig. 10. (a) CTW test sequence superposed on the measured strain and (b) calcu-
lated strain rate, for the test conditions 6060-T4 with 9% 4043 dilution and 0.067
mm/s cross-head speed

When welds were made using base welding parameters on 6060-T6

coupons (Table 3), it was observed that cracking always initiated 25
mm after the weld start, which corresponds to the position of the weld
pool mushy zone at the instant of transverse strain application (step 2 in
Table 5).

Alloy Heat / Temper Effect

Aluminium 6060 extrusions from two different heats and tempers were
evaluated with the CTW test using the same base welding parameters
(Table 3). A critical strain rate – dilution map comparing 6060-T4 and
6060-T6 is shown in Fig. 11 as discussed in the background. 6060-T6 has
been studied at 0 and 17% 4043 dilution, whereas data for 6060-T4 exists
from previous work (Fig. 3) [5] where CTW tests were run over a broad
range of dilutions (0%, 5%, 9%, 11%, 14%, and 16% 4043). A large dif-
ference in critical dilution is observed at high strain rates, where higher
4043 dilution is required to avoid cracking in 6060-T6. For example, at a
local strain rate of 0%/s, 7% 4043 dilution is required to avoid cracking in
6060-T4, while 17% dilution is required for 6060-T6. It is not known at
this point whether this difference in weldability is due to differences in
 Weld Parameter and Minor Element Effects 295

hardness (40 versus 83 HV0.5) or composition (Table 6). This demon-

strates, however, how material with the same alloy designation can exhibit
different weldability depending on specific conditions.

Fig. 11. Critical strain rate-dilution map: (a) 6060-T4, previous data from [5] and
(b) 6060-T6, with data points shown for 0 and 17% 4043 dilution

Travel Speed Effect

CTW tests were made on aluminium 6060-T6 using the base parameters
(Table 3), but varying the current and weld travel speed. In addition, the
welding current was varied so as to maintain a constant weld pool size.
Weld travel speed was run at 2, 4, and 6 mm/s, and the current was varied
respectively from 80A to 145A. Data for 4 mm/s was obtained from a pre-
vious test (Fig. 11). Dilutions studied were 0%, and either 18% or 16%
4043 respectively for 2 and 6 mm/s weld travel speeds. The crack – no
crack boundaries for these conditions are compared against the boundary
found using the base welding parameters (Table 3), i.e. 4 mm/s weld travel
speed and 110A, as shown in Figs. 12 and 13. Increasing weld travel speed
should improve weldability following the relationship of Chihosky [30,
31], and this appears to be the case, but only at high strain rates. Specifi-
cally, at 18% 4043 dilution, solidification cracking occurs for a +0.12%/s
local strain rate at 2 and 4 mm/s weld travel speed, while no crack occurs
at +0.25%/s strain rate for a 6 mm/s travel speed. On the other hand, 6060-
T6 is less weldable at 0% 4043 dilution with increasing weld travel speed.
296 N. Coniglio, C.E. Cross

This is likely due to a change in weld metal grain structure, where stray
centreline grains were observed to occur only at 4 and 6 mm/s weld travel
speed.

Fig. 12. Critical strain rate – dilution map comparing crack-no crack boundaries at
weld travel speeds of (a) 4 mm/s and (b) 2 mm/s when welding 6060-T6. Data
points are shown for 2 mm/s

Fig. 13. Critical strain rate – dilution map comparing crack no crack boundaries at
weld travel speeds of (a) 4 mm/s and (b) 6 mm/s when welding 6060-T6. Data
points are shown for 6 mm/s
 Weld Parameter and Minor Element Effects 297

Insert Control Test

Weld coupons with 6060 inserts containing no extra alloying elements

were evaluated with CTW testing thus providing a control to determine the
influence of the insert and corresponding weld coupon preparation on
weldability. The base welding parameters were used (Table 3), but the cur-
rent was decreased to 95A to maintain a constant weld pool size. Cracking
susceptibility was measured for 0 and 17% 4043 dilution. Figure 14 com-
pares the crack-no crack boundary boundaries for 6060-T6 with base weld-
ing parameters and 6060-T6 with 6060 inserts. A slight difference is ob-
served which may come from the methodology to determine these
boundaries (i.e. fixed increasing step of the cross-head speed). The use of
inserts appears to have no significant influence on the crack-no crack
boundary. Thus, weldability differences due to insert alloy additions (fol-
lowing sections) will be assumed to represent the effect of the alloy addi-
tion, and not the insert itself.

Fig. 14. Critical strain rate – dilution map comparing crack-no crack boundary for
(a) 6060-T6 and (b) 6060-T6 with 6060 insert. Data points are shown for 6060 in-
serts
298 N. Coniglio, C.E. Cross

Iron Impurity Effect

Inserts at high iron content were machined from a 6060+1.7%Fe ingot

(Table 6). The influence of iron content on cracking susceptibility was first
observed during the casting of ingots. While 6060 ingots cracked at the
base of the pouring spout, there was no cracking in 6060+1.7%Fe ingots.
This supports observations of Lu and Dahle [25], where high iron content
prevented cracking in cast Al-Mg-Si alloys. The welding parameters were
the same as those used for 6060 inserts. The 4043 dilutions studied in-
cluded 0% and 20%. The weld pool composition resulting from using the
high iron insert had a corresponding high iron content (i.e. increase from
0.2 to 0.4 wt.%Fe) as given in Table 6. The crack-no crack boundaries are
compared in Fig. 15. Further investigations are needed to better define
these boundaries, but it appears there is little effect of iron.

Fig. 15. Critical strain rate – dilution map comparing crack-no crack boundary for
6060-T6 with (a) 6060 insert and (b) 6060+1.7%Fe insert. Data points are shown
for 6060+1.7%Fe insert

Grain Refinement Effect

CTW tests were performed with inserts high in TiBor content machined
from a 6060+0.8%Ti+0.16%B ingot (Table 6). A few scattered macro-
pores were observed in the weld cross-sections. The influence of TiBor on
cracking susceptibility was first observed in cast ingots, where 6060 ingots
cracked at the base of the pouring spout, and 6060+0.8%Ti+0.16%B ingots
 Weld Parameter and Minor Element Effects 299

did not crack. The welding parameters were the same as those used for
6060 inserts. The 4043 dilutions studied were 0% and 15%. The weld pool
composition with the Ti rich insert differed by its higher titanium content
(from 0.02 to 0.14 wt.%Ti) as given in Table 6. Crack-no crack boundaries
are compared for both insert compositions in Fig. 16. The grain refiner ad-
dition significantly improved weldability, particularly at low 4043 dilu-
tions. This is not surprising considering the potent effect of refinement re-
ported for aluminium weldability improvement [29]. Moreover, it seems
that TiBor improves weldability with more efficiency than 4043, the
boundary being completely moved to higher strain rates for all dilutions.
Finally, the boundary is almost vertical; revealing that 4043 dilution has
little influence on the weldability of grain refined 6060 welds. In essence,
the grain refinement resulted from TiBor exceeds any metallurgical effects
resulting from 4043 filler dilution.
4043 filler is known to reduce the mechanical properties of 6xxx base
metal, and so the improved weldability comes at a price. This suggests that
it may prove feasible to develop a filler that is high in grain refiner, while
low in Si. It is interesting to note that Ti content in 4043 filler tends to be
very high (0.2 wt.%Ti) compared to the initial Ti content in 6060-T6 (0.02
wt.%Ti). Increasing 4043 filler dilution increases the Ti content in the
weld metal, promoting grain refinement [5]. Thus, improved grain refine-
ment should be achieved with increasing 4043 dilution, which may explain
why the crack-no crack boundaries tend to be closer at high 4043 dilutions
(Fig. 16). Future studies are required to determine the improvement in
weldability when increasing the only Si content of the weld without in-
creasing the Ti content.

Oxygen Contamination Effect

CTW tests were performed using an oxygen containing shielding gas (flow
rate: 0.33 L/s He + 0.03 L/s Ar + 1%O2). This required a decrease in the
welding current to 100A to keep a constant weld pool size. 4043 dilutions
of 0% and 18% were studied. High quantities of oxides were observed at
the weld surface. Crack-no crack boundaries are compared between nor-
mal and oxygen-containing gas in Fig. 17, where use of oxygen is found to
slightly improve weldability. At 18% 4043 dilution and a local strain rate
of +0.09%/s, a crack should occur when welding with helium, but does not
initiate with oxygen added to the helium. This trend goes counter to the be-
lief that oxygen forms oxides films that can nucleate cracks [24]. However,
300 N. Coniglio, C.E. Cross

it may just be that a heavy oxide is formed at the pool surface and does not
get mixed into the weld pool.

Fig. 16. Critical strain rate – dilution map comparing crack-no crack boundary for
6060-T6 with (a) 6060 insert and (b) 6060+0.8%Ti insert. Data points are shown
for 6060+0.8%Ti insert

Fig. 17. Critical strain rate - dilution map comparing crack-no crack boundary for
6060-T6 welded with (a) 0.33 L/s He and (b) 0.33 L/s He + 0.03 L/s (Ar+1%O2).
Data points are shown for He+(Ar+1%O2) shielding gas
 Weld Parameter and Minor Element Effects 301

Hydrogen Contamination Effect

CTW tests were performed using a hydrogen containing shielding gas

(flow rate: 0.33 L/s He + 0.27 L/s Ar + 2%H2). This required a decrease in
the welding current to 105A to keep a constant weld pool size. Dilutions
were studied at 0 and 18% 4043. Limited interdendritic pores were ob-
served in the weld cross section (Fig. 18). Since no oxides were observed
on the weld surface, it is assumed that hydrogen served as an effective re-
ducing element. Crack-no crack boundaries are compared between welds
made with helium and helium-argon-hydrogen in Fig. 19. Hydrogen is
found to improve weldability, especially at high 4043 dilution. In fact, at
18% dilution and +0.15%/s local strain rate, 6060-T6 cracked when
welded with helium, but did not crack when welded with helium-argon-
hydrogen. This suggests that hydrogen pores may be feeding shrinkage and
thus reducing the pressure drop. This leads to an interesting choice be-
tween having cracking or porosity.

Sensitivity of CTW Weldability Test

One of the aims of the present work was to use the CTW test to evaluate
how small variations in alloy composition may affect weldability. In pur-
suing this aim, the sensitivity and limitations of this new weldability test
have also been established. Some of the inherent problems associated with
this test are discussed below.

Position of Extensometer

In executing the CTW test sequence (Table 5), the extensometer used to
measure local weld strain was positioned at weld mid-length. However,
ideally the best place to measure strain is at the crack initiation site, which
was found to always occur at the point where the mushy zone sits when
strain is first applied, 25 mm after the weld start. In order to determine
how strain at these two locations may differ (i.e. 25 mm versus 50 mm),
four tests were performed with the extensometer positioned 25 mm from
the weld. These values were then compared against the standard mid-
length measurements, as shown in Table 8. These values show good
agreement (less than 0.05%/s difference), except for one condition at high
cross-head speed where there was a large difference (0.29%/s). This sug-
gests that the standard sequence should be modified for future tests, to re-
locate the extensometer to the 25 mm position.
302 N. Coniglio, C.E. Cross

Fig. 18. Un-etched weld cross sections of 6060-T6 coupon made with He+Ar+H2
and 0% 4043 filler dilution revealing interdendritic pores (arrows)

Fig. 19. Critical strain rate – dilution map comparing crack-no crack boundary for
6060-T6 welded with (a) 0.33 L/s He and (b) 0.33 L/s He + 0.27 L/s (Ar+2%H2).
Data points are shown for He+( Ar+2%H2) shielding gas
 Weld Parameter and Minor Element Effects 303

Table 8. Comparison between strain rates based upon extensometer measurements

made 25 mm from weld start and at weld mid-length

Cross- Measured strain rate (%/s)

Cracking Dilution
head speed 25 mm from Weld mid-
susceptibility (%)
(mm/s) weld start length
No crack 0 0 –0.18 –0.20
Crack 0 0.033 –0.09 –0.06
No crack 17 0.033 –0.05 +0.00
Crack 17 0.083 +0.26 +0.55

Negative Strain Rate

As can be seen in any of the strain rate-dilution maps, cracks that occur at
low 4043 dilutions involves a negative critical strain rate (i.e. inward
movement of material to feed shrinkage). At first hand this appears counter
intuitive, but may actually reflect upon the material’s very poor weldabil-
ity. Even with the inward movement of base material, the solidification
shrinkage is apparently not sufficiently compensated to avoid cracking.
Data from previous work has shown similar behavior, where cracking may
occur even with inward movement of base material [5].

Crack Boundary

The nature of the crack-no crack boundary for the 6060/4043 alloy system
is such that variations in critical dilution of 0–20% corresponds to a maxi-
mum difference in critical strain rate, expressed in terms of cross-head
speed, of roughly 0.1 mm/s. In terms of test sensitivity, it would be prefer-
able that this number be much larger. As it is, this means that test incre-
ments approaching 0.01 mm/s are needed to accurately identify the loca-
tion of a boundary (i.e. one tenth of the full range). In this work,
increments of 0.017 mm/s were used, which only roughly determined the
boundary for each experimental condition. Thus, the use of finer incre-
ments should be considered for future tests on this alloy. Other alloy sys-
tems will likely exhibit their own unique sensitivity to strain rate.

Interacting Factors

As is usual for welding experiments, it is difficult to vary one parameter

systematically and not have numerous other factors change indirectly. For
CTW testing, the use of 4043 filler changes not only the silicon content of
304 N. Coniglio, C.E. Cross

the weld metal, but increases the size and shape of the weld bead. Thus, it
becomes difficult to separate the metallurgical effects from the mechani-
cal. Likewise, variations in travel speed or changes in shielding gas will al-
ter the heat input and, although corrected with adjustments to current, there
may be subtle changes to pool shape that can affect grain structure. Thus, it
becomes difficult to know if the effect of travel speed on weldability is due
to a change in local strain rate or grain structure. The application of model-
ling may help to answer these questions in the future.

Conclusion

The newly developed CTW test, evaluated in terms of critical strain rate –
dilution maps, has been successfully used to quantify the weldability of the
aluminium 6060/4043 alloy system. In particular, the effect of minor alloy
additions and travel speed on solidification cracking susceptibility has
been studied by observing how these changes shift the crack-no crack
boundary in a strain rate-dilution map. Small additions of oxygen or iron
were found to have little or no effect on weldability. Improvements in
weldability were experienced with the addition of hydrogen or a grain re-
finer, or with the use of faster travel speeds.

Acknowledgement

The authors are grateful to BAM for internal funding of this project, and
specifically wish to thank Th. Böllinghaus and Th. Kannengiesser for their
insightful support of this research effort. In addition, technical support at
BAM from A. Hannemann, P. Friedersdorf, K. Scheideck, R. Breu, M.
Lammers, and M.Richter was greatly appreciated. The authors are also
grateful to Metallurg Aluminum for donation of master alloys.

References

1. Dudas JH, Collins FR (1966) Preventing Weld Cracks in High-Strength Alu-

minum Alloys. Welding Journal 45:241s–249s
2. Metzger GE (1967) Some Mechanical Properties of Welds in 6061 Aluminum
Alloy Sheet. Welding Journal 46(10):457s–469s
 Weld Parameter and Minor Element Effects 305

3. Mousavi MG, Cross CE, Grong Ø, Hval M (1997) Controlling Weld Metal
Dilution for Optimized Weld Performance in Aluminum. Sci. Tech. Weld.
Joining 2:275–278
4. Liptak JA, Baysinger FR (1968) Welding Dissimilar Aluminum Alloys.
Welding Journal 47:173s–180s
5. Coniglio N, Cross CE, Michael Th, Lammers M (2008) Defining a Critical
Weld Dilution to Avoid Solidification Cracking in Aluminum. Welding Jour-
nal (in review)
6. Pellini WS (1952) Strain Theory for Hot Tearing. Foundry 80:125–199
7. Prokhorov NN (1956) The Problem of the Strength of Metals While Solidify-
ing During Welding. Svarochnoe Proizvodstvo 6:5–11
8. Prokhorov NN, Bochai MP (1958) Mechanical Properties of Aluminum Al-
loys in the Crystallization Temperature Range during Welding. Svarochnoe
Proizvodstvo 2:1–6
9. Prokhorov NN, Gavrilyuk MN (1971) Strain Behaviour of Metals during So-
lidification after Welding. Welding Production 13(6):8–13
10. Zacharia T (1994) Dynamic Stresses in Weld Metal Hot Cracking. Welding
Journal 73(7): 164s–172s
11. Feng Z (1994) A Computational Analysis of Thermal and Mechanical Conditions
for Weld Metal Solidification Cracking. Welding in the World 33:340–347
12. Kannengiesser T, McInerney T, Florian W, Böllinghaus T, Cross CE (2002)
The influence of local weld deformation on hot cracking susceptibility. In:
Mathematical modelling of weld phenomena 6, Maney, pp 803–817
13. Feurer U (1977) Influence of Alloy Composition and Solidification Condi-
tions on Dendrite Arm Spacing, Feeding, and Hot Tear Properties of Alumi-
num Alloys. In: Proc. international symposium engineering alloys, Delft,
131–145
14. Senda T, Matsuda F, Takano G, Watanabe K, Kobayashi T, Matsuzaka T
(1971) Experimental Investigations on Solidification Crack Susceptibility for
Weld Metals with Trans–Varestraint Test. Trans. of JWRI 2(2):141–162
15. Arata Y, Matsuda F, Nakata K, Shinozaki K (1977) Solidification Crack Sus-
ceptibility of Aluminum Alloy Weld Metals (Report II) – Effect of Straining
Rate on Cracking Threshold in Weld Metal during Solidification. Trans. of
JWRI 6(1):91–104
16. 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. of JWRI 8:85–95
17. Tamura H, Kato N, Ochiai S, Katagiri Y (1977) Cracking Study of Aluminum
Alloys by the Variable Tensile Strain Hot Cracking Test. Trans. JWS 8:16–22
18. Rappaz M, Drezet JM, Gremaud M (1999) A New Hot–Tearing Criterion.
Met. Mat. Trans. 30A:449–455
19. Rindler W, Kozeschnik E, Enzinger N, Buchmayr B (2002) A Modified Hot
Tearing Criterion for Steels. In: H Cerjak (ed) Mathematical Modelling of
Weld Phenomena 6, Woodhead Publishing Limited, pp 819–835
306 N. Coniglio, C.E. Cross

20. Jennings PH, Singer ARE, Pumphrey WI (1948) Hot-Shortness of some High-
Purity Alloys in the Systems Al-Cu-Si and Al-Mg-Si. J. Inst. Metals 74:227–248
21. Mondolfo LF (1976) Aluminum Alloys–Structure & Properties, Butterworths,
London
22. Coniglio N, Cross CE (2006) Characterization of Solidification Path for Alu-
minum 6060 Weld Metal with Variable 4043 Filler Dilution. IIW-1755–06,
Welding in the World 50(11/12):14–23
23. The Aluminum Association (2006) International Alloy Designations and
Chemical Composition Limits for Wrought Aluminum and Wrought Alumi-
num Alloys
24. Van Horn KR (1967) Aluminum Properties, Physical Metallurgy, and Phase
Diagrams, vol.1. American Society of Metals, Ohio
25. Lu L, Dahle AK (2005) Iron-rich Intermetallic Phases and their Role in Cast-
ing Defect Formation in Hypoeutectic Al-Si Alloys. Met. Mat. Trans.
36A:819–835
26. Wang L, Makhlouf M, Apelian D (1995) Aluminum Die Casting Alloys : Al-
loy Composition, Microstructure, and Properties-Performance Relationships.
International Materials Reviews 40(6):221–238
27. Mbuya TO, Odera BO, Ng’ang’a SP (2003) Influence of Iron on Castability
and Properties of Aluminum Silicon Alloys: Literature Review. International
Journal of Cast Metals Research 16(5):451–465
28. Singer ARE, Jennings PH (1947) Hot-Shortness of some Aluminum-Iron-
Silicon Alloys of High Purity. J. Inst. of Metals 73:273–284
29. Mousavi MG, Cross CE, Grong Ø (1999) Effect of Scandium and Titanium-
Boron on Grain Refinement and Hot Cracking of Aluminum Alloy 7108. Sci-
ence and Technology of Welding and Joining 4(6):381–388
30. Chihoski RA (1972) The Character of Stress Field Around a Weld Arc Mov-
ing on Aluminum Sheet. Welding Journal 51(1):9s–18s
31. Chihoski RA (1979) Expansion and Stress Around Aluminum Weld Puddles.
Welding Journal 58(9):263s–276s
32. Hunziker O, Dye D, Roberts SM, Reed RC (2005) A Coupled Approach for
the Prediction of Solidification Cracking during the Welding of Superalloys.
In: Böllinghaus Th, Herold H (eds) Hot Cracking Phenomena in Welds,
Springer-Verlag Berlin Heidelberg, pp 299–319
33. Shibahara M, Serizawa H, Murakawa H (2005) Finite Element Method for
Hot Cracking Analysis using Temperature Dependent Interface Element. In:
Böllinghaus Th, Herold H (eds) Hot Cracking Phenomena in Welds, Springer-
Verlag Berlin Heidelberg, pp 253–267
34. Cicala E, Duffet G, Andrzejewski H, Grevey D, Ignat S (2005) Hot Cracking
in Al-Mg-Si alloy laser welding – operating parameters and their effects. Ma-
terials Science and Engineering 395A:1–9
35. Savage WF, Aronson AH (1966) Preferred orientation in the weld fusion
zone. Welding Journal 45:85s–89s
36. Campbell J (1991) Castings, Butterworth-Heinemann, Oxford, Boston
37. Westengen H (1982) Formation of Intermetallic Compounds During DC Cast-
ing of a Commercial Purity Al-Fe-Si Alloy. Z. Metallkde. 73:360–368
 Weld Parameter and Minor Element Effects 307

38. Cross CE, Olson DL, Edwards GR (1993) The Role of Porosity in Initiating
Weld Metal Hot Cracks. In: Zacharia Th (ed) Int. Conf. Proc. On Modeling
and Control of Joining Processes, pp 549–557
39. Talbot DEJ (2004) The Effects of Hydrogen in Aluminum and Its Alloys,
Maney Publishing, London
40. M´Hamdi M, Mo A (2005) On Modelling the Interplay between Microporos-
ity Formation and Hot Tearing in Aluminium Direct-Chill Casting. Materials
Science and Engineering A:105–108
41. Arata Y, Matsuda F, Nakata K, Shinozaki K (1977) Solidification Crack Sus-
ceptibility of Aluminum Alloy Weld Metals (Report III) - Effect of Straining
Rate on Crack Length in Weld Metal. Trans. of JWRI 6(2):47–52
42. Borland JC (1960) Generalized Theory of Super Solidus Cracking in Welds
(and Castings). British Welding Journal: 508–512
43. Pumphrey WI, Jennings PH (1948) A Consideration of the Nature of Brittle-
ness and Temperature Above the Solidus in Castings and Welds in Aluminum
Alloys. J. Inst. Metals 75:235–256

Appendix

The following tables reveal the different measurements realized for all the
studied experimental conditions. They explicitly relates the calculated dilu-
tion to the 4043 filler feeding speed, the weld pool shape measurements
according to Fig. 8, and dimensional cross sectional characteristics of the
weld pools. Each table correspond to one experimental condition at low
and high 4043 filler dilutions.

Table A.1. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T4 [5]. Distances (A), (B), (C) are according to Fig. 8
Distance Distance in front
Filler Wire 4043
Top Width behind of electrode
Base metal Speed Dilution
(A) (mm) electrode (B) (C) (mm)
(mm/s) (%)
(mm)
6060-T4 0 0 7.5 6.1 4.7
6060-T4 41.7 16 8.1 5.9 4.8
Cross
Bead Root Over-Bead Over-Bead
Section
Thickness Width Width Curvature
Area
(mm) (mm) (mm) (mm–1)
(mm²)
25.1 4.2 5.5 7.5 –0.068
31.7 5.0 6.5 8.1 +0.011
308 N. Coniglio, C.E. Cross

Table A.2. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6. Distances (A), (B), (C) are according to Fig. 8
Distance Distance in
Filler Wire 4043
Top Width behind front of
Base metal Speed Dilution
(A) (mm) electrode electrode
(mm/s) (%)
(B) (mm) (C) (mm)
6060-T6 0 0 7.5 6.3 5.6
6060-T6 41.7 17 7.8 5.9 5.3
Cross Sec- Bead Root Over-Bead Over-Bead
tion Area Thickness Width Width Curvature
(mm²) (mm) (mm) (mm) (mm–1)
26.1 4.8 5.8 7.8 –0.066
31.5 5.0 6.9 8.1 +0.009

Table A.3. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 with 6060 insert. Distances (A), (B), (C) are according to Fig. 8
Distance Distance in
Filler Wire 4043
Top Width behind front of
Base metal Speed Dilution
(A) (mm) electrode electrode
(mm/s) (%)
(B) (mm) (C) (mm)
6060-T6 0 0 8.0 5.6 6.1
6060-T6 41.7 17 7.9 6.0 5.0
Cross Bead Root Over-Bead
Over-Bead
Section Thickness Width Curvature
Width (mm)
Area (mm²) (mm) (mm) (mm–1)
24.2 4.3 5.7 8.1 –0.056
28.7 5.1 6.0 8.0 +0.025

Table A.4. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 with 6060+1.7%Fe insert. Distances (A), (B), (C) are according to
Fig. 8
Distance Distance in
Filler Wire 4043
Top Width behind front of
Base metal Speed Dilution
(A) (mm) electrode (B) electrode
(mm/s) (%)
(mm) (C) (mm)
6060-T6 0 0 8.3 6.0 6.4
6060-T6 41.7 20 7.5 5.1 5.5
Cross
Bead Root Over-Bead
Section Over-Bead
Thickness Width Curvature
Area Width (mm)
(mm) (mm) (mm–1)
(mm²)
26.7 4.3 5.7 8.5 –0.053
29.6 5.3 6.5 7.6 +0.025
 Weld Parameter and Minor Element Effects 309

Table A.5. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 with 6060+0.8%Ti insert. Distances (A), (B), (C) are according to
Fig. 8
Distance Distance in
4043
Filler Wire Top Width behind front of
Base metal Dilution
Speed (mm/s) (A) (mm) electrode electrode
(%)
(B) (mm) (C) (mm)
6060-T6 0 0 8.9 6.2 6.3
6060-T6 41.7 15 9.0 6.0 5.6
Cross Bead Root Over-Bead
Over-Bead
Section Area Thickness Width Curvature
Width (mm)
(mm²) (mm) (mm) (mm–1)
25.2 4.3 4.5 8.4 –0,025
28.7 5.1 5.1 7.9 +0.015

Table A.6. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 welded at a torch travel speed of 2 mm/s. Distances (A), (B), (C)
are according to Fig. 8
Distance Distance in
Filler Wire 4043
Top Width behind front of
Base metal Speed Dilution
(A) (mm) electrode electrode
(mm/s) (%)
(B) (mm) (C) (mm)
6060-T6 0 0 8.2 6.6 7.2
6060-T6 20.8 18 8.3 6.3 6.7
Cross Bead Root Over-Bead Over-Bead
Section Thickness Width Width Curvature
Area (mm²) (mm) (mm) (mm) (mm–1)
25.7 4.0 6.3 8.3 –0.074
28.5 5.2 6.1 8.1 +0.000

Table A.7. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 welded at a torch travel speed of 6 mm/s. Distances (A), (B), (C)
are according to Fig. 8
Distance Distance in
Filler Wire 4043 Top Width behind front of
Base metal
Speed (mm/s) Dilution (%) (A) (mm) electrode electrode
(B) (mm) (C) (mm)
6060-T6 0 0 7.3 5.4 5.6
6060-T6 62.5 16 7.2 5.4 4.3
Cross Bead Over-Bead Over-Bead
Root Width
Section Thickness Width Curvature
(mm)
Area (mm²) (mm) (mm) (mm–1)
25.1 4.2 6.5 7.6 –0.107
30.1 5.1 6.7 7.9 –0.015
310 N. Coniglio, C.E. Cross

Table A.8. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 welded with He+(Ar+1%O2) shielding gas. Distances (A), (B), (C)
are according to Fig. 8
Distance Distance in
Filler Wire
4043 Top Width behind front of
Base metal Speed
Dilution (%) (A) (mm) electrode electrode
(mm/s)
(B) (mm) (C) (mm)
6060-T6 0 0 8.2 5.8 6.4
6060-T6 41.7 18 7.6 6.0 6.4
Cross Bead Over-Bead
Root Width Over-Bead
Section Area Thickness Curvature
(mm) Width (mm)
(mm²) (mm) (mm–1)
27.6 4.2 7.1 8.4 –0.108
29.1 5.1 6.4 7.6 +0.000

Table A.9. 4043 filler dilution and dimensional characteristics for CTW welds of
alloy 6060-T6 welded with He+(Ar+2%H2) shielding gas. Distances (A), (B), (C)
are according to Fig. 8
Distance Distance in
Filler Wire
4043 Top Width behind front of
Base metal Speed
Dilution (%) (A) (mm) electrode electrode
(mm/s)
(B) (mm) (C) (mm)
6060-T6 0 0 8.7 5.9 6.2
6060-T6 41.7 18 8.4 5.8 6.0
Cross Bead Over-Bead
Root Width Over-Bead
Section Area Thickness Curvature
(mm) Width (mm)
(mm²) (mm) (mm–1)
26.4 4.3 6.7 8.8 –0.084
31.4 5.1 6.4 8.5 +0.015
Using Simulation for Investigations of Hot
Cracking Phenomena in Resistance Spot Welding
of 6xxx Aluminum Alloys (AA6016 and AA6181)

A. Eder, S. Jaber, N. Jank

Fronius International, Wels-Thalheim, Austria

Abstract

This study presents how Numerical Simulation, by using the Finite Ele-
ment Method (FEM), can help to avoid hot cracking phenomena for resis-
tance spot welding of 6xxx AlMgSi alloys. It is well known that aluminum
spot welding is a difficult task especially for 6xxx alloys. Small silicon
contents in these alloys increase the cracking sensitivity dramatically in
comparison with 5xxx alloys. In order to reduce tedious experimental work
and to understand such problems SORPAS®, a commercial available Finite
Element Method software for resistance welding, was used. Virtual simu-
lation experiments for the innovative resistance spot welding process
DELTA SPOT™ led to optimized welding parameters in order to get the
desired nugget size without surface cracks.

Introduction

Using the innovative resistance spot welding process DELTA SPOT™ [1]
one can weld a wide range of metal combinations like three-sheet alumi-
num, three-sheet steel and dissimilar metals like aluminum-steel. All this is
made possible by using a coated process tape in combination with a resis-
tance spot welding gun – Figs. 1 and 2.
312 A. Eder et al.

Fig. 1. DELTASPOT™ electrodes with a process tape (courtesy of FRONIUS Inc.

Welding technology. Austria)

The tape is posed between the electrodes and the sheets to be joined.
The resistance caused by the tape generates an additional heat input into
the work piece. It not only allows dissimilar material welding but also
eliminates work piece-coating and contaminant effects. It leads to ideal
contact, splash-free welds and high electrode lifetime. After every weld the
process tape moves to its next position. The contact surface therefore al-
ways remains clean. Additionally the fine coating of the tape optimizes the
electrical contact to the work piece, especially for welding aluminum
where it prevents spatter formation. Each welding spot is reproducible –
there is always the same quality as at the start.
DELTA SPOT™ was designed with the automotive and supplier indus-
tries very much in mind. One goal is to solve difficult welding tasks aris-
ing from the use of new materials in the automotive industry. Therefore
welding aluminum, especially high strength aluminum alloys, is an impor-
tant topic. For resistance spot welding of aluminum the process tape has
another important effect. It can be used to control the heat input into the
workpiece. The additional heat due to the resistance of the process tape
compensates the negative effect of the high conductivity of aluminum. The
tape leads to high process efficiency and reliability.
This study concentrates on the problems arising by resistance spot weld-
ing of aluminum and its alloys – especially the problem of hot cracking.
The difference between welding with and without process tape is demon-
strated for AlMg3 (AA 5754). As an example for difficult to weld 6xxx
aluminum alloys, AlMg0.4Si alloys are investigated. For these alloys it is
demonstrated how simulation can be used to find an optimal process curve
for resistance spot welding in order to avoid hot cracking.
 Using Simulation for Investigations of Hot Cracking Phenomena 313

Fig. 2. DELTASPOT™ spot welding gun (courtesy of FRONIUS Inc. Welding

technology. Austria)

Welding High Strength Aluminum Alloys

Aluminum is widely used for light weight constructions in automotive and

aerospace industries due to its perfect properties: low density, third most
abundant element, highly formable (face centered cubic crystal structure).
The low strength and stiffness is compensated by alloys with other ele-
ments. A brief overview of aluminum alloys used for automotive applica-
tions is given in [2]. See [3] for aerospace applications.
But, most high strength aluminum alloys are considered to be unweld-
able. One of the major metallurgical problems of high strength aluminum
alloys is hot cracking. Cracks arise when the thermal stresses generated
during cooling exceed the strength of the almost solidified metal. The ten-
dency of cracking grows with the solidification range of the aluminum al-
loys – Fig. 3.
The cracking sensitivity is depending on the alloying element contents.
Si and Mg have a strong effect on the weld cracking sensitivity [4]. Simple
phase diagrams for crack prediction are not possible because real commer-
cial aluminum alloys contain many alloying elements.
Strategies for avoiding hot cracking are e.g. preheating, a controlled
shrinkage of the parts or low heat input. Spot welding with a process tape
is another possibility which is demonstrated in the following section.
314 A. Eder et al.

Fig. 3. Melting and solidification interval of aluminum alloys

Spot Welding of Aluminum With and Without a Tape

In order to demonstrate the difference between welding aluminum with

and without a process tape we made resistance spot welding experiments
for AlMg3 (AA5754). The process parameters for the 1 mm sheets were
3 kN electrode force and approximately 16 kA welding current in one
pulse mode. In the case without a process tape a higher welding current
than in the process tape case was necessary. During the welding series the
current was held constant independent of the electrode contamination. The
welding time was set to 400 ms (incl. upslope and down-slope time). R40
electrodes of 12 mm in diameter were used.
Due to the rapid cooling of the molten weld pool hot cracking and
sometimes solidified eutectics appear when joining with the conventional
spot welding process. Figure 4 shows the macro section for spot welded
AlMg3 sheets without a process tape. The weld nugget has an elliptic form
and is of approximately 4 mm in diameter. The cracks start at the weld
nugget and are directed outward in a typical angle [5]. Within a few spot
welds there is a contamination of the electrodes with aluminum. The con-
tamination dramatically affects the appearance of the welded spot. This
means there has to be done a cleaning of the electrodes after three to five
spot welds.
 Using Simulation for Investigations of Hot Cracking Phenomena 315

Fig. 4. Welding AlMg3 without a process tape

These phenomena do not appear for welding AlMg3 with a process tape
(in this case PT1407). The process parameters where chosen as in the con-
ventional spot welding example, except the current which could even be
reduced. In the macro-section there are just some small eutectics visible.
Compare Fig. 5. The nugget size is larger and it has a cylindrical form
which leads to a higher tensile strength – Fig. 6 [6]. The cylindrical form
of the weld nugget is caused by the additional heat from the process tapes.
This changes the cooling behavior and the stress distribution in the sheets
which helps reduce cracking sensitivity. It also reduces the gap adjacent to
the weld nugget.
As mentioned above for spot welding with a process tape less current is
necessary due to the additional heat from the tape. It makes a weld more
economically. Additionally the electrodes stay clean due to the protection
of the tape. Therefore a high quality of the sheet surface appearance can be
reached.

Fig. 5. Welding AlMg3 with a process tape

316 A. Eder et al.

Fig. 6. Tensile test results for AlMg3

Hot Cracks in Spot Welding of 6xxx Alloys

Spot welding of the high strength aluminum alloys is much more sophisti-
cated to solve. Test welds of 6xxx (AA6016) alloys with the process tape
PT 1200 were made. The welding parameters were 3 kN with one pulse of
approx. 17 kA for two 1 mm sheets. The standard electrode R70 was used.
Even with the aid of a process tape cracks appeared at the surface and in
the nugget – see Fig. 7. Due to the large solidification interval of AA6016
and the critical concentration of the alloying elements a high risk of crack-
ing is present.
In order to understand and to solve this problem the temperature profile
during cooling of the spot weld was needed. Therefore a series of Finite
Element simulations with the commercial available resistance welding pro-
gram SORPAS® were made.
 Using Simulation for Investigations of Hot Cracking Phenomena 317

Fig. 7. Hot cracks in spot welded AA 6016 sheets

Resistance Spot Welding Simulation with a Tape

The simulation of the resistance spot welding process involves strong in-
teractions between the work pieces, electrodes, interfaces, process tape,
coatings and machine settings. Figure 8 [7, 8] shows the main components
which have to be considered for resistance spot welding with a process
tape.

Fig. 8. Simulation components for the DELTASPOT™ process

318 A. Eder et al.

SORPAS® takes into account the metallurgical, electrical, mechanical

and thermal process that will occur during the welding process using Finite
Element modeling. The software simulates the heat generated by the weld-
ing current, the resistance of the electrodes, the tape, the coatings and the
work piece, the heat transfer across materials, the metallurgical phase
transformation (solid to liquid) caused by temperature change, the defor-
mation and strain distribution across the contact areas and the process win-
dow. All this information is then presented in graphic form to the user.
For standard materials a material database is included in the software
package, but also new materials can be added if necessary. Special care has
to be taken for the process tape and the coatings since they are relatively
thin and have a different physical behavior compared to the work-piece
and the electrodes. The coatings have an influence on the contact resis-
tance and are usually evaporated during the process. Additionally some
machine setting parameters like electrodes, and current and force curves
are needed to complete the process model.

Fig. 9. FEM mesh for resistance spot welding with a process tape
 Using Simulation for Investigations of Hot Cracking Phenomena 319

A typical FEM mesh for solving a resistance spot welding process with an
additional tape is shown in Fig. 9. An adaptation of the standard program
settings was necessary to be able to include the thin tape and coatings.

Mathematical Models for Spot Welding Simulation

In order to calculate the time dependent distributions of voltage, current

and temperature in materials and electrodes we need the following govern-
ing axi-symmetric differential equations [7] using cylindrical coordinates
(r ,φ ) .
The differential equation for the electrical potential φ is

1 ∂ § ∂φ · ∂ § ∂φ · (1)
¨ rσ r ¸ + ¨σ z ¸ = 0,
r ∂r © ∂r ¹ ∂z © ∂z ¹
where σ is the electrical conductivity of the materials.
The temperature distribution T is a result of the transient heat transfer
differential equation with internal heat source Q

1 ∂§ ∂T · ∂ § ∂T ·  ∂T (2)
¨ rλr ¸ + ¨ λz ¸+Q = ρ C ,
r ∂r © ∂r ¹ ∂z © ∂z ¹ ∂t
where λ is the thermal conductivity, ρ is the mass density and C is the
heat capacity. All material data is given in a temperature dependent way,
which allows phase transformation calculations.
The deformation, the stress, strain distribution and the contact areas at
interfaces are calculated by the variation approach for the functional of the
potential energy
∂π (3)
π = ³ σ εdV − ³ FvdS , =0
V S
∂v
The first term on the right hand side is the potential energy of the bulk
deformation. The second term respects the boundary conditions due to ex-
ternal load or velocity and friction etc. Wanheim and Bay’s friction theory
for real contact areas [9] is used for the contact resistivity:

§σ ·§ ρ1 + ρ 2 ·
ρ contact = 3¨¨ s _ soft ¸¸¨ + ρ calibration ¸ ,
© σn ¹© 2 ¹ (4)
320 A. Eder et al.

where σ s _ soft is the flow stress of the softer metal, σ n is the contact
normal pressure at the interface. The subscripts 1 and 2 indicate the two
base metals in contact. To take account for surface contaminants such as
oil, water, dirt and oxides etc. a term called ρ calibration is introduced in the
model. The quantitative influence of the surface contaminants on the con-
tact resistance is usually not known. Therefore the parameter ρ calibration in
Eq. (4) can be used for calibration. In order to complete the mathematical
model, appropriate boundary and initial conditions are used for the electri-
cal potential, temperature field and mechanical model.

Solver Strategy

Equations (1–4) represent strongly coupled models, an electrical, a ther-

mal, a metallurgical and a mechanical one. These models are all influenced
by the dynamic behavior of the materials, interfaces, the machines and the
processes.

Fig. 10. SORPAS® procedure to solve the governing equations

FEM techniques are applied for Eqs. (1–4) to obtain a numerical discretiza-
tion. Figure 10 shows the solver strategy, designed to obtain efficient spot
welding simulations [7].
 Using Simulation for Investigations of Hot Cracking Phenomena 321

Parameter Optimization Process

During resistance spot welding the main process parameters, welding cur-
rent and electrode force, can be varied in order to get high quality welds.
At the beginning of a spot weld the force increases in a linear manner
which leads to a good electrical contact between the work pieces. After
this preload the current pulse starts. Depending on the material there can
be used one or several current pulses. For example it can be useful to set a
preheating pulse to avoid critical cooling rates [10].
The process engineer has to choose the right process curve for the mate-
rials to be welded. Due to the variety of process curves this can be a diffi-
cult and time consuming task. SORPAS® allows any force and current
form as well as material combinations which makes it suitable for process
optimization. Several simulations can be run successively in a batch proc-
ess for parametric studies. The optimization loop applied for solving diffi-
cult spot welding problems is shown in Fig. 11.

Fig. 11. Virtual resistance spot welding optimization loop

Usually the simulation starts on the basis of a first practical trial. Ac-
cording to a spot weld macro-section resulting from a first guess of a
current-force pulse the simulation model is calibrated. The calibration usu-
ally is done via the contact resistance model – compare Eq. (4). Then vir-
tual parameter studies by variation of the shape of the current-force pulse
have to be done. This results in a series of work piece temperature profiles.
These temperature histories in combination with metallurgical knowledge
322 A. Eder et al.

(e.g. CCT-diagrams, optimal cooling rates) can be used to get suggestions

for a new practical trial. If the following welding experiment is of suffi-
cient quality the loop is finished, otherwise it starts again (new adapted
model, other strategies, and so on). In order to get reliable results a strong
collaboration between Numerical Simulation and experimentation engi-
neers has to be guaranteed.

Parameter Studies for High Strength Aluminum Alloys

The goal for resistance spot welding the aluminum alloys AA 6016 and
AA 6181 was to minimize hot cracking as well as to get the desired nugget
size. A virtual parameter study for different types of welding process
curves (current and force) was done in SORPAS®. Exemplary are some re-
sulting temperature histories close to the work piece surface on the sym-
metry axis for AA 6016 as shown in Fig. 12.

Fig. 12. Simulation of temperature history close to the work piece surface
 Using Simulation for Investigations of Hot Cracking Phenomena 323

Fig. 13. Cracks at the surface for an optimized AA 6016 weld

The simulated temperature curves are results of different current-force

welding process curves. In Fig. 12 the solidification interval is marked by
bold black horizontal lines. Trials 1 and 2 show the temperature history of
the first simulated trial process curves which were not sufficient for crack
prevention. At the end of the parameter study the lower cooling rates of the
optimized curves Opt1 and Opt2 are expected to prevent from crack for-
mation. The optimized cooling slope allows the healing of the surface
cracks due to the force applied during solidification. Figure 13 shows the
surface cracks of an AA 6016 spot weld at the beginning of the parameter
study and the crack-free weld with an optimized process curve. The real
and the virtual macro-sections of the spot welding are in good agreement
as can be seen in Fig. 14.

Fig. 14. Validation of the simulation results

324 A. Eder et al.

Fig. 15. Surface of an AA 6181 weld before and after optimization

For resistance spot welding of 2 mm AA 6181 sheets with the process

parameters 24 kA, 220 ms, 5 kN, electrode R150 and process tape PT1407
the same procedure was performed. The successful optimization of the
welding task can be seen in Fig. 15. Even the extreme cracks at the surface
disappear.

Conclusion

Resistance spot welding of aluminum alloys is a difficult task. Hot crack-

ing is one of the main problems to be solved. The goal in this study was to
minimize cracking phenomena especially for the 6xxx alloys AA6016 and
AA 6181. The solution strategy was based on numerical simulation to-
gether with an innovative resistance spot welding process including a proc-
ess tape.
First the difference between spot welding with and without a process
tape was demonstrated by the use of DELTA SPOT™ for resistance spot
welding of AlMg3. Additionally SORPAS®, a commercial resistance weld-
ing simulation software that uses finite element techniques, was introduced
as a tool for spot welding simulation including a process tape.
For the difficult to weld AA 6016 and AA 6181 aluminum alloys virtual
parameter studies for resistance spot welding with a process tape were per-
formed. Validation and analysis of the simulations led to optimized tem-
perature histories and finally to crack free welding process curves.
Solving difficult spot welding problems is not anymore just a question
of trial and error experiments. Simulation can help to investigate difficult
spot welding problems by supplying fast and reliable answers.
 Using Simulation for Investigations of Hot Cracking Phenomena 325

References

1. Stieglbauer W (2006) DeltaSpot – Spot welding with process tape. IIW

Doc.III-1381-06
2. Irving B (2000) The auto industry gears up for aluminum. Welding Journal,
11:63–68
3. Robson J (2002) Designing high strength aluminum alloys for aerospace ap-
plications. Manchester Materials Science Centre.
4. Krüger U (1994) Weldability. TALAT Lecture 4202, Schweißtechnische
Lehr- und Versuchsanstalt Berlin. EAA – European Aluminium Association.
5. Senkara J, Zhang H (2000) Cracking in spot welding aluminium alloy
AA5754. Welding Journal 7:194–201
6. Sun X, Stephens EV, Davies RW, Khaleel MA, Spinella DJ (2004) Effects of
fusion zone size on failure modes and static strength of aluminum resistance
spot welds. Welding Journal 11:308–318
7. Zhang W (2003) Design and implementation of software for resistance weld-
ing process simulations. SAE 2003 World Congress, Detroit, USA
8. Zhang W (2005) Recent advances and improvements in the simulation of re-
sistance welding processes. IIW Doc.III-1345-05
9. Wanheim T, Bay N (1987) A model for friction in metal forming processes.
Annals of the CIRP 27:189–194
10. Matthes KJ, Richter E (2006) Schweißtechnik, Schweißen von metallischen
Konstruktionswerkstoffen. Carl Hanser Verlag München Wien. ISBN-10:
3-446-40568-2
 Part IV
Liquation Cracking
Evaluating Hot Cracking Susceptibility of Ni-Base
SAW Consumables for Welding of 9% Ni Steel

L. Karlsson, E-L Bergquist, S. Rigdal, N. Thalberg

ESAB, Göteborg, Sweden

Abstract

A study of Ni-base weld metal hot cracking resistance was made compar-
ing results from Modified Varestraint-Transvarestraint Testing (MVT),
metallographic evaluation and mechanical testing. Submerged arc weld
metals were produced with Alloy 625, Alloy 59 and Alloy C-276 filler
wires at two levels of heat input. Welds were subjected to MVT testing,
three-point and wrap-around bend testing as well as metallographic inspec-
tion of weld cross-sections. All test methods generally ranked the lower
heat input welds as more crack resistant than those produced with a higher
heat input. However, MVT testing ranked alloy types in a different order
than mechanical testing and metallographic evaluation. Alloy 59 type weld
metals were judged most crack resistant with MVT testing and C-276
welds were ranked as slightly better than those produced with Alloy 625
wires. Mechanical testing and metallographic evaluation, on the other
hand, ranked the Alloy 625 type weld metals as most crack resistant and
Alloy C-276 weld metals as least resistant. Discrepancies between results
of the different test methods and practical experience can most likely be at-
tributed to a number of factors including metallurgical aspects and test
procedures. It is concluded that test methods simulating actual application
welding conditions as closely as possible are most likely to accurately pre-
dict hot cracking susceptibility of real weldments. An assessment of the
hot cracking susceptibility of welding consumables should therefore pref-
erably be based on results from a combination of test methods.

Introduction

Large vessels and tanks for transportation and storage of liquefied natural
gas (LNG) at cryogenic temperatures are since many years commonly
constructed from 9% Ni steel. One of the most important steps in the
330 L. Karlsson et al.

fabrication of storage tanks is welding. Obviously both productivity and

consistent high weld quality are key issues. Submerged arc welding
(SAW) offers the highest productivity and is therefore usually the pre-
ferred method when geometry and welding position permits [1, 2].
The weld metal should provide low temperature properties matching
those of the 9% Ni base material. Since many years, it is therefore standard
practice to use Ni-base consumables. These safely comply with require-
ments on mechanical properties, including strength and ductility, and have
a thermal expansion coefficient closely matching that of 9% Ni steels.
Several types of Ni-based consumables have been developed and the prin-
cipal compositions are now standardised (e.g. solid wires for SAW accord-
ing to AWS A5.14 and EN 18274). However, in practice only a few are
able to meet the very high requirements for 9% Ni applications. Basic
submerged arc welding fluxes are considered mandatory to produce a clean
deposit with a low amount of microslag inclusions improving toughness
and minimising the risk of hot cracking. Many of these Ni-base alloys are
also used for welding of highly austenitic stainless where overalloying is
needed to match parent material corrosion resistance [3].
Several hundreds of large storage tanks and vessels fabricated from 9%
Ni steel and welded with Ni-base consumables are in operation. Many
years of failure-free experience from these show that fit-for-purpose welds
can be produced but different opinions exist as to which filler materials are
most suitable.
Hot cracking tests are rarely parts of welding consumable qualification
and resistance to hot cracking is generally not specifically discussed. It is
nevertheless an important aspect in the selection and qualification of con-
sumables and is sometimes identified as a problem through evaluation of
metallographic specimens. The main influence is however more indirect
sometimes causing unacceptable cracking of transverse and longitudinal
bend test specimens or low elongation and premature fracture of tensile
test specimens. A complication is that both the filler materials inherent hot
cracking susceptibility and the welding procedure influence these results.
Since a badly designed welding procedure can cause hot cracking in any
type of weld metal it can be difficult to agree on whether the consumable
or the procedure is to be blamed. Reliable methods for evaluation of hot
cracking susceptibility producing results comparable to those from fabrica-
tion experience are therefore needed.
Many different hot cracking test methods have been developed over the
years [4]. In principle all of these can be classified into two major groups:
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 331

• Self Restrained Tests where the specimen loading is produced by

stresses developed during welding of a restrained weldment and
• Externally Loaded Tests that require the use of equipment to impose a
strain on the specimen during welding or some other thermal cycle.
Some tests are fully standardised whereas others are only described in
EN/ISO technical reports [4, 5]. Externally loaded tests such as the Vare-
straint and the Transvarestraint tests have, although not being fully stan-
dardised, the advantage of rapid testing, low scatter of results and good re-
producibility (with a single machine). Useful data can also be generated
from a small number of tests.
The chapter presents results from a study comparing results from Modi-
fied Varestraint-Transvarestraint Testing (MVT), metallographic evalua-
tion and mechanical testing. SAW weld metals produced with three types
of Ni-base filler wires (Alloy 625, Alloy 59 and Alloy C-276) at two levels
of heat input were tested. Test results are presented and the influence of
test method on hot cracking susceptibility ranking is discussed.

Experimental

Three types of Ni-base SAW weld metals were subjected to MVT testing,
metallographic evaluation and bend testing.

Weld Metal

SAW weld metals were produced with three types of 1.6 mm diameter Ni-
base filler wires (Alloy 625, Alloy 59 and Alloy C-276). Welding was
done with a basic flux at two levels of heat input for each filler wire type.
Typical all-weld metal composition and properties for the three SAW
wire/flux combinations are presented in Tables 1 and 2.

Table 1. SAW consumables for 9% Ni-steel. Typical all-weld metal composition

(wt-%)
Filler wire
C Si Mn Cr Ni Mo Nb W Fe
alloy type
59 <0.01 0.2 3 22 Bal. 14 – – 3
625 0.01 0.2 1.5 21 Bal. 8.5 3 – 2
C-276 0.01 0.2 1.9 15 Bal. 14 – 3.5 7
332 L. Karlsson et al.

Table 2. Consumable classification and typical all-weld metal properties

Consumable Tensile properties Charpy-V
Filler wire classification -196°C
alloy type AWS A5.14 / Rp0.2 Rm A5
EN 18276 [MPa] [MPa] [%] [J]
ERNiCrMo-13 /
59 470 675 45 70
S Ni 6059
ERNiCrMo-3 /
625 440 720 40 90
S Ni 6625
ERNiCrMo-4 /
C-276 480 700 35 75
S Ni 6276

A standard AWS joint configuration with 60º included angle, 13 mm

root opening and a backing plate was used. Joint faces were buttered prior
to deposition of the welds to minimise dilution with parent material. Welds
were produced with a heat input of 0.9 kJ/mm and 1.5 kJ/mm. The weld
metals were designated starting with the type of filler wire alloy followed
by L or H for the lower and the higher heat input level, respectively. For
example 625-H stands for a weld metal produced with an Alloy 625 wire
with a heat input of 1.5 kJ/mm.
One set of welds was subjected to MVT testing followed by additional
metallographic evaluation whereas the other was studied by three-point
and wrap-around bend testing as well as inspection of weld cross-section.

Metallographic Evaluation

Weld cross-sections were examined for the presence of cracks. Three

specimens were extracted from each of the six welds, near the start and
stop ends and from the centre. A limited study was also done on some
specimens subjected to MVT testing.
The cross-sections were prepared using standard metallographic
techniques, electrolytically etched in 70% H3PO4 and examined using a
stereo microscope and at higher magnification in a light optical
microscope. Cracks were counted and classified based on their location
relative to bead fusion boundaries. Those connected to a fusion boundary
of a bead were classified as liquation (and /or ductility dip) cracks and
cracks in the interior of beads as solidification cracks (Fig. 1).
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 333

Fig. 1. Weld cross sections illustrating classification of cracks based on location

relative to bead fusion boundaries
(a) Solidification cracks in the interior of beads in weld 276-L
(b) Liquation cracks at bead fusion boundary in weld 59-L

Bend Testing

The presence of cracks was also evaluated using three-point bend and wrap
around bend testing. Three-point bend testing was done with 10 mm thick,
transverse side bend specimens and longitudinal face bend test specimens.
The diameter of the former was in both cases 40 mm and the bending an-
gle was 180°.
Wrap around testing was done using 5 mm thick, transverse side bend
test specimens. The diameter of the former was 24 mm and the bending
angle was 180°. All testing was done in accordance with SS-EN 910.

MVT Testing

In MVT testing a GTA torch is used to remelt the machined surface of the
weld metal to be tested. As the arc passes a predefined point the specimen
is bent at a constant rate around a die. The specimen surface is thereby
subjected to a strain of well-defined level and rate. As a result of the
334 L. Karlsson et al.

stress/strain distribution during bending the hot cracks form almost exclu-
sively at or near the specimen surface. On completion of the test the
specimen surface is visually examined for cracks at a magnification of
X25. The number and total length of all visible cracks are determined and
plotted as a function of the bending strain [5].
Specimens (10×40×100 mm) for MVT testing were machined 2 mm
below the weld top surface resulting in a test surface located largely in the
top layer of beads. Welding was done under Ar and with a nominal energy
input of approximately 0.75 kJ/mm. Bending was done transverse to the
welding direction at a bending rate of 1.8 mm/s with bending radii selected
to produce strains of 1%, 2% and 4%. One specimen was tested per weld
and strain level.

Results

Mechanical testing and metallographic evaluation of hot cracking tendency

produced the same ranking of filler wires irrespective of heat input.
However, MVT testing resulted in an entirely different ranking.

Metallographic Evaluation

The metallographic evaluation of cross-section produced the same very

clear ranking of the three types of filler wires regardless of whether solidi-
fication, liquation (and ductility dip) or the total number of cracks was
considered (Table 3). 625 welds contained only a few small cracks (1-2 per
cross-section), the 59-type welds an average of 4-5 cracks and the C-276
welds typically 10 cracks.
A representative cross section is shown for each type of filler wire in
Fig. 2. As seen in the figure, the C-276 type welds not only contained a
larger number of cracks but the average size also tended to be larger.
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 335

Fig. 2. Examples of cross-sections used for metallographic evaluation of hot

cracking tendency. (a) 59L, (b) 625-L, (c) 276-L
336 L. Karlsson et al.

Table 3. Number of hot cracks as determined by metallographic evaluation on

three weld cross-sections from each weld
Weld Number of cracks (on 3 cross-sections)
Solidification cracks Liquation/ductility dip Total
cracks
59-L 4 8 12
59-H 0 15 15
625-L 1 2 3
625-H 1 4 5
276-L 22 9 31
276-H 10 21 31

Bend Tests

Results from bend testing are summarised in Table 4. The two types of
side bend tests and the longitudinal bend test were in complete agreement.
Welds 625-L and 625-H passed without remarks except in side bend wrap
around testing where one small crack was detected at the edge of the
specimen. Only a limited number of small cracks shorter than 0.5 mm
(here called micro-cracks) were seen in 59-L and 59-H specimens (Fig. 3).
All 276-L and 276-H specimens contained numerous micro-cracks and in
most cases also larger cracks (Fig. 4). One 276-H 3-point side bend
specimen even broke at a bending angle of 120°.

Table 4. Bend test results

Weld Side bend, Longitudinal Side bend,
3-point face bend wrap around
59-L micro-cracks micro-cracks micro-cracks
59-H micro-cracks micro-cracks micro-cracks
625-L no remark no remark no remark
625-H no remark no remark 1 micro-crack at edge
276-L large cracks and large cracks several micro-cracks
micro-cracks
276-H large crack and micro- micro-cracks several micro-cracks
cracks, one specimen
broken at 120°
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 337

Fig. 3. Micro-cracks on strained surface of 59-H bend test specimens

(a) 3-point side-bend specimen
(b) Longitudinal face-bend specimen
338 L. Karlsson et al.

Fig. 4. Large cracks and numerous micro-cracks on strained surface of 276-L bend
test specimens
(a) 3 point side-bend specimen
(b) Longitudinal face-bend specimen

MVT Tests

Results from MVT testing is presented as tables giving total cracks lengths
(Table 5) and total number of cracks (Table 6) for all strain rates, as a
standard crack length versus strain diagram (Fig. 5) and as a diagram
presenting number of cracks against strain (Fig. 6).
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 339

Table 5. Results of MVT tests presented as combined crack lengths for all strains
Weld Combined crack length for 1%, 2% and 4% strain (mm)
GTA WM Weld metal GTA WM + “Distant Total
HAZ Weld metal cracks”
HAZ
59-L 10.2 3.8 14.0 5.3 19.3
59-H 6.8 4.1 0.9 2.1 13.0
625-L 16.8 3.1 19.9 1.6 21.5
625-H 17.6 2.6 20.2 2.2 22.4
276-L 12.1 5.6 17.7 4.0 21.7
276-H* 15.3 6.8 22.1 10.3 32.4
* only tested at 2% and 4% strain

Table 6. Results of MVT test presented as the total number of cracks for all
strains
Weld Number of cracks
GTA WM Weld metal GTA WM + “Distant Total
HAZ Weld metal cracks”
HAZ
59-L 21 10 31 18 59
59-H 16 12 28 7 35
625-L 29 14 43 10 53
625-H 38 19 57 4 61
276-L 22 14 36 17 53
276-H* 20 18 38 37 75
*only tested at 2% and 4% strain

Newly formed cracks, at a distance from the GTA bead fusion boundary
(ductility dip cracks), could not easily be distinguished from hot cracks
already present in the weld metals before MVT testing. The “distant”
cracks are therefore included in the tables but excluded from the data used
to construct diagrams.
Using combined crack lengths or total number of cracks can produce
various rankings of hot cracking sensitivity depending on whether cracks
in weld metal, HAZ or distant cracks are considered. Looking only at weld
metal cracks the 59 type weld metals were least crack sensitive and the 625
welds most prone to cracking. Ranking in HAZ depends on whether num-
ber of cracks or crack length is used as criteria. However, including or ex-
cluding distant cracks has the most dramatic effect on ranking. The C-276
type weld metals were clearly most crack prone at a distance from the
GTA bead whereas the 625 were least crack sensitive in this region.
340 L. Karlsson et al.

Including all cracks thereby shifted the ranking suggesting the 625 or 59 type
weld metals to be most resistant and the C-276 type more likely to crack.

15,0
Varestraint specimen:
Varestraint specimen:
HAZ
HAZ 59-L
14,0 l l1 1
WM
WM

13,0 HAZ
59-H
HAZ

Crack length measurement

12,0 Crack
¦ length
l ges =
¦ measurement
n

i =1
li , SG +
m

i =1
li ,WEZ
625-L
n m
+
l ges
tot =
tot
¦ liWM,SGWM ++ ¦ lHAZHAZ
i ,WEZ
11,0 i =1 i =1
625-H

10,0
276-L
Total crack length [mm]

9,0
276-H
8,0

7,0

3
6,0
2
5,0
1
4,0

3,0

2,0 Sector:

1: resistant to hot cracking

1,0 2: increasing risk of hot cracking
3: high risk of hot cracking
0,0
0 1 2 3 4 5

Surface bending strain [%]

Fig. 5. Results of MVT testing. Cracks formed at a distance from the GTA re-
melted bead were not included.

Results are presented in Fig. 5 in the standard format presenting total

crack length versus strain. The ranking is varying with strain but the ten-
dency is that the 59 type weld metals are most crack resistant and the
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 341

625 weld metals most prone to hot cracking. According to the classification
normally used when interpreting MVT diagrams the 625 type weld metals
would be considered restricted weldable and the 59 and C-276 types weld-
able or restricted weldable depending on strain rate and heat input [5].

40,0

59-L

35,0
59-H

625-L
30,0

625-H

25,0
276-L
No. of cracks

276-H
20,0

15,0

10,0

5,0

0,0
0 1 2 3 4 5

Surface bending strain [%]

Fig. 6. Results of MVT testing presented as number of cracks versus surface

strain. Only cracks in GTA re-melted bead and adjacent heat affected material
were considered

A presentation of the results as the number of cracks against bending

strain (Fig. 6) resulted in virtually the same ranking as in the previous fig-
ure. The general tendency was again that the 625 weld metals were most
prone to hot cracking and the 59 type least. The main difference was at the
342 L. Karlsson et al.

intermediate strain were the largest number of cracks were found in the
276-H weld metal.
A limited metallographic study of cross-sections of MVT specimens
revealed that most but far from all cracks were surface breaking. Although
statistics is limited there appeared to be a tendency that relatively fewer of
the hot cracks formed in C-276 type welds penetrated the top surface (Fig. 7).
This was also the only alloy type where quite a few rather large cracks
were seen at a distance from the GTA bead fusion boundary.

Fig. 7. Cross section of a C-276 type MVT specimen showing non-surface break-
ing cracks and a crack at some distance from the GTA re-melted bead. The broken
line outlines the MVT GTA weld bead

Discussion

Considering the complex hot cracking mechanisms, variations of stress

and strain fields within welds etc. it is not surprising that various test
methods or tests on different welds with nominally identical compositions
can give different results. However, it is necessary to understand when
well-established test methods can be used as a tool for consumable
selection and when not. The following paragraphs will therefore discuss
possible reasons for discrepancies between the hot cracking test results
presented above.

Ranking of Hot Cracking Susceptibility

Weld metals were ranked with respect to hot cracking susceptibility for
each test (Table 7). As noted earlier, evaluation based on metallography
and bend testing was in very good agreement ranking the 625 weld metals
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 343

as best and C-276 weld metals as worst. All methods also ranked the lower
heat input welds as more crack resistant. As noted earlier these results
were in poor agreement with MVT test results.

Table 7. Comparison of hot cracking tendency ranking based on results from met-
allographic evaluation, bend testing and MVT testing. One is most cracking
resistant and six most prone to cracking
Test method Criteria Weld
59 59 625 625 276 276
L H L H L H
Solidification cracks 4 1 2 2 6 5
Liquation & ductility dip 3 5 1 2 4 6
Metallography
cracks
Total # of cracks 3 4 1 2 5 5
Side bend,
3 3 1 1 5 6
3-point
# and size (semi-
Side bend,
quantitative) of cracks 3 4 1 2 5 5
wrap around
Face bend 3 4 1 1 6 5
Av. ranking of metallographic and bend tests 3.2 3.5 1.2 1.7 5.2 5.3

# of cracks in GTA WM 3 1 5 6 4 2
# of cracks in GTA HAZ 1 2 3 6 3 5
# of cracks in GTA WM &
2 1 5 6 3 4
HAZ
# of “distant” cracks 5 2 3 1 4 6
Total # of cracks 4 1 2 5 2 6
MVT:
all strain rates Crack length in GTA WM 3 1 5 6 4 2
Crack length in GTA HAZ 1 2 3 6 3 5
Crack length in GTA WM
2 1 5 6 3 4
& HAZ
Crack length “distant”
5 2 3 1 4 6
cracks
Total crack length 4 1 2 5 2 6
# of cracks at 1% strain 1 1 4 3 5 n.d
MVT: # of cracks at 2% strain 1 2 4 5 3 6
Cracks in # of cracks at 4% strain 3 1 5 6 4 2
GTA WM Crack length at 1% strain 1 1 4 3 5 n.d
and HAZ Crack length at 2% strain 2 1 6 5 3 4
Crack length at 4% strain 4 1 3 5 6 2
Av. ranking MVT tests 2.6 1.3 3.9 4.7 3.6 4.2
n.d. = no data
344 L. Karlsson et al.

Evaluation of MVT test results did show more scatter in the ranking
compared to metallography of cross-sections or bend testing. However,
there was fairly good agreement between different criteria in that the 59
type weld metals were most crack resistant. On an average the C-276
welds were ranked as slightly better than the 625 type weld metals
although the difference was small and ranking varied significantly with
criteria. With the exception of the 59 welds, also MVT testing ranked
lower heat input specimens as less crack susceptible than those produced
with higher heat input.
It is notable than none of the criteria used to evaluate the MVT
specimens produced a similar ranking as metallography of cross-sections
or bend testing.

Hot Cracking Test Methods and Cracking Mechanisms

The MVT ranking of the alloy types is in good agreement with previous
results from MVT testing predicting Alloy 59 base material to be most
crack resistant and 625 as most likely to crack [6]. Weld metals of 625 and
C-276 type were ranked in the same order as parent material and higher
heat input during testing produced generally more and longer cracks but
did not change the ranking order. However, a more recent comparison
reported Alloy 59 parent material to be significantly more susceptible to
cracking than Alloy 625 at higher strains although the difference was very
small at 1% or 2% strain [7]. Obviously care should be taken when
comparing tests performed on different heats of the same material.
Alloy 59 and Alloy 625 type weld metals have good reputation for
resistance to hot cracking in practical applications although problems are
sometimes encountered [3]. Opinions about Alloy C-276 type filler metals
vary but also these are successfully used in numerous applications. Com-
paring practical experience to test results it appears that there is agreement
on 59 type weld metals being relatively crack resistant. On the other hand,
metallography, bend testing and practical experience suggests a reasonably
good hot cracking resistance whereas MVT ranks 625 type weld metals as
the least resistant of the three alloys tested.
A number of factors contribute most likely to discrepancies between
results of the different test methods and practical experience. These
include metallurgical aspects such as solidification, segregation and
precipitation behaviour of the alloys, tendencies to form non-surface
breaking cracks and differences in heat input between actual welding
procedures and MVT testing.
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 345

Both Alloy 625 and C-276 are known to have rather complex
solidification behaviour [8, 9, 10]. Significant segregation occurs and
solidification is completed by precipitation of topologically close-packed
P-phase in Alloy C-276 whereas a eutectic-type reaction between the
austenitic matrix and various Nb-rich phases such as NbC and Laves phase
occurs in Alloy 625 [3, 11]. However, Alloy 59 is expected to solidify
largely as single-phase austenite with less segregation of alloying elements
and thereby less extension of the solidification range. Alloys 625 and C-276
are also significantly more prone to precipitation during heat treatment as
often seen in reheated regions of weldments [12].
A larger solidification temperature range of is contributing to a greater
risk of solidification cracking as is often observed when comparing 625
and 59 type weldments [3]. Back-filling can on the other hand contribute
to healing of solidification cracks thereby making the effect on cracking
tendency of real weldments smaller. This could partly explain the larger
cracking tendency in MVT testing of 625 welds compared to
metallography and bend testing. Strains are larger and solidification more
rapid during MVT testing, than in actual SAW applications, due to the
lower heat input. The strain rate will also be different during MVT testing
as compared to in actual welding. Another consequence of differences in
heat input is that slower cooling and larger weld pools lead to more
pronounced segregation. The segregated regions will be those where
liquation cracking occurs most easily due to a lower melting temperature.
The importance of heat input was confirmed by test results for all weld
metals, with one exception, ranking lower heat input welds as more crack
resistant (Table 7).
Apart from effects of cracking mechanisms the relatively greater
number of non-surface penetrating cracks in Alloy C-276 welds will
clearly confuse results in MVT testing. The reason is presently not
understood since the fraction of liquation cracks, identified from
metallographic cross-sections (Table 3 and Fig. 2), was equal or higher in
C-276 weld metals than for the other two types.
In conclusion there are indications that the MVT test weld pool size
and cooling rate provokes relatively more solidification cracking in
Alloy 625 compared to in submerged arc welding and thereby predicts
a higher cracking sensitivity than observed in practice. Quite contrary
significant subsurface cracking in C-276 weld metals renders MVT test
results misleading in the opposite direction. Although being a technique
capable of producing useful data from a small number of tests it is clear
that the MVT method should be complemented with other tests
simulating actual application welding conditions to permit a fair
assessment of welding consumables.
346 L. Karlsson et al.

Conclusions

• The hot cracking resistance of submerged arc weld metals, produced

with Alloy 625, Alloy 59 and Alloy C-276 Ni-base filler wires at two
levels of heat input, was evaluated with MVT testing, metallographic
evaluation and bend testing.
• Metallography and bend testing ranked the 625 weld metals as most
crack resistant and C-276 weld metals as least resistant. All methods
generally ranked the lower heat input welds as more crack resistant.
However, MVT testing judged 59 type weld metals as being most crack
resistant and ranked C-276 welds as slightly better than those produced
with Alloy 625 wires.
• Several factors are believed to contribute to discrepancies between
results of the different test methods. These include metallurgical aspects
such solidification, segregation and precipitation behaviour of the
alloys, tendencies to form non-surface breaking cracks and differences
in strain, strain rate and heat input between actual welding procedures
and MVT testing.
• Test methods simulating actual application welding conditions as
closely as possible are most likely to accurately predict hot cracking
susceptibility of real weldments. MVT testing should therefore
preferably be complemented with other methods to permit a fair
assessment of welding consumables.

Acknowledgement

BAM (Bundesanstalt für Materialforschung und -prüfung, Berlin) is

gratefully acknowledged for performing MVT tests and for commenting
on results.

References

1. L.-E. Stridh, L. Karlsson, S. Rigdal and N. Thalberg, “New methods in

Welding of 9% Nickel Steel for LNG Applications”, Proc. IIW 58th Annual
Assembly, International Conference “Benefits of new methods and trends in
welding to economy, productivity and quality”, 14–15 July 2005, Prague,
pp. 39–53.
 Evaluating Hot Cracking Susceptibility of Ni-Base SAW Consumables 347

2. L. Karlsson, S. Rigdal L.-E. Stridh and N. Thalberg, “Efficient Welding of 9%

Nickel Steel for LNG Applications”, Proc. Stainless Steel World 2005 Con-
ference & Expo, 8–10 Nov 2005, Maastricht, pp. 138–142.
3. L. Karlsson, S. Rigdal and S.L. Andersson, “Welding of highly alloyed
austenitic and duplex stainless steels”, Welding in the World, 1997, 39 (2)
pp. 99–110.
4. Hot cracking phenomena in welds, Editors Th. Böllinghaus and H. Herold,
Springer-Verlag Berlin Heidelberg, 2005.
5. CEN ISO 17641 Part 1–3, Destructive tests on welds in metallic materials –
Hot cracking tests for weldments – Arc welding processes.
6. U. Brill, T. Hoffman and K. Wilken, “Solidification cracking: Super stainless
steels and nickel base alloys”, Proc. Materials weldability symposium,
Materials week 1990, Detroit, ASM International, 1990, pp. 99–105.
7. T. Kannengiesser, M. Wolf and H. Schobbert, “Recent developments in nickel
base material welding considering the influence of shielding gas on the hot
cracking resistance”, IIW Doc. II-1554–05.
8. M.J. Cieslak, T.J. Headley and A.D. Jr. Romig, “The welding metallurgy of
HASTELLOY Alloys C-4, C-22 and C-276”, Met. Trans. A, 1986, 17,
pp. 2035–2047.
9. M.J. Cieslak, “The welding and solidification metallurgy of Alloy 625”,
Welding Journal, 1991, 70, pp. 49s–56s.
10. J.N. DuPont, “Solidification of an Alloy 625 weld overlay”, Metallurgical and
Materials Transactions A, 1996, 27A, pp. 3612–3620.
11. M. Raghavan, B.J. Berkowitz and J.C. Scanlon, “Electron microscopic
analysis of heterogeneous precipitates in Hastelloy C-276”, Met. Trans. A,
1982, 13A, pp. 979–984.
12. U. Heubner and M. Köhler, “Das Zeit-Temperatur-Ausscheidungs- und das
Zeit-Temperatur-Sensibilisierungs-Verhalten von hochkorrosionsbeständigen
Nickel.Chrom-Molybdän-Legierungen”, Werkstoffe und Korrosion, 1992, 43
(5) pp. 181–190.
Assessment of HAZ Hot Cracking
in a High Nitrogen Stainless Steel

K. Stelling, M. Lammers, D. Meinel

Federal Institute for Materials Research and Testing, Berlin, Germany

Abstract

Fully austenitic materials, such as nickel base alloys and several stainless
steels, can be prone to hot cracking in the heat affected zone during
welding. Such cracks, due to their small size commonly denoted as micro-
cracks, were found in laser hybrid welds of a high nitrogen stainless steel
(UNS S 34565 / German material no. 1.4565).
Mechanical testing, such as tensile and Charpy V-notch tests as well as
cyclic loading, has been carried out on hot crack afflicted laser plasma
hybrid welds. The test results show that the ductility of the material is
decreased, whereas the strength is not influenced at all. SEM micrographs
of the fracture surfaces revealed that micro-cracks in the HAZ were
associated with failure.
For a thorough evaluation of the cracking mechanism, the orientation of
the micro-cracks alongside the fusion line of the hybrid welds is of great
importance. Thus, computer tomography using micro focus X-ray has been
carried out. Using this new approach, the orientation of the cracks towards
the fusion boundary was ascertained. It turned out that the critical strain
emerged at specific sites of the hybrid weld. The critical strain was
oriented tangentially to the fusion boundary and perpendicularly to the
welding direction. X-ray cone beam computer tomography turned out as a
very useful tool for the investigation of the crack distribution and hence,
for getting qualitative information about the critical stress/strain
distribution in real welds.

Introduction

Fully austenitic stainless steels offer high performance for many

applications, such as offshore technology, chemical industries and
wherever high corrosion resistance and/or special mechanical, e.g. high
350 K. Stelling et al.

ductility at cryogenic temperatures, or special physical properties, e.g. low

magnetic permeance, are required. In these steels, nitrogen as an alloying
element can effectively reduce the material costs if it is used as a substitute
for nickel. Thus, the nitrogen alloyed stainless steels represent a very cost-
saving option compared to nickel base alloys. Furthermore, nitrogen
additions strongly increase the yield strength of these materials [1, 2].
However, fully austenitic materials can suffer from high sensitivity to
hot cracking, i.e. solidification cracking, liquation cracking and ductility-
dip cracking (DDC) [3-6].
Hot cracking occurs during welding if a critical strain is imposed on an
interface within a so called brittle temperature range (BTR) or ductility dip
temperature range (DTR) in case the of DDC [5, 7].
In addition to solidification cracking, cracking within the BTR implies
also material separation in the partially melted zone. This type of cracking
is denoted as liquation cracking because it arises from low melting phases
on the grain boundaries. It is the temperature as a function of time, i.e. heat
input, which controls diffusion and thus dissolution and/or precipitation of
low melting phases. If the peak temperature enters the BTR, then the
material cannot sustain critical localised strain during the cooling cycle
which results in hot tearing [8, 9].
Different mechanisms for liquation cracking are described in literature.
[11, 12, 13]
In contrast to that, DDC occurs below the solidus temperature and thus,
is rather a solid-state cracking mechanism. DDC sensibility is increased if
long, straight grain boundaries prevail which have low resistance to grain
boundary sliding [7].
It is clear that both, mechanical and metallurgical factors must be
present to generate hot cracking. The welding process, the welding
parameters and the thermal properties of the material govern the thermal
cycle, the weld pool shape, and hence, the thermo-mechanical loading of
the material adjacent to the weld pool. Another important factor regarding
mechanical loading would be the internal or external restraint.
However, the entire mechanism controlling crack formation in the heat
affected zone (HAZ) is still not completely understood. Besides this, it is
not definitely known whether these micro-cracks can have an unfavorable
effect on the in-service behavior of the weld. Tensile tests carried out on a
micro-fissured austenitic weld metal showed that micro-cracks influence
especially the ductility of the weld metal [14].
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 351

Furthermore, it is not clear how liquation cracking can successfully be

avoided by adapting the welding parameters. It is commonly recommended
to lower the heat input. But in this context, the weld pool shape is rarely
taken into consideration.
However, it is clear that the weld pool shape has a strong influence on
the stress/strain pattern [8, 12].
For example, a necked weld pool shape can have a detrimental effect on
the hot cracking behaviour in the HAZ of fully austenitic materials. This
weld pool shape is typically found in laser and laser arc hybrid welds.
Some authors already started to use numerical simulation of temperature
and strain distribution in the partially melted zone in order to quantify
critical conditions for liquation cracking in mere laser welds [15].
The term “hybrid welding” describes the combination of two welding
processes in a common process zone. A laser welding process is coupled
with arc welding processes in order to profit by the special attributes of the
single processes. This means that higher welding speed can be achieved
compared to mere arc welding due to the deep penetration of the laser.
Compared to mere laser welding, gap bridging is improved because of the
broader weld pool and the addition of filler metal. During laser arc hybrid
welding of fully austenitic materials it was often observed that micro-
cracks occurred along the fusion line, obviously depending on the
characteristic weld geometry.
However, in order to get a deeper insight into the mechanism of hot
cracking during welding it would be important to know where critical
conditions prevail in real welds. Additionally, it has to be established how
the cracks are oriented relative to the weld pool because that would
indicate in which direction critical strain/stress act during welding.
In this study, computer tomography with micro focus X-ray has been
used for obtaining more information about the distribution and the
orientation of the liquation cracks in a laser plasma hybrid weld of a high
nitrogen stainless steel.
Additionally, tensile testing, Charpy V-notch testing and cyclic tensile
loading has been used for assessing the mechanical behaviour of liquation
crack afflicted welds. Subsequently, scanning electron microscopy (SEM)
was used for examination of the fracture surface after testing in order to
make clear whether these cracks were involved in specimen failure.
352 K. Stelling et al.

Welding Experiments

Laser plasma hybrid welds of high nitrogen stainless steel plates of 6.6 mm
plate thickness were produced for the subsequent examinations. The
composition of the base metal UNS S 34565 (German material no. 1.4565)
and the filler powder are summarised in Table 1.
Hybrid welding has been carried out using a diode pumped Nd:YAG
laser. A scheme of the arrangement of laser beam and plasma torch can be
seen in Fig. 1. The parameters for the welds are summarised in Table 2.

Table 1. Chemical composition of the UNS S 34565/1.4565 stainless steel plate

and the filler powder, % by weight
Material C Cr Ni Mn Mo Si
UNS S 34565
0.026 23.7 18.66 6.084 5.565 0.162
6.6 mm plate

filler powder 0.015 27.25 22.4 7.75 4.45 0.53

S P N Nb others
UNS S 34565
<.001 0.016 0.466 0.010 Al: 0.024
6.6 mm plate
W: 0.34
filler powder 0.004 0.003 0.340 –
Cu: 1.01

Fig. 1. Arrangement of laser beam and plasma torch in the hybrid welding process
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 353

Table 2. Welding parameters

Tensile test specimen Charpy test specimen
Current 270 A 230 A
Voltage 23 V 20 V
Laser power 4.3 kW 4.2 kW
Welding speed 1.4 m/min 1.5 m/min
Nominal heat input 4.5 kJ/cm 3.5 kJ/cm
Focal point position 2.5 mm 3.0 mm
under the plate surface under the plate surface
Plasma gas Argon Argon

Microscopic Examination

The cross sections of the welds are shown in Figs. 2a and 3a. The shape of
the hybrid weld is typically divided into two parts. The broader upper part
results from the plasma arc, whereas the lower part is tapered by the deep
penetration of the laser. The particular shape of the weld pool is influenced
by the applied welding parameters, such as arc current, focal point position
and plasma gas flow. In hybrid welding, these parameters can be varied
within a comparatively wide range, thus, a variety of weld pool geometries
can be produced.

Fig. 2. (a) Cross section of the laser plasma hybrid weld used for the tensile test
and cyclic tensile loading, (b) hot cracks in the HAZ of the base metal
354 K. Stelling et al.

Fig. 3. (a) Cross section of the laser plasma hybrid weld used for the Charpy test,
(b) hot cracks in the HAZ of the base metal

However, the microscopic examination revealed that hot cracks

developed in the base metal adjacent to the fusion line, see Fig. 2b and
Fig. 3b. The length of the cracks is about 200 μm, and thus in the
dimension of the grain size. These intergranular cracks typically emerged
in the plasma part of the weld, close to the transition area towards the laser
part. The crack distribution was found to be asymmetric in welds with an
asymmetric shape. This presumably results from different stress/strain
patterns which depend on the orientation and/or the curvature of the fusion
boundary. It could also be seen from the microscopic examination that
some of the hot cracks have been in contact with the molten pool, Fig. 2b,
others seem to emerge in a distance from the fusion line, Fig. 3b. The
cracks appear to be liquation cracks and/or ductility-dip cracks resulting
from critical thermo-mechanical strain within a BTR and DTR,
respectively.
In the next step, mechanical testing was carried out to determine
whether these micro-cracks have a detrimental effect on the mechanical
properties of the weld.
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 355

Mechanical Testing

Tensile Test

Tensile test specimens were prepared according to Fig. 4. The specimen

surface has been ground until a uniform specimen thickness was reached in
order to avoid notch effects.
For comparison, base metal specimens and weld metal specimens were
fabricated. The loading direction was perpendicular to the rolling direction
of the plate and also to the welding direction regarding the weld metal
specimens (transverse tensile test). The tensile test was performed at room
temperature according to DIN EN 10 002. [16]

Fig. 4. Specimen geometry for the tensile test, loading direction was perpendicular
to the welding direction

Figure 5 shows the average values for the base metal specimens and the
weld metal specimens. The measured 0.2%-proof stress ranged from 400
to 450 MPa and the tensile strength ranged from 830 to 860 MPa,
respectively. There was no significant difference in the mechanical
strength between base metal and welded specimen.
In contrast, the ductility, represented by the reduction of area,
appreciably decreased from 65% in the base metal specimens to 40% in
both welded specimens. It has to be pointed out that the absolute value for
the ductility is still high. However, the welded specimens failed along the
fusion line and that would actually be a disqualifying criterion for quality
assurance.
356 K. Stelling et al.

Figure 6 shows the fractured weld metal tensile test specimen. From
Fig. 6b which shows the base metal side of the fracture, it is obvious that
the fracture surface has two different appearances. The upper part of the
fracture surface shows shiny sites which may indicate brittle fracture,
whereas the lower part shows ductile fracture. From the SEM micrographs
in Fig. 7, the different surfaces, i.e. the flat appearance of the shiny fracture
areas which might indicate portions of brittle fracture (a) and the dimple
structure of the ductile fracture (b), become apparent.

Fig. 5. Results of the tensile test, weld metal specimens fractured along the fusion
line

Fig. 6. (a) Fractured tensile specimen, (b) fracture surface of (a), base metal side
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 357

Fig. 7. SEM micrographs of the surface in Fig. 9b, (a) fracture surface within the
upper part, (b) ductile (dimple) fracture in the lower part

Charpy Test

The ductility of materials can be assessed by the Charpy V-notch test.

Normally, face centred cubic (fcc) metals such as austenitic steels provide
high ductility, especially at cryogenic temperatures. However, weld
imperfections can reduce low temperature toughness. Thus, the notch
impact resistance, i.e. the absorbed energy measured in the Charpy test,
was used to evaluate the influence of micro-cracks on the fracture
toughness. Preparation of the V-notch specimens has been carried out
according to DIN EN 10045 [17].
In Fig. 8a, the specimen geometry and the position of the notch in
relation to the weld metal zone are shown. At least three specimens were
tested at three different temperatures, i.e. RT, –60°C and –120°C.
From the diagram in Fig. 9 it becomes obvious that the toughness, i.e.
the absorbed energy in the Charpy test, decreased towards low test
temperatures. The examination of a specimen tested at –120°C revealed
that the crack proceeded along the fusion line in the upper part of the weld,
i.e. the plasma part.
The path along which the specimens fractured is marked in the cross
section of the weld in Fig. 8b.
In order to ascertain the reason for the toughness drop at low
temperature, fractographic examination of both fracture surfaces of a V-
notch specimen has been carried out.
358 K. Stelling et al.

Fig. 8. (a) V-notch specimen according to DIN EN 10045 and (b) estimated
fracture line during testing at –120°C

Fig. 9. Absorbed energy of the test specimen at different temperatures

 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 359

Fig. 10. Survey of the fracture surface of the V-notch specimen which cracked
partially along the fusion line, on the left: weld metal side, on the right: base metal
side

Figure 10 shows the corresponding fracture surfaces of a specimen

tested at -120°C. Since the specimen fractured along the fusion line in the
upper part (plasma part of the weld), the two halves of the specimen
represent the weld metal side (on the left) and the base metal side (on the
right) of the fusion line region.
SEM examination revealed that the major portion of the surface has a
dimple structure which is typical of ductile fracture. However, at the top
side of the specimen, i.e. in the plasma part of the weld, sites could be
found which indicate pre-existing intercrystalline cracks in the specimen.
These cracks are supposed to be hot cracks that had already been observed
in microscopic examination. The appearance of these cracks can be seen in
detail in Figs. 11 and 12.
Due to the fact that hot cracking within BTR involves surface liquation,
it can normally be distinguished from DDC by the freely solidified surface
structure, whereas the structure of DDC usually shows sharper ridges and
also reveals a slight surface roughness at higher magnification. [18]
Considering this, just some of these cracks can clearly be identified as
liquation cracks. A freely solidified surface inside the crack can be seen in
the detail of section no. 2. This kind of surface structure was observed
especially on the weld metal side of the specimen.
Other cracks show a more even structure which seems neither
completely freely solidified nor completely cracked in the solid-state, see
details of sections no. 1 and 3. This structure is not typical of liquation
cracks. It more resembles the surface of a ductility-dip crack.
Concluding from the different surface structures of the hot-cracks, i.e.
freely solidified surface and also smooth surface with just slight
indications of melted material, cracking presumably occurred at different
stages/temperatures within the BTR and/or DTR. The degree of liquation
360 K. Stelling et al.

and thus, the surface appearance of the cracks obviously depends on the
distance from the fusion zone.

Fig. 11. SEM pictures, section sites are marked in Fig. 10, weld metal side

Fig. 12. SEM pictures, section site is marked in Fig. 10, base metal side
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 361

The above described hot cracks are surrounded by an area which has a
flat, transcrystalline appearance without pronounced dimple structure. This
surface structure supposedly results from a complex stress state which was
induced by the micro-cracks during the impact test.
Therefore, at least four different surface structures can be distinguished:

1. transcrystalline fracture with dimple structure (major part)

2. transcrystalline fracture without dimples, probably due to a
complex stress state around the hot cracks,
3. intercrystalline hot crack with a smooth surface structure
4. intercrystalline hot crack with a freely solidified surface

The first three of the different surface types are indicated in section no. 4,
Fig. 13. The fourth one was shown in detail of section no. 2, Fig. 11.

Fig. 13. SEM micrograph of section no. 4, demonstration of the different types of
fracture surfaces, section site is marked in Fig. 10, base metal side

The SEM examination revealed that the reduced ductility of the weld at
lower temperatures can be attributed to the micro-cracks along the fusion
line. Although ductile dimple fracture represents the major part of the
fracture surface, the micro-cracks were obviously involved in failure and
governed the fracture path during testing, especially at low temperatures.
362 K. Stelling et al.

Cyclic Tensile Loading

Fatigue failure can be induced by surface imperfections such as notches or

cracks. Fatigue represents the most dangerous failure mechanisms due to
the often unnoticed progress.
In a very first attempt, the hot crack afflicted hybrid welds were
subjected to cyclic loading using oscillating tensile stress in order to see
whether it is possible to provoke fatigue failure starting from a micro-
crack.
The specimen geometry according to Fig. 4 was the same geometry as
for the tensile test. Here too, the specimen surface was ground until a plane
specimen surface was reached without any surface notches visible with the
naked eye.
Three specimens were tested under the following conditions: Oscillating
tensile stress with a peak stress of 300 MPa (Vmin = 0 MPa or stress ratio
R = 0) was imposed on the specimen. It has to be emphasised that the
applied maximum stress was about 75% of the yield strength which means
a high impact for the material. The oscillation frequency was chosen to be
30 Hz. If the specimen did not fail after 2u106 cycles, then testing was
terminated.
Just one of the three specimens failed before reaching the maximum
number of cycles, see Table 3.

Table 3. Results of cyclic tensile loading

Maximum stress Number of
Specimen no. After test
(i.e. stress amplitude), MPa load cycles
1 300 2.00 u 106 no failure

2 300 2.00 u 106 no failure

cracking, initiation
3 300 0.73 u 106
at hot crack

The major part of the fracture surface was at a right angle to the
direction of the tensile stress which indicates fatigue failure. In Fig. 14b
and Fig. 15, the portion of fatigue fracture and of final catastrophic failure
characterised by strong plastic deformation can clearly be distinguished.
An examination with the naked eye revealed that cracking initiated on the
top side of the specimen surface close to the fusion line. The macroscopic
fracture propagation direction is indicated by directional markings on the
fracture surface.
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 363

Fig. 14. Macrographs of the failed test specimen, (a) side view, position of the
weld is distinguished by means of a characteristic surface structure of the
specimen after extensive plastic deformation, (b) top view

Fig. 15. Directional markings on the fracture surface of the failed specimen
indicate the crack initiation and the propagation direction of the fracture.

At higher magnification, the initiation site is identified as a micro-crack

which was opened towards the ground specimen surface. The micro-crack
obviously spans two to three grains. A single exposed grain becomes
clearly visible in the SEM micrograph in Fig. 16b. The surface of this
micro-crack reveals a smooth and even structure. Slip line formation
occurred on the free crack surface. The slip lines are oriented in a distinct
manner depending on the crystallographic orientation of the grain, see
Fig. 16c. Ductility dip cracks can have such surface appearance. So this
could be an indication for a DDC mechanism. However, it is also possible
that the slip lines originate from cyclic loading during testing.
In Fig. 17a, the transition area of the micro-crack towards the fatigue
fracture surface is shown. The transcrystalline fatigue fracture in the base
metal can be seen in Fig. 17b. Since the fracture has propagated on distinct
multiple fracture paths, fatigue striations with different orientations can be
364 K. Stelling et al.

observed on the fracture surface, i.e. they are oriented according to the
crystallographic planes of the underlying metal.
Resuming the results of the cyclic tensile loading tests it can be stated
that under extreme loading conditions, which have been applied to the
welded specimens, micro-hot cracks can lead to fatigue failure if they are
opened towards the surface.

Fig. 16. SEM micrographs showing the initiation site of the fracture at different
magnification, (a) survey, (b) surface of the intergranular micro-crack, (c) smooth
surface structure revealing slight slip line formation

Fig. 17. (a) transition area of the micro-crack and the transcrystalline fatigue
fracture surface, (b) transcrystalline fatigue fracture surface

X-ray Cone Beam Tomography

From the examination of the fractured specimens it was possible to get

information about the surface structures of the micro-cracks, and also
about the orientation towards the fracture surface. However, the orientation
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 365

of the cracks relative to the fusion line which would indicate the direction
of the critical strain during the welding process did not become clear either
from the SEM examination or from the two dimensional micrographs.
For that reason, a weld section was examined using three-dimensional
computer tomography. This technique is known to be a very useful tool for
revealing the distribution of imperfections inside a component.
The principle of three-dimensional or cone beam tomography (CT) is
shown in Fig. 18. The tomograph was developed at BAM in the
department for non-destructive testing.

Fig. 18. Principle of three-dimensional or cone beam tomography

During testing, the sample is turned in the cone beam of a micro focus
X-ray tube and manifold slices are measured using an area detector. With
the usage of magnification techniques, a spatial resolution of less than
10 μm could be reached depending on the object size.
In order to get highest resolution and quality of the CT images, the
specimen should have an almost cylindrical geometry and a low geometric
measure in the radiation direction.
For this reason, a small piece has been cut off from the weld. Then the
specimen was oriented such that the lowest specimen thickness was in
radiation direction, i.e. the turning axis of the specimen was perpendicular
to the welding direction as well as to the plate normal. For the analysis of
crack distribution, the cracks which can be identified in the tomography
due to their lower attenuation were manually marked in every single plane
of the CT in order to create a three-dimensional picture of the crack-
distribution.
366 K. Stelling et al.

Fig. 19. Three-dimensional drawing of the weld with embedded transparent CT

segment showing the distribution of the liquation cracks along the fusion
boundary of the weld

Figure 19 shows the final, transparent three-dimensional computer

tomography (cuboid) which was embedded in a schematic drawing of the
weld specimen. The cracks were marked with white color so that it
becomes visible how they are distributed across the fusion boundary. The
distribution as well as the size and orientation of the cracks are more
clearly displayed in Fig. 20.
The distribution of the cracks and their orientation in relation to the
fusion boundary becomes clear from the viewing direction perpendicularly
to the fusion boundary which is shown in Fig. 20b. This sight reveals that
the crack planes are oriented almost perpendicularly to the fusion line and
parallel to the welding direction. In other words, the critical strain acted
tangentially to the fusion boundary and perpendicularly to the welding
direction. Schematically, this is demonstrated in Fig. 20c. No cracks could
be detected in the lower, i.e. the laser part, of the weld.
The critical conditions obviously prevail in the broader, upper region of
the weld. Suggested reasons for that are:

1. This part of the weld undergoes heat accumulation because it is

the region of the highest heat input and because of the broad,
slightly and inwardly curved shape of the weld bead. Thus, the
partially melted zone or the zone of grain boundary brittleness is
broader than in the lower part of the weld.
2. The volume of the liquid and, thus, the solidification shrinkage as
well as further cooling shrinkage of the weld bead is increased
towards the upper part of the weld.
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 367

3. The ratio of the fusion boundary area and the underlying base
metal volume is highest in the inwardly curved regions of the
weld. If shrinkage starts during cooling tangential strains are
imposed on these regions.

However, inwardly curved regions are found in the lower part of the
weld, i.e. the laser part, too. Nevertheless, hot cracks in these regions are
rarely found. Therefore, it must be a combination of increased local
shrinkage strains and adverse temperature fields, i.e. adverse metallurgical
conditions, causing the typical crack distribution.

Fig. 20. (a) Crack distribution along the weld flank in the upper weld part and
(b) cracks visualized by computer CT with the viewing direction perpendicular to
the fusion boundary; (c) schematic illustration of the direction of critical strain
towards the fusion boundary
368 K. Stelling et al.

Conclusions
1. The high nitrogen stainless steel UNS S 34565 is susceptible to hot
cracking in the heat affected zone of the base metal during laser plasma
hybrid welding.
2. Cracking typically occurred in the upper part of the weld produced by
the trailing plasma process during hybrid welding.
3. Transverse tensile testing and Charpy V-notch testing of welded
specimens indicate a decrease in ductility. But the strength of the
material is obviously not affected.
4. SEM surface examination of fractured specimens revealed that hot
cracks were involved in failure. The intergranular hot cracks show
different surface structures. Some grains exhibited an almost dendritic
surface, whereas others revealed a more even but still very smooth
surface. Thus, it is concluded that cracking occurred at different stages
of liquation and/or grain boundary ductility depending on the distance
from the fusion line.
5. By means of X-ray computer tomography it was found out that the
crack planes were located perpendicularly to the fusion boundary and
parallel to the welding direction. It could be shown in this contribution
that computer tomography is a very useful tool for the evaluation of
critical regions for hot cracking inside the material.

References
1. Brooks JA, Thompson AW, Williams JC (1980) Weld Cracking of Austenitic
Stainless Steels. In: R. Kossowsky (ed) Physical Metallurgy of Metal Joining,
Warrendale, Pennsylvania, pp 117–136
2. Berns H, Gavriljuk VG (1999) High Nitrogen Steels. Springer, Berlin
Heidelberg New York
3. Lin W, Lippold JC, Baeslack WA III (1993) An Evaluation of Heat-Affected
Zone Liquation Cracking Susceptibility, Part I: Development of a Method for
Quantification. Welding Research Supplement, April:135s–153s
4. Savage WF, Lundin CD (1965) The Varestraint Test. Welding Journal
34(10):433s–442s
5. Nissley NE, Collings MG, Guaytima G, Lippold JC (2002) Development of
the Strain-to-Fracture Test for Evaluating Ductility-Dip Cracking in
Austenitic Stainless Steels and Ni-Base Alloys. Welding in the World:32–40
6. Lundin CD, Lee CH, Menon R, Osorio V (1988) Weldability Evaluation of
Modified 316 and 347 Austenitic Stainless Steels: Part I – Preliminary
Results. Welding Research Supplement February:35s–46s
 Assessment of HAZ Hot Cracking in a High Nitrogen Stainless Steel 369

7. Collins MG, Ramirez AJ, Lippold JC (2004) An Investigation of Ductility-

Dip Cracking in Nickel-Based Weld Metals – Part III. Welding Journal
February:39s–49s
8. Borland JC, Younger RN (1960) Some Aspects of Cracking in Welded Cr-Ni
Austenitic Steels. British Welding Journal January:22–59
9. Kou S (2003) Solidification and Liquation Cracking Issues in Welding. JOM
June:37–42
10. Kou S (1987) Welding Metallurgy. John Wiley and Sons, New York, pp
239–262
11. Radhakrishnan B, Thompson RG (1992) A Model for the Formation and
Solidification of Grain Boundary Liquid in the Heat-Affected Zone (HAZ) of
Welds. Metallurgical Transactions A 23A:1783–1799
12. Pepe JJ, Savage WF (1967) Effects of Constitutional Liquation in 18-Ni
Maraging Steel Weldments. Welding Journal 46(9): 411s–422s
13. Radhakrishnan B, Thompson RG (1991) Modeling of Microstructural
Evolution in the Weld Heat Affected Zone (HAZ). In: Cieslak MJ, Perepezko
JH, Kang S, Glicksman ME (eds) The Metal Science of Joining, TMS, Ohio,
pp 31–40
14. Cui Y, Lundin CD, Hariharan V (2006) Mechanical behaviour of austenitic
stainless steel weld metals with microfissures. Journal of Materials Processing
Technology 171:150–155
15. Nishimoto K, Woo I, Shirai M(2002) Analyses of temperature and strain
distributions in laser welds – Study of the weldability of Inconel 718 cast
alloy (Report 6). Welding International 16(4):284–292
16. DIN EN 10 002 (2001): Tensile Test – Part 1: Test Method (at Room
Temperature)
17. DIN EN 10 045 (1991): Charpy impact test – Part 1: Test Method
18. Matsuda F, Nakagawa H (1977) Some Fractographic Features of Various
Weld Cracking and Fracture Surfaces with Scanning Electron Microscope.
Transactions of JWRI 6(1):81–90
Crack Appearance in Hot Rolled Billets

S.T. Mandziej1, G. Krallics2

1
Advanced Materials Analysis, Enschede, The Netherlands
2
Budapest University of Technology and Economics, Budapest, Hungary

Abstract

Hot cracks may appear in metal alloys on heating or on cooling when the
tensile strains and related stresses, caused by thermal expansion or
shrinkage and usually enhanced by various restraints, cannot be
compensated by a local plastic deformation of an alloy. In general, these
metal alloys, which have large thermal expansion coefficients, are
susceptible to hot cracking during casting, welding or e.g. re-heating for
hot rolling. The metallurgical quality of an alloy affects its susceptibility to
hot cracking, in particular the chemical and microstructural
inhomogeneities influence this susceptibility. Additional plastic strains,
applied in hot forming, usually extend the cracks leading to damage. In
this work an attempt is made to describe the micro- and macro-
mechanisms of the cracking as well as microstructural and mechanical
factors assisting the damage, based on observations of continuously cast
ingots in which internal hot cracks and cavities were formed during initial
steps of hot rolling. Characteristic of these ingots was chemical
segregation resulting in differences of hot ductility in different parts of the
ingots. Thus during plastic deformation, due to interaction between the
microstructurally and mechanically different regions, cracks and cavities
did nucleate and grow. Discussed here are physical data necessary to
adequately describe the behaviour of material during deformation and
cracking, as well as physical simulation methods to gain these data. The
gained data and identification of the damage processes are then used for
computer modelling with an aim to determine critical conditions of
controlling the application process in order to avoid the cracking and to
assure the manufacture of sound products. The micro-scale conditions
characteristic of e.g. welds and the macro-scale situations typical for
rolling of steel billets are discussed.
372 S.T. Mandziej, G. Krallics

Introduction

During hot rolling after continuous casting of steel various cracks may
occur resulting from the size and shape of the product as well as from
temperature, strain and strain rate of the applied processing. Thus during
rolling of 410 mm diameter continuously cast ingots into square 200 mm
cross section billets, periodic large internal cavities (bursts) were observed.
They often appeared in a medium-carbon Cr-Mo construction steel,
however never in any plain carbon steel of similar carbon content,
although the casting and rolling conditions were identical.

Fig. 1. Drawing stress (Vxf) as a function of reduction (r%), with indicated burst-
sensitive zone [1]
 Crack Appearance in Hot Rolled Billets 373

Relative Reduction, ∈

Relative Thickness, hc /Rc

Fig. 2. Relation between cross-reduction and size of rolls as a factor affecting the
central bursting in bar rolling [2]

Fig. 3. Schematic of rolling with marked area where the pulling force is created
due to friction [3]

The central bursts have been characteristic of drawing processes run at

transient cross-reduction conditions, as schematically presented in Fig. 1
[1]. They were also reported in the case of hot rolling, at conditions like
these presented in Fig. 2 [2]. During rolling, the axial force pulling in the
direction of material’s flow is mainly due to the friction between rolls and
billet, Fig. 3 [3]. Accordingly, in the situation shown in Fig. 3, the burst in
rolling is expected to form rather on entry to the rolls than on exit. The
magnitude of this type of central internal crack is shown in Fig. 4; the
picture taken from the cross-section of 200 mm hot-rolled billet.
374 S.T. Mandziej, G. Krallics

Fig. 4. Cross-section of 200 mm billet with marked locations from which samples
were taken for hot tensile testing

What is shown in Fig. 4 appeared in the hot rolling process designed

according to the rules described in the professional literature, which has
been always valid for the plain carbon steels but not always for the
considered here low-alloy 34CrMo4 grade steel, containing 0.35%C-
0.81%Mn-0.24%Si-1.01%Cr-0.21%Mo-0.029%Al-0.0073%N. Certainly
additional factors operated, like e.g. segregation, which could not be
predicted from only chemical composition of the material and designed
technological parameters of its processing.

The Hot Ductile Fracture

Fracture in bulk deformation processing usually appears as ductile and

rarely as brittle fracture. Depending on temperature and strain rate, the
details of the ductile fracture mechanism vary. Figure 5 illustrates in a
schematic way the “classical” modes of ductile fracture obtained in tensile
tests over a wide range of strain rates and temperatures. At temperatures
below approximately one-half the melting point of given material, a typical
dimpled type of ductile fracture prevails. At very high temperatures, a
rupture type of fracture occurs in which the material recrystallises rapidly
and pulls down to point, with nearly 100% reduction in area. A
transgranular creep type of failure occurs at temperatures less than those
 Crack Appearance in Hot Rolled Billets 375

causing rupture. Intragranular voids form, grow, and coalesce into internal
cavities, which result in a fracture with a finite reduction in area.

Fig. 5. “Classic” modes of ductile fracture

A commonly used representation of possible fracture mechanisms is the

(Ashby) map, Fig. 6. Such a map shows the area of dominance in terms of
normalized stress versus normalized temperature for the dominant fracture
mechanism. The maps are constructed chiefly by using the best
mechanistic models of each fracture process.
Failure in deformation processing around one-half of the melting
temperature 0.5 TM, occurs by ductile mode I. The three stages of such
ductile fracture are shown schematically in Fig. 7. The first stage is void
initiation, which usually occurs at second-phase particles or inclusions.
Voids are initiated because particles do not deform, and this causes the
ductile matrix around the particles to deform more than average. This in
turn produces more strain hardening, thus creating higher stress in the
matrix near the particles. When the stress becomes sufficiently large, the
interface may separate or the particle may crack. As result, ductility is
strongly dependent on the size and density of the second-phase particles,
as shown in Fig. 8 [4]. The second stage of ductile fracture is void growth,
which is a strain-controlled process. Voids elongate as they grow, and the
ligaments of matrix between the voids became thinner. The final stage of
ductile fracture is the cavity formation by the separation of the ligaments
that link the growing voids.
376 S.T. Mandziej, G. Krallics

Fig. 6. Fracture mechanism map

Fig. 7. Three stages of ductile fracture: (a) Void initiation at hard particles (b) Void
growth (c) Void linking

The described fracturing by fibrous tearing (mode I) occurs by void

growth in the crack plane that is essentially normal to the tensile axis. In
mode II of the ductile failure the voids grow in arrays at an oblique angle
to the crack plane under the influence of shear strain. This type of shear-
band tearing can be observed on the surface of the cone in ductile cup-and-
cone tensile fracture. It commonly occurs in deformation processing in
which friction and/or geometric conditions procure inhomogeneous
deformation, leading to local shear bands. Localization of deformation in
these shear bands leads to an adiabatic temperature increase that causes
local softening.
Compressive stresses superimposed on tensile or shear stresses by the
deformation process can have a significant influence on closing small
cavities and crack or limiting their growth and thus enhancing
workability. Because of this important role of the stress state, it is not
possible to express workability in absolute terms. Workability depends
 Crack Appearance in Hot Rolled Billets 377

not only on material characteristics but also on process variables, such as

strain, strain rate, temperature, and stress state.

Fig. 8. Effect of volume fraction of second–phase particles on the ductility of

plain low-carbon steels [4]

Many deformation processes for metals and alloys are performed at

temperatures above 0.5 TM i.e. in the hot working range. Three principal
benefits accrue from hot working. First is the lower flow stress at higher
temperatures, which offers an economy method for size reduction of large
work pieces. Second is the fact that metals at high temperatures are
generally capable of achieving larger deformation strains without fracture
than at lower temperatures. However, for many alloy systems, the
temperature and strain rate must be appropriately chosen. Lastly, high-
temperature deformation assists homogenization of the ingot structure
(chemical segregation) and the closing of internal voids. These benefits
come at the cost of surface oxidation, poorer dimensional accuracy, and
the need to appropriately heat and cool the work piece. In general, the
mechanisms of hot working are rather complex and may vary considerably
from alloy to alloy.

Characteristic of Central Burst-Susceptible Steel

The material under consideration here in its initial form is a round ingot of
410 mm diameter made of 34CrMo4 grade steel, which has been
continuously cast in vertical direction with travel speed of 0.5 m/min at a
casting temperature 1390°C. Cutting off a 6 m long piece from the con-
cast ingot appeared after cooling to about 1050°C and transferring to re-
heating/soaking oven at inlet temperature of about 900°C. The exit
378 S.T. Mandziej, G. Krallics

temperature from this furnace to the rolling track was 1280–1290°C, and
rolling with the rate of about 5/s at the temperature of 1240°C into 200 mm
square billet. After this procedure, internal burst cavities were found by
NDT, and typical for these cavities was their periodicity in the length
direction; they were repeating after each 1.2–1.5 m.
To check for the susceptibility to liquation cracking/grain boundary
sensitization, strain-induced crack opening (SICO) tests were run on 10 mm
diameter samples after re-heating to nil strength temperature (NST). The
SICO test was developed by Dynamic Systems Inc. in USA [5], for
determining formability limits at the situations when high thermal
gradients occur. It comprises heating a rod-like sample mounted in “cold”
Gleeble’s jaws, so that in the middle of the sample’s span the uniformly
heated zone is formed. This central hot zone is then axially compressed to
form a bulge. The test has been dedicated to Gleeble physical simulators,
in which the use of high-rate electric resistance heating may cause a non
controlled overheat of anvil-work piece interface in a conventional
uniaxial compression test. The SICO procedure was in the first instance
applied here for determining the melting and solidification parameters. For
this 25 mm diameter bars were used and from them the melting (solidus)
temperature was found to be ~1390°C, the nil strength temperature
~1350°C and the nil ductility temperature ~1300°C. Then samples for the
10 mm SICO tests were taken from different locations on the cross section
of the 200 mm square billet:
Sample A - taken ~30 mm from the central burst cavity,
Sample B - taken ~15 mm from the central burst cavity,
Sample C - taken exactly at the central burst cavity.
These locations have been marked on the cross-section of the billet shown
in Fig. 4. The results obtained from SICO tests are given in Table 1.

Table 1. SICO tests for hot/reheat crack susceptibility

Peak Deformation Strain Strain Results:

Material temp temperature rate number of
(NST) cracks
Sample A 1345 1290 1.25 6/s 6s-cracks
Sample B 1345 1290 1.25 6/s No cracks
Sample C1 1345 1290 1.25 6/s 2s-cracks
Sample C2 1345 1290 1.25 10/s 4m-cracks +
17s-cracks
 Crack Appearance in Hot Rolled Billets 379

The highest susceptibility to the liquation cracking appeared on the

samples taken from the close vicinity of the burst cavity, especially for the
strain rate exceeding the parameters of the real process. Then additionally
cracks were found on sample taken at about 30 mm from the burst cavity,
i.e. from the transition region of globular to columnar primary crystals. No
crack sensitivity was found in the region of about 15 mm from the burst
cavity. Representative SICO samples from locations A and C are presented
in Figs. 9 and 10.

Fig. 9. SICO sample A showing a small Fig. 10. SICO sample C2 showing a
crack near to perimeter of the bulge chain of small cracks along perimeter

Following the above SICO results, additional samples for hot tensile
testing were taken from the as continuously cast 410 mm diameter ingot,
from the locations adequate to these which showed different hot crack
sensitivity in the SICO tests. These samples were marked as given in
Table 2.
In the hot tensile tests the nil strength temperature (NST) of 1345°C was
taken as the maximum of the heating cycle and from it the samples were
cooled down to the pulling temperatures. Two samples were taken from
the center of the ingot, location (1) where a small shrinkage cavity as well
as visible porosity appeared with dendritic crystals inside. The mechanical
results from these samples are of course irrelevant nevertheless the
samples were useful for observation of fracture mode.
380 S.T. Mandziej, G. Krallics

Table 2. Results of hot tensile tests

Sample Distance Temperature YS Strain RA remarks
number from [°C] [MPa] Hl [%]
centre reheat test
1 0 mm 1345 1240 43 0.37 16 s. cavity
1A 0 mm 1345 1200 81 0.56 62 porosity
2 30 mm 1330 1200 95 0.60 86
2A1 30 mm 1345 1240 85 0.61 88
2A2 30 mm 1345 1200 96 0.65 94
2A3 30 mm 1345 1240 79 0.54 83
3 60 mm 1330 1200 95 0.64 96
3A1 60 mm 1345 1240 76 0.55 75
3A2 60 mm 1345 1200 11 0.07 -8 brittle
3A3 60 mm 1345 1240 83 0.66 78

The results of the hot tensile testing on the sound samples showed that
at the temperature of 1240°C, i.e. equivalent to the application conditions,
the material from location (3) deformed slightly slower in the crosswise
direction than this from location (2), while their yield strengths were
comparable. As the procedure of checking for susceptibility to liquation
cracking comprises heating-up a sample with fixed Gleeble’s jaws that due
to thermal expansion causes slight bulging and increase of sample’s
diameter in the central uniformly heated zone, it appeared on one sample
from the location (3), that it begun already cracking on heating in a brittle
manner, failing during subsequent pulling at very small extension (Hl) and
with “negative” reduction in area. This odd behaviour was caused by a
liquid phase, which already on heating under the compression due to
thermal expansion spread along the grain boundaries as a semi-continuous
film. Except for this, all other samples tested at 1200°C showed high
ductility, i.e. the RA of more than 85%. The case of brittle failure and then
the scatter of results indicate the influence of segregation and/or porosity
on the results of the hot tensile testing.

Metallographic Examination

In the first instance the central burst cavities were examined. As the
internal defects were found in the ingots, one might suspect that the
formation of the burst cavity might be related to the porosity in the middle
of the ingot, however it is not. The walls of the burst cavity appeared to be
 Crack Appearance in Hot Rolled Billets 381

entirely of shear fracture type, Fig. 11, with numerous fibre-like “pillars”
breaking by tensile necking, Fig. 12.

Fig. 11. Shear-type separation surfaces Fig. 12. Fibre-like pillar with defined
dominating inside the burst cavity necking portion in the burst cavity

When a sample containing shrinkage porosity is taken from the middle

of the ingot and then hot tensile tested, in the cavity extending during the
test numerous dendrites remain intact, while the walls of the cavity show
the tensile shear mode, Fig. 13. The same type of shear can be observed
on the outer surface of hot tensile samples, Fig. 14, and almost exactly
the same shear-slip features can be found on all inner walls of the burst
cavity. In the last, due to tri-axiality of the generated strains, a number of
fibre-like pillars occur standing up from the main inner wall planes,
Figs. 15 and 16.

Fig. 13. Dendritic crystals in the cavity Fig. 14. Outer surface of the tensile
of hot tensile sample 1 sample 1 showing shear-slip lines
382 S.T. Mandziej, G. Krallics

Fig. 15. Shear-slip lines on the surface Fig. 16. A cross section through a
of the burst cavity in the 200 mm fibre-pillar protruding from wall of the
square billet burst cavity; nital etched - DF

Characteristic of these pillars in the investigated medium-carbon Cr-Mo

steel is their austenite grain size; it is much coarser than a few millimetres
below their roots, which indicates that they were formed already on the
entry to the rolls and this way could not accumulate the major amount of
strain exerted during the rolling. The following two pictures taken in a
dark-field mode (DF) show the austenite grain size in the pillar, Fig. 17,
and at the distance of ~10 mm below the pillar’s root, Fig. 18.

Fig. 17. Coarse prior austenite grains Fig. 18. Finer prior austenite grains at
in the fibre-like pillar; nital etched – about 10 mm below the pillar’s root;
dark field image nital etched - DF

On the shear-slip type walls of the burst cavity evidences can be found
that during their formation hard particles existed which assisted the
deformation and separation process of the fibres and the walls, as well as
that some fibres had different ductility than the others. In the subsequent
Fig. 19 this situation is shown. After a slight electro-polishing and etching
of such shear-slip surface, the particles, which left the sharp shear traces on
the surfaces, can be revealed, in the size and amount adequate to explain
the appearance of “scratches” on the burst cavity walls, Fig. 20.
 Crack Appearance in Hot Rolled Billets 383

Fig. 19. Shear-slip type of the burst Fig. 20. Fragment of polished and
cavity wall with sharp “scratches” etched surface of the burst cavity
caused by solid hard inclusion particles showing microstructure with inclusion

In general, as seen on the metallographic samples taken from the vicinity

of cavity of the 200 mm square billet, the material after deformation still
consists of coarser grain patches embedded in finer-grained surrounding,
Fig. 21, and the cracks seem to easier propagate through the coarser-grained
material while retard when enter the finer-grained regions. Near to such
cracks in the pearlitic-ferritic microstructure of the billet, finer than ferrite
grains bright particles could be found, Fig. 22.

Fig. 21. Region of coarser prior Fig. 22. Pearlitic-ferritic microstructure

austenite grains near to burst cavity in with bright AlN particles in ferrite grains
200 mm square billet; nital etched, and black spheroidal inclusions; nital
dark-field image etched

During examination of the grip portion of sample 3A2, which sample

was taken from the crack-sensitive zone of the ingot, and which sample
failed during the hot tensile test in a brittle manner i.e. at little elongation
and without necking, the AlN particles tracing boundaries of the primary
columnar and globular crystals were also found, Fig. 23, as well as some
porosity, Fig. 24.
384 S.T. Mandziej, G. Krallics

Fig. 23. A chain of AlN particles Fig. 24. Porosity near to dendrite
tracing primary dendrite boundary in boundary in as-cast microstructure of
the crack-sensitive zone (3) of the the ingot in zone (3); nital etched
ingot; nital etched

Discussion

The cavitation described here, in particular the formation of the central

burst cavity is evidently controlled by the large strains occurring in the
metal forming process. It is also evidently assisted by inhomogenity of the
primary microstructure i.e. segregation, as well as by the solidification or
reheat defects of porosity type. The columnar crystals in the as-solidified
structure of the continuously cast ingots are large – they reach the lengths
of up to 50 mm, and during their crystallization various contaminants are
moved with the crystallization front towards the center of the material.
Among the contaminants also the grain-refining elements and particles are
moved. Accordingly, the region of dendrites appears “cleaner” and coarse-
grained. After homogenizing annealing before hot rolling the intensity of
segregation becomes certainly reduced on micro-scale, nevertheless it does
not seem so as regards the zone segregation of the AlN and other
inclusions. In result, during rolling the cavitation occurs, which initiates in
the regions where higher density of these inclusions can be found.
In general, the material under consideration with its 0.013% P and
0.007% S is clean, most probably much cleaner than the materials, which
were available at the time when the first books and papers on hot ductility
and central bursting in drawing, extrusion and rolling were produced
(compare with standards in e.g. Metals Handbook 1961, permitting
0.040% P and 0.040% S for a similar grade of steel [6]). For sure the
investigated here material contained less sulfide inclusions than the ones
manufactured in 1960s. So the historical models of the ductile cracking, by
the formation of voids around inclusions, in particular the sulfides, and
 Crack Appearance in Hot Rolled Billets 385

then coalescence of these voids, not always can be applied without reserve.
The same remark may refer to the deformability maps.
Nevertheless, the cavitation during metal forming is still an existing
important phenomenon, which often causes premature failures of the
material during processing. It appears in a large number of steels, which
generate microscopic voids (or cavities) when subjected to large strains
under tensile loading. The formation of microscopic cavities often occurs
along the grain boundaries of primary i.e. as-cast structure during the
initial operations of high-temperature deformation. It is believed that the
cavitation may lead to premature failure at levels of deformation far less
than those at which the flow localization controlled failure would occur
[7]. For a given material, the extent of cavitation depends on the specific
deformation conditions like the amount of crosswise strain, strain rate and
temperature, as well as on the surface condition, which may affect the
friction between tool (rolls) and workpiece (billet) and this way the level
of tensile strains and dynamic stresses. An important requirement for
cavitation during flow under hot working is the presence of tensile stress.
On the other hand, under conditions of homogeneous compression,
cavitation is not observed. In fact, it is believed that the cavities, which
might be produced at the tensile flow situation, can be removed during
subsequent compressive flow. Due to this last belief, the metal-formers
often do not care too much about the fine porosities and cavities, which
may nucleate during the forming process, unless the coalescence of these
cavities leads to total failure of the product.
In the situation of billets containing the large central burst cavities, two
major factors should be highlighted. The first is the tensile stress/strain
situation that may occur on the entry of the ingot to the rolling stand and
which is assisted by the friction between the rolls and the oxidized surface of
the ingot. The oxide scale on the investigated Cr-Mo steel does not easily
strip-off as well as it limits the possibility of sliding between billet and rolls,
thus contributing to high tensile forces. The second factor is the segregation
of inclusions in particular this of AlN. For the purpose of de-oxidation and
grain refinement the investigated steel contained almost 300 ppm of Al and
100 ppm of N. Concentration of these and other inclusion particles in certain
regions of the ingot structure resulted in differences of hot deformability, in
particular in capacity of material to flow with different rates at the same
applied external stress. This phenomenon is consistent with the formation of
the large central burst cavities having shear-type walls.
In this chapter the whole complexity of the problem was neither
addressed nor solved. Our aim was mainly to describe the phenomena,
which assist the formation of the cracks – the central burst cavities. It
appeared that the differences in local ductility under dynamic deformation
386 S.T. Mandziej, G. Krallics

conditions could create the situation, which gets out of control. The hot
deformation tests: SICO and hot tensile testing, gave results showing the
differences in local material properties, while metallography revealed the
effects: micro-structure and cracks, as well as indicated the distribution of
particles which might assist the problem.
In result, measures were taken at steelworks to eliminate the problem, in
particular by reducing the amount of AlN in the cast as well as introducing
stirring to reduce segregation in the ingots.

Conclusions

1. Large diameter continuously cast steel ingots have different hot

formability/ductility properties in different zones of their cross-section.
2. When hot rolling such materials comprising regions of different
ductility, internal cavities may form due to differences in local flow
rates of the material.
3. High density of grain-refining particles and inclusions evidently
strengthens the material and reduces its hot ductility.
4. Internal burst cavities initiate in the high strength, low ductility zones,
in which due to segregation of strengthening particles high probability
of voids formation around these particles occurs.
5. Dynamic tensile stresses of the hot rolling process, especially these
created due to friction between the rolls and the work piece/billet,
assist the formation of the central burst cavities.

References

1. Avitzur B, in: “Metal Forming – the Application of Limit Analysis”, Publ M

Dekker Inc, New York, 1980, pp. 108–109.
2. Turucz S, “The effect of the Roll-Gap Shape Factor on Internal Defects in
Rolling”, J Mater Process Technol, Vol. 60, 1996. pp. 275–282.
3. Backofen WA, “Deformation Processing”, Addison-Wesley Publ Co Inc, 1972.
4. Gladman T, Holmes B, and McIvor LD, “Effect of Second–Phase Particles on
the Mechanical Properties of Steel”, J Iron and Steel Institute, 1971, p. 78.
5. Ferguson HS; “Fundamentals of Physical Simulation”, in Proc Intl
Symposium on Physical Simulation, TU Delft 1992, pp. 1–21.
6. Metals Handbook 8th Edition, Lyman T (ed), Vol. 1, “Properties and
Selection of Metals”, ASM, Metals Park OH, 1961, p. 61.
7. Nicolaou PD, Semiatin SL and Lombard CM; “Simulation of the Hot-tension
test under Cavitating Conditions”, Metall Mater Trans A, Vol. 27, 1996, pp.
3112–3119.
 Part V
Ductility-Dip Cracking
Effect of Filler Metal La Additions
on Micro-Cracking in Multi-Pass Laser
Overlay Weld Metal of Alloy 690

K. Nishimoto1, K. Saida1, M. Sakamoto1, W. Kono2

1
Osaka University, Division of Materials & Manufacturing Science,
Osaka, Japan
2
Toshiba Co., Power & Industrial Systems R&D Centre, Japan

Abstract

The effect of La addition on microcracking susceptibility in the multipass

laser overlay welds of alloy 690 (Inconel 690) was investigated. Multipass
laser overlay welding of alloy 690 was conducted on alloy 132 (Inconel
132) in air (in-air welding) and under water (underwater welding) by using a
YAG laser with the power of 2 kW. Argon gas shielding was applied in both
cases with the gas flow rate of 20–60 L/min. Five kinds of filler metals,
which were varied in La content from 0 to 0.09 mass%, were used.
Microcracks in both weld metals produced by in-air and underwater
welding were categorised as ductility-dip cracks. Ductility-dip cracks
occurring in the weld metal could be completely prevented by using
0.01–0.03 mass% La containing filler metals in both in-air and underwater
welding. However, excessive La addition more than 0.04 mass% to the filler
metal adversely led to the occurrence of weld metal solidification/liquation
cracks also in both in-air and underwater welding. The improved
ductility-dip cracking susceptibility was attributed to the desegregation of
impurity elements of P and S to grain boundaries due to the scavenging
effect of La. Excessive La addition to the weld metal caused the formation
of a Ni-La intermetallic compound with low melting point, resulting in
solidification/liquation cracking. Thermo elasto-plastic analysis indicated
that the plastic strain-temperature curve slightly intersected the ductility-dip
temperature range (DTR) of the La-free weld metal in the cooling stage of
overlay welding, however, the plastic strain-temperature curves did not
cross the DTRs of the La-added weld metals. The analysed results were
fairly consistent with the experimental results that a few ductility-dip cracks
390 K. Nishimoto et al.

occurred in the La-free weld metal, while no ductility-dip cracks occurred in

the La-added weld metal of multipass laser overlay welds.

Introduction

The Ni-base superalloy, alloy 600 (Inconel 600), has been commonly used
in the reactor vessel, steam generator and pressuriser of nuclear power
plants (pressurised-water nuclear reactors). However, serious damage such
as the primary water stress corrosion cracking (PWSCC) has been found in
alloy 600 components [1]. Another Ni-base superalloy, alloy 690 (Inconel
690), is the material of choice for pressurised-water nuclear reactor
components because of its superior PWSCC resistance compared to alloy
600. It is planned that alloy 690 will replace the existing alloy 600 in ageing
nuclear power plants [2, 3]. In particular, the cladding of alloy 690 onto the
existing alloy 600 components and welds is one of the most practical
countermeasures for the preventive maintenance and/or repair of PWSCC
[4]. Recently, the laser cladding technique, especially when performed
underwater, has been developed and applied in practice for overlay welding
and seal welding of reactor components [5]. This technique possesses the
advantages of low heat input and reliable welding, compact welding
machine dimensions for welding in narrow spaces and a wide choice of
weld materials.
On the other hand, it has been pointed out that alloy 690 is highly
susceptible to hot cracking under heavy-restraint conditions such as the
welding of thick components [6]. Microcracking (microfissuring) in the
reheated weld metal during multipass welding can be of particular concern.
According to previous studies [7, 8, 9, 10, 11], microcracking in alloy 690
welds has been characterised as ductility-dip cracking. The authors have
reported that the primary cause of microcracking (ductility-dip cracking) in
alloy 690 was the reduction of hot ductility during cooling in welding
attributable to the grain boundary segregation of impurity elements such as
P and S [12, 13]. Furthermore, we have also proposed an effective method
for preventing microcracks in alloy 690 welds using La-containing alloy
690 filler metals, and confirmed that microcracks in multipass gas tungsten
arc (GTA) welds of alloy 690 could be completely prevented using the filler
metal with 0.01 mass% La added [14].
The objectives of the present study are to evaluate the microcracking
susceptibility in the multipass laser overlay weld metal of alloy 690, and to
 Effect of La Additions on Micro-Cracking 391

clarify the effect of La addition to the filler metal with the objective of
improving microcracking susceptibility of the weld metal. We have
conducted microstructural analysis of the weld metal to clarify the
mechanism by which La addition improves microcracking behaviour.
Furthermore, in order to comprehend the occurrence of ductility-dip crack
during laser overlay welding, we have made theoretical approaches,
employing numerical analysis, to model the behaviours such as the grain
boundary segregation of P, S and the development of thermal strain in laser
overlay welds.

Materials and Experimental Procedures

Materials

The base metal used in this study to simulate alloy 600 weld metal was alloy
132 (Inconel 132, DNiCrFe-1J) overlay weld metal (shielded metal arc
welding: SMAW, 4 layers), surface-ground to approx. 10 mm thick, on a 70
mm thick carbon steel, with the dimensions shown in Fig. 1. Five kinds of
filler metals of alloy 690/alloy 52 (Inconel 52, ERNiCrFe-7) in which the La
content varied between 0 and 0.07 mass% were used. The commercial filler
metal NF0 (La-free) was used as a standard of comparison for
microcracking susceptibility. The chemical compositions of the base metal
and filler metals are shown in Tables 1 and 2, respectively. The diameters of
the filler metals are 1.2 mm.

Alloy 132 overlay weld

~10mm

metal (SMAW, 4 layers)

m
0m
31mm

15

Carbon steel

60mm

Fig. 1. Schematic illustration and dimensions of the base metal

392 K. Nishimoto et al.

Table 1. Chemical composition of base metal alloy 132 (DNiCrFe-1J) used (mass%)
C Si Mn Ni Cr Cu Fe P* S* Nb+Ta
DNiCrFe-1J
(Alloy 132) 0.04 0.21 2.80 Bal. 14.43 0.02 8.88 40* 30* 1.86

* : ppm

Table 2. Chemical compositions of filler metals alloy 690/alloy 52 (ERNiCrFe-7)

used (mass%)
Filler metal C Si Mn Ni Cr Co Mo Ti Al Fe P* S* B* La
NF0 0.01 0.22 0.25 Bal. 29.74 0.02 0.02 0.11 0.10 10.31 50* 10* <10* –
NF1 0.01 0.22 0.25 Bal. 29.9 0.02 0.02 0.10 0.10 10.30 60* 10* <10* 0.01
NF2 0.01 0.21 0.25 Bal. 29.6 0.02 0.02 0.10 0.09 10.23 60* 10* <10* 0.03
NF3 0.01 0.22 0.25 Bal. 29.6 0.02 0.02 0.10 0.10 10.27 60* 10* <10* 0.04
NF4 0.01 0.22 0.26 Bal. 29.5 0.21 0.02 0.11 0.09 10.13 70* <10* <10* 0.07
* : ppm

Experimental Procedures

Figure 2 illustrates the set-ups for multipass laser overlay welding (3 layers,
10-9-8 passes) of alloy 690 on alloy 132 was carried out in air and under
water using a YAG laser, and the conditions are summarised in Table 3. The

Optical fiber

YAG laser (power : 4kW)

Optical fiber
Numerical
control
Shield gas (Ar)
YAG laser (power : 4kW) Wire feeder

Welding head Water

Laser optics system Water tank
Wire feeder

Shield gas (Ar)

Shield gas (Ar)
Specimen Work table

Specimen
Jig
Controller
Numerical control

(a) In-air welding (b) Underwater welding

Fig. 2. Schematic illustration of laser overlay welding apparatus

 Effect of La Additions on Micro-Cracking 393

Table 3. Laser overlay welding conditions used

Cladding condition A B C D E
Laser power (W) 1000 1000 1100 1200 1200
Laser travelling 5.0 11.7 8.3 5.0 11.7
speed (mm/s)
Wire feeding 10.0-18.3
speed (mm/s)
Shiled gas flow 20 (in–air)
rate (Ar, L/min) 60 (underwater)

welding conditions were varied within the following ranges: laser power of
1.0–1.2 kW, laser travelling speed of 5.0–11.7 mm/s, wire feeding speed of
10.0–18.3 mm/s and shield gas flow rates (Ar) of 20 L/min for in-air
welding and 60 L/min for underwater welding. The microcracking
susceptibility was evaluated as the total crack length in the cross-section of
the overlay weld metal. The weld metal was sectioned and prepared for
optical microscope and scanning electron microscope (SEM) observation by
grinding and then polishing with #400–#1500 grit emery paper. The final
polishing was performed with 1 μm alumina paste. Electrolytic etching was
performed with a 10% oxalic acid ethanol solution at 1.5–3.0 V for 25–30 s.
The element distribution in the weld metal was analysed by EPMA.

Microcracking Behaviour in Overlay Weld Metal

Characterisation of Microcracks

The cross-sectional views of the laser overlay weld metal with varying La
content, using Condition A in Table 3, are shown in Fig. 3. In the case of
using the La-free filler metal (NF0), a few microcracks occur in the weld
passes of the second and third weld layers where they were reheated by the
subsequent weld beads (designated as “under-bead cracking”). Several large
microcracks can be seen in the weld metal (NF4) containing the high
proportion of La (0.07 mass%), mainly in the final weld layer. However, no
microcracks can be observed in the weld metal in the intermediate case
using 0.01 mass% La added to the filler metal (NF1). Figure 4 shows the
microstructures of microcracks in the weld metals of NF0 and NF4. A
microcrack in the La-free weld metal (NF0) occurs at a small distance from
the fusion line and is propagated at the austenitic grain boundary (G.B.).
From the fact that the crack surface indicates smooth and intergranular
fracture without any trace of melting, the majority of microcracks in the
La-free weld metal can be characterised as being ductility-dip cracks. On
the other hand, a microcrack in the weld metal containing 0.07 mass% La
(NF4) originates at the fusion line and is propagated at the solidification
394 K. Nishimoto et al.

Laser overlay welding : Condition A

1mm

Crack

Filler metal : NF0 (La-free)

1mm

Filler metal : NF1 (0.01mass%La)

1mm

Crack
Filler metal : NF4 (0.07mass%La)

Fig. 3. Macrostructures of laser overlay weld metals with various La contents

boundary. The surface morphology of this crack indicates a dendritic

structure with a trace of a liquid film. It follows that the majority of
microcracks in the weld metal containing the larger amount of La can be
characterised as being solidification/liquation cracks.

Effect of La addition to Filler Metal on Microcracking Susceptibility

The effect of La addition to the filler metal on microcracking susceptibility

in the overlay weld metal is shown in Fig. 5. The total crack length was
evaluated for various combinations of the filler metal and welding condition.
The types of microcracking were not distinguished in the present evaluation,
however, the microcracking susceptibility in the La-free weld metal can be
 Effect of La Additions on Micro-Cracking 395

Cross section Crack surface

Fusion line
Filler metal : NF0

High temp.
(La-free)

Low temp.
20μm 100μm 5μm
G.B.

High temp.
Filler metal : NF4
(0.07mass%La)

Fusion line
Low temp.

50μm 100μm 5μm

Fig. 4. Microstructures of cracks in laser overlay weld metals

approximately regarded as ductility-dip cracking susceptibility and that in

La-added weld metal as solidification cracking susceptibility.
Microcracking susceptibility in the La-free weld metal is not very high and
microcracks occur for only a fraction of the welding conditions. No
microcracks occur in the weld metal using 0.01–0.03 mass% La added to the
filler metal (NF1 and NF2) at any welding condition including the weld
conditions where microcracks occurred in the La-free weld metal. However,
microcracking susceptibility increases again when the La content of the
filler metal exceeds 0.04 mass%, except for Conditions D and E. In the
present experimental welding conditions, microcracking susceptibilities at
low laser power (Conditions A and B) tend to be higher than those at high
laser power (Conditions D and E). Furthermore, there is not much difference
in microcracking susceptibility between underwater overlay welding and
in-air overlay welding.

Microcracking Mechanism in Overlay Weld Metal

Microstructural Analysis of Weld Metal with Added La

In order to elucidate the mechanism of change in microcracking

susceptibility with La addition, microstructural analyses of the overlay weld
metals were carried out. The preliminary SEM observation of the overlay
396 K. Nishimoto et al.

12
Total clack length (mm)

Condition A (underwater)
10
Condition B (underwater)
Condition C (underwater)
8
Condition D (underwater)
Condition E (underwater)
6
Condition C (in-air)
4

2
(0)(0)(0) (0)(0)(0)(0)(0)(0) (0)(0)(0)(0)(0)(0) (0)(0)(0)(0) (0)(0)
0
NF0 NF1 NF2 NF3 NF4
(La-free) (0.01mass%La) (0.03mass%La) (0.04mass%La) (0.07mass%La)

Filler metal
Fig. 5. Effects of La-addition and welding conditions on microcracking susceptibility in
laser overlay weld metal

weld metals containing La revealed the presence of a cellular structure in all

the weld metals; further, there was no significant difference in the
morphology and evenness of grain boundaries between the overlay weld
metals with different La contents [14]. The mapping of elements in the
overlay weld metals NF2 and NF4 analysed by EPMA is shown in Figs. 6
and 7, respectively. P, S and O tend to be found at the same locations as La
in the weld metal NF2 (arrows in Fig. 6), and La tends to be found also at
different locations from P and S in the weld metal NF4 (arrow in Fig. 7).
Taking these results together with the quite similar results reported in the
previous study [14], we could identify the products in the La-added weld
metals as La phosphide, sulphide and/or oxide in the La-added weld metal,
and a Ni-La intermetallic compound might also be present in the overlay
weld metal NF4.

Mechanism of Improvement in Microcracking Susceptibility

by La Addition

The improvement mechanism of microcracking in the alloy 690 overlay

weld metal by the La addition to the filler metal is discussed. By analogy
with the microcracking mechanism in the multipass GTA weld metal, we
expect the ductility-dip cracking in the overlay weld metal would also be
 Effect of La Additions on Micro-Cracking 397

caused by the reduction of hot ductility attributed to grain boundary

segregation of impurity elements such as P and S. From the fact that La
phosphide and sulphide are formed in the La-added weld metals,
microsegregation of P and S to a grain boundary would be depressed. In
other words, La scavenges P and S in the weld metal [14]. Consequently, we
deduce that ductility-dip cracking susceptibility would be improved by the
amelioration of hot ductility of the weld metal [12-14]. On the other hand,
excessive La-addition to the filler metal leads to the unwanted solidification
and/or liquation cracking in the overlay weld metal. Excessive La-addition
also leads to the formation of a Ni-La intermetallic compound in the weld
metal which has a low eutectic point compared to the Ni matrix [15].
Therefore, cracks would occur in the weld metal for an excessive La content
due to the enlargement of BTR (the brittle temperature range between
solidus and liquidus) and/or the occurrence of local liquation of a Ni-La
intermetallic compound during reheating [14].

EPMA analysis COMP

Laser overlay weld metal

Filler metal : NF2
(0.03mass%La)

2μm

La P

S O

Fig. 6. Element analysis by EPMA for La-added weld metal NF2

398 K. Nishimoto et al.

EPMA analysis COMP

Laser overlay weld metal

Filler metal : NF4
(0.07mass%La)

2μm

La P

S Ni

Fig. 7. Element analysis by EPMA for La-added weld metal NF4

Theoretical Approaches to Ductility-Dip Cracking

in Mutipass Laser Overlay Weld Metal

In order to better understand the conditions under which ductility-dip

cracking occurs during laser overlay welding, we have used numerical
analysis for theoretical approaches to such ductility-dip cracking behaviour
as grain boundary segregation of P, S and the development of thermal strain
in multipass laser overlay welds.

Grain Boundary Segregation in Weld Metal during Multipass Laser

Overlay Welding

In order to estimate grain boundary segregation of P and S during multipass

laser overlay welding (reheating) process, the theoretical analysis was
 Effect of La Additions on Micro-Cracking 399

conducted under the condition that the overlay weld metal was subjected to
welding thermal cycles generated by two successive overlay weld passes.

Theoretical Model of Microsegregation in Reheated Weld Metal

As described above, ductility-dip cracks in the alloy 690 overlay weld metal
would propagate preferentially along austenitic grain boundaries which
were coincident with solidification boundaries (dendrite boundaries) [12].
As a result of this conclusion, the theoretical model for microsegregation
used in the present simulation considered solidification segregation as
well as grain boundary segregation. The calculation procedures for
microsegregation are similar to those used previously for multipass GTA
welding [13, 16, 17]. Specially, segregation is assumed to happen during the
solidification stage of welding; subsequent segregation during the following
cooling/reheating stage occurs under the initial condition of non-uniform
distribution of segregating species formed during solidification segregation.
In order to simplify the computation procedure, segregation behaviour of P
and S to a grain boundary was calculated for pseudo-binary systems of
(Ni-30Cr-10Fe)-P and (Ni-30Cr-10Fe)-S using the finite differential
method (FDM) scheme. Furthermore, the co-segregation effect of P with S
was not considered in the present simulation. The model analysed here uses
one-dimensional diffusion in a rectangular triangle assuming that the
morphology of a dendrite cell is basically a hexagonal prism. Further details
of the theoretical models used in the present study can be found in Refs. [13,
16, 17]. The conditions used in the calculations are summarised in Table 4,
i.e., the initial P and S contents in the weld metal are 50 ppm and 16 ppm,
respectively, and the cooling and reheating rates were varied between four

Table 4. Calculation conditions for computer simulation of grain boundary segregation

of P and S

P 0.0050
Initial concentration
(mass%) S 0.0016
GTA welding 30(20) / 50
Laser overlay 60(40) / 100
Cooling rate / welding-1
reheating rate Laser overlay
(K/s) 150(100) / 250
welding-2
Laser overlay 300(200) / 500
welding-3
( ) : Cooling rate between liquidus and solidus
400 K. Nishimoto et al.

levels in order to simulate laser overlay welding and GTA welding. Material
constants used for the numerical simulation are also given in Ref. [13].

Microsegregation to Grain Boundary during Multipasss Welding

The computer simulation was carried out with the peak temperature in both
the second and third cycles being 1500 K. Figure 8 shows the relation
between the elapsed time in welding thermal cycles and the calculated P and
S concentrations at a grain boundary for cooling/reheating rates of 150/250
K/s. In the solidification process, P and S concentrations in the rest liquid
phase increase with the progress of solidification, resulting in remarkably
high amount of P and S bring segregated to a grain boundary at the
completion of solidification. However, P and S concentrations are
drastically reduced to their equilibrium concentrations when the weld metal
cools only by approx. 200 K. After that, they increase again with further
cooling (drop in temperature) from approx. 1400 K to 700 K, and they
saturate at the constant level below 700 K. In the reheating process, P and S
concentrations at a grain boundary keep constant temporarily or increase
slightly with a rise in temperature. However, they decline rapidly at
temperatures over approx. 1100 K. They once more increase again during
the cooling stage in the reheating process, and then change cyclically
thereafter in synch with the thermal cycle. The effects of different cooling

Cooling rate : 150K/s (100K/s)

Calculated P & S concentrations

Reheating rate : 250K/s

3.0 1800
Temp. S
at grain boundary (mass%)

1600
2.5 P
1400
Temperature (K)

2.0 1200
1000
1.5
800

1.0 600
400
0.5
200
0 0
0 5 10 15 20 25 30 35
Time (s)

Fig. 8. Calculated P and S concentrations at grain boundary during multipass

laser overlay welding thermal cycle
 Effect of La Additions on Micro-Cracking 401

and reheating rates on the P and S concentrations at a grain boundary at

1500 K, which represents the typical temperature range of ductility-dip
cracking, are shown in Fig. 9. The calculated P and S concentrations
increase slightly with increases in the cooling and reheating rates as well as
the number of weld passes. However, the values of the cooling and
reheating rates do not much affect grain boundary segregation for the range
of conditions used here. The approximate P and S concentrations at a grain
boundary attained during welding were respectively about 70 and 25 times
(approx. 3500 ppm and 390 ppm) as high as their initial concentrations. We
conclude that microsegregation of P and S to a grain boundary affects
microcracking susceptibility significantly, but grain boundary segregation
tends to be reduced at higher temperatures during the reheating thermal
cycle.

Thermal Strain Analysis during Multipass Laser Overlay Welding

It has been generally understood that hot cracking problems, such as

solidification and ductility-dip cracks occur when the strain-temperature

Initial P concentration : 0.0050mass%

Initial S concentration : 0.0016mass%
: P (2nd pass) : S (2nd pass)
: P (3rd pass) : S (3rd pass)
0.36 0.0410
grain boundary at 1500K (mass%)

grain boundary at 1500K (mass%)

C.R. : Cooling rate

H.R. : Reheating rate
Calculated P concentration at

Calculated S concentration at

0.35 0.0405

0.34 0.0400

0.33 0.0395

0.32 0.0390

0.31 0.0385

0.30 0.0380
GTAW LOW-1 LOW-2 LOW-3
C.R. : 30K/s C.R. : 60K/s C.R. : 150K/s C.R. : 300K/s
H.R. : 50K/s H.R. : 100K/s H.R. : 250K/s H.R. : 500K/s

Fig. 9. Effect of welding condition (cooling and reheating rates) on P and S

concentrations at grain boundary at 1500 K during multipass gas tungsten arc
welding (GTAW) and laser overlay welding (LOW)
402 K. Nishimoto et al.

curve crosses the brittle temperature ranges (solidification brittle

temperature range (BTR)/ductility-dip temperature range (DTR)) in the
welding process. In the present study, we have used thermal stress/strain
analysis to predict the occurrence of ductility-dip cracks in the multipass
laser overlay weld metal.

Model Details

The high temperature stress and strain in multipass welds were analysed by
two-dimensional thermo elasto-plastic FEM (calculation code: Quick
Therm). The model used for analysis of multipass laser overlay welds (3
layers, 6 passes) on alloy 690 is illustrated in Fig. 10. The deposited metal of
each weld pass modelled in the present analysis has the rectangular shape
with 1873 K (estimated temperature of the molten metal) dimensioned as 2
mm wide and 0.5 mm thick as illustrated in figures; the penetration of each
weld pass is consequently determined by the heat conduction from this
rectangular heat source. The position used for the analysis of stress/strain in
the multipass weld metal is also indicated in the bottom figure; it is the
reheated location in the first weld pass adjacent to the fusion boundaries of
the subsequent weld passes. The temperature dependencies of the various
material constants of the alloy 690 employed in the FEM analysis are
summarised in Fig. 11. The solidification shrinkage of the molten metal was
introduced as a volume change of 3%, and the melting temperature of alloy
690 was set at 1660 K. The interpass temperature used in multipass welding
used for the FEM analysis is 473 K.

Multipass laser overlay

weld metal
5mm

x Alloy 690
10mm

2mm
0.5mm

Analysed position Pass-6

Pass-4 Pass-5

Pass-1 Pass-2 Pass-3

Multipass laser overlay weld metal

Fig. 10. Analysed models for stress and strain in multipass laser overlay welds
 Effect of La Additions on Micro-Cracking 403

8
Specific heat (×10-3J/g•K)
7
Physical constants

5
Poisson's ratio (×10-1)
4

3
Thermal conductivity
2 (×10J/m•s•K)

Thermal expansion Young's

1 coefficient (×10-5/K) modulus
(×102GPa)
Yield stress (×102MPa)
0
273 773 1273 1773
Temperature (K)
Fig. 11. Temperature dependencies of material constants of alloy 690 used in FEM
analysis

Change in Temperature, Stress and Strain during Multipass

Welding Process

The calculated change in temperature, stress Vx (perpendicular to the weld

line/crack) and total strain (thermal strain + elastic strain + plastic strain) at
the analysed position during multipass laser overlay welding are shown in
Fig. 12. The peak temperature of subsequent weld passes decreases
gradually until the third weld pass, and then rises again before decreasing
after the fourth weld pass. The peak temperature is located in the
ductility-dip cracking temperature range (approx. 1630–1300 K) in the
second and fourth weld passes. The crack opening stress (Vx) changes
cyclically in synch with the weld thermal cycle and increases with number
of weld passes, i.e., it is compressive in the reheating stage, while rapidly
changing to tension in the cooling stage of welding. The total strain also
changes cyclically corresponding to the weld thermal cycle and tends to be
compressive.

Prediction of Ductility-Dip Cracking

The calculated results of the plastic strain at the analysed position in the first
weld bead perpendicular to the weld line/crack during laser overlay welding
404 K. Nishimoto et al.

2273
Temperature (K)

1773

1273

773

273
20
Stress, σx (MPa)

10

-10

-20
0.02
0.01
Total strain

0
-0.01
-0.02
-0.03
-0.04
-0.05
0 50 100 150 200 250 300 350 400
Elapsed time (s)

Fig. 12. Change in temperature, crack opening stress Vx and total strain during
multipass laser overlay welding

is shown in Fig. 13. In the present study, the plastic strain was expressed as
the increment in plastic strain [18] from the moment of solidification
(approx. 1660 K) of the first weld pass. A solid line indicates the increment
in plastic strain under the condition that the crack opening stress Vx is tensile
(corresponding nearly to the cooling stage) and a dashed line indicates that
the crack opening stress Vx is compressive (corresponding nearly to the
reheating stage). The plastic strain increases monotonically with cooling in
the first weld pass. However, in subsequent welds, once in the reheating
stage, it decreases, and then increases again in the cooling stage. It follows
that the plastic strain in the weld metal accumulates with increasing number
of weld passes. The relationship between the plastic strain-temperature
curve and the DTR, as a function of the La content in the weld metal are also
shown in Fig. 13. In the present study, the DTRs of the laser overlay weld
 Effect of La Additions on Micro-Cracking 405

BTR
0.7
Base metal : Alloy 690
0.6 Filler metal : NF0,NF1
DTR Laser overlay welding
Plastic strain (%)

(0.01mass%La)
0.5
DTR
(0.003mass%La)
0.4
DTR
0.3 (La-free) ng
Reheati ass
ng in in 4th p
Cooli ass
0.2 4th p
Cracking
s
in d pas
0.1 ling
Coo pass ating
in 2n 3rd pass
2nd Rehe 1 s tp a s s
Coolin
g in Crack-free
0
1673 1473 1273
Temperature (K)
Fig. 13. Relation between plastic strain-temperature curve and ductility-dip
temperature ranges (DTRs) in multipass laser overlay welding

metals with various La contents were assumed to be comparable to those of

the GTA weld metals determined by the Varestraint test [19]. The plastic
strain-temperature curve of the fourth weld pass slightly intersects the DTR
of the La-free weld metal in the cooling stage of overlay welding, i.e.,
ductility-dip cracking probably occurs. However, no weld passes have the
plastic strain-temperature curves that cross the DTRs of the weld metals
with added La. Assuming that ductility-dip cracking will occur when the
plastic stain-temperature curve intersects the DTR when the crack opening
stress Vx is tensile, these calculated results are reasonably consistent with the
experimental results that a few ductility-dip cracks occurred in the La-free
weld metal, while no ductility-dip cracks occurred in the La-added weld
metal of multipass laser overlay welds.
We conclude that ductility-dip cracking occurs during multipass laser
overlay welding when the plastic strain-temperature curve intersects the
DTR under tensile stress. The present prediction suggests that ductility-dip
cracking can be prevented by reducing the increment in plastic strain during
multipass welding.
406 K. Nishimoto et al.

Conclusions

In the present study, microcracking susceptibility in the multipass laser

overlay alloy 690 weld metal was evaluated, and the effect of La addition to
the filler metal was clarified in order to reduce such microcracking. The
results obtained may be summarised as follows:
(1) In the case of using the La-free filler metal, a few microcracks
occurred in the weld passes of the second and third weld layers. For the weld
metal with 0.07 mass% added La, several large microcracks occurred
mainly in the final weld layer. However, in the case of the filler metal with
0.01 mass% La added, no microcracks formed in the weld metal. The
majority of microcracks in the La-free weld metal could be characterised as
being ductility-dip cracks, and those in the weld metal containing the larger
amount of La could be characterised as being solidification/liquation cracks.
(2) Microcracking susceptibility in the La-free weld metal was not very
high and microcracks occurred for only some welding conditions. No
microcracks occurred in the weld metal with 0.01–0.03 mass% La added to
the filler metal under any welding conditions. However, microcracking
susceptibility increased again when the La content of the filler metal
exceeded 0.04 mass%. Furthermore, there was not much difference in
microcracking susceptibility between underwater overlay welding and
in-air overlay welding.
(3) The EPMA analysis revealed that P, S and O tended to be distributed
at the same locations as La in the weld metal NF2, and La also tended to be
distributed at different locations from P and S in the weld metal NF4. The
products in the La-added weld metals were identified as La phosphide,
sulphide and/or oxide, and a Ni-La intermetallic compound was additionally
identified in the overlay weld metal containing the larger amount of La.
(4) The ductility-dip cracking susceptibility was improved by the
amelioration of hot ductility of the weld metal due to the desegregation of
impurity elements of P and S to grain boundaries by the scavenging effect of
La. Excessive La addition resulted in the formation of a Ni-La intermetallic
compound, therefore, solidification and liquation cracks could occur in the
weld metal because of the enlargement of the BTR and/or the occurrence of
local liquation during the reheating process.
(5) Theoretical analysis of microsegregation of P and S to a grain
boundary during the multipass weld thermal cycle revealed that grain
boundary P and S concentrations in the ductility-dip cracking temperature
range respectively attained values about 70 and 25 times of their initial
concentrations. However, the cooling and reheating rates did not greatly
 Effect of La Additions on Micro-Cracking 407

affect grain boundary segregation of P and S for the range of conditions used
here.
(6) The occurrence of ductility-dip cracking in the multipass alloy 690
laser overlay weld metal was predicted by simulation using a mechanical
approach. The plastic strain-temperature curve in the La-free weld metal did
not cross the DTR in the first weld pass, however, it intersected slightly the
DTR during the cooling portion of subsequent welds. On the other hand, for
the weld metals containing La, none of the plastic strain-temperature curves
of any weld passes crossed the DTR. The laser overlay welding test revealed
that ductility-dip cracking would occur in the overlay weld metal when the
plastic strain-temperature curve intersected the DTR.

References

1. B.P. Miglin and G.J. Theus: “Stress Corrosion Cracking of Alloy 600 and 690
in All Volatile Treated Water at Elevated Temperature”, EPRI Report
NP-5761M, Electric Power Research Institute, Polo Alto, USA, (1988).
2. S.D. Strauss: “Inconel 690 is Alloy of Choice for Steam-Generator Tubing”,
Power, 140-2 (1996), p. 29–30.
3. J. Miyaguchi, K. Tabuchi and T. Yamamoto: “Towards Maintenance Service
Supporting Secure Nuclear Energy”, Mitsubishi Juko Giho, 40-2 (2003),
p. 92–95 (in Japanese).
4. Y. Nakamura, K. Dambayashi, H. Sakamoto, Y. Fujiya and Y. Tsukamoto:
“Field Application of the Cladding of PWR Reactor Vessel Outlet Nozzle”,
Mitsubishi Juko Giho, 43-1 (2006), p. 4–5 (in Japanese).
5. Y. Kanazawa and M. Tamura: “Underwater YAG Laser Welding Technique”,
Toshiba Review, 60-10 (2005), p. 36–39 (in Japanese).
6. W. Wu and C.H. Tsai: “Hot Cracking Susceptibility of Fillers 52 and 82 in
Alloy 690 Welding”, Metall. Mater. Trans., 30A-2, (1999), p. 417–426.
7. V.R. Davé, M.J. Cola, M. Kumar, A.J. Schwartz and G.N.A. Hussen: “Grain
Boundary Character in Alloy 690 and Ductility-Dip Cracking Susceptibility”,
W.J., 83-1, (2004), p. 1s–5s.
8. N.E. Nissley and J.C. Lippold: “Development of the Strain-to-Fracture Test”,
W.J., 82-12, (2003), p. 355s–364s.
9. M.G. Collins and J.C. Lippold: “An Investigation of Ductility Dip Cracking in
Nickel-Based Filler Materials - Part I”, W.J., 82-10, (2003), p. 288s–295s.
10. M.G. Collins, A.J. Ramirez and J.C. Lippold: “An Investigation of Ductility
Dip Cracking in Nickel-Based Filler Materials - Part II”, W.J., 82-12, (2003),
p. 348s–354s.
11. M.G. Collins, A.J. Ramirez and J.C. Lippold: “An Investigation of Ductility
Dip Cracking in Nickel-Based Filler Materials - Part III”, W.J., 83-2, (2004),
p. 39s–49s.
408 K. Nishimoto et al.

12. K. Nishimoto, K. Saida and H. Okauchi: “Microcracking Susceptibility in

Reheated Weld Metal, - Microcracking in Multipass Weld Metal of Alloy 690
(Part 1) -”, Sci. Technol. Welding & Joining, Vol. 11 No. 4 (2006), p. 455–461.
13. K. Nishimoto, K. Saida, H. Okauchi and K. Ohta: “Microcracking Mechanism
in Reheated Weld Metal, - Microcracking in Multipass Weld Metal of Alloy
690 (Part 2) -”, Sci. Technol. Welding & Joining, Vol. 11 No. 4 (2006),
p. 462–470.
14. K. Nishimoto, K. Saida, H. Okauchi and K. Ohta: “Prevention of
Microcracking in Reheated Weld Metal by La Addition to Filler Metal, -
Microcracking in Multipass Weld Metal of Alloy 690 (Part 3) -”, Sci. Technol.
Welding & Joining, Vol. 11 No. 4 (2006), p. 471–479.
15. K. Nishimoto, I. Woo and M. Shirai: “Underlying Mechanism in the
Improvement of HAZ Cracking Susceptibility by Rare Earth Metal Addition, -
Study on Weldability of Cast Alloy 718 (Report 8) -”, Quarterly J. JWS, 20-1,
(2002), p. 87–95 (in Japanese).
16. K. Nishimoto, H. Mori and H. Hirata: “Effect of Grain Boundary Segregation
of Sulphur on Reheat Cracking Susceptibility in Multipass Weld Metal of
Fe-36%Ni Alloy”, Proc. of the 7th Int. Symp. JWS, Kobe (2001), p. 827–838.
17. K. Nishimoto, H. Mori, K. Esaki, S. Hongoh and M. Shirai: “Effect of Sulphur
and Thermal Cycles on Reheat Cracking Susceptibility in Multipass Weld
Metal of Fe-36%Ni Alloy”, IIW Doc. IX-1934-99 (1999).
18. M. Shibahara, S. Sone, H. Serizawa, H. Murakawa, K. Nishimoto and K. Saida:
“FEM Analysis for Ductility-Dip Cracking Under Multi-Pass Welding”,
J. Kansai Soc. Naval Architects, Japan, 241 (2004), p. 177-186 (in Japanese).
19. K. Saida, M. Sakamoto and K. Nishimoto: “Mechanical Approach for
Prediction of Microcracking in Multipass Weld Metal of Ni-Base Alloy 690”,
International Welding/Joining Conference-Korea 2007, Seoul, Korea (2007),
p. 24–25.
Ductility-Dip Cracking in High Chromium,
Ni-Base Filler Metals

J.C. Lippold1, N.E. Nissley2

1
The Ohio State University Columbus, Ohio, USA
2
Exxon Mobil Upstream Research Houston, Texas, USA

Abstract

Ductility-dip cracking (DDC) is an elevated temperature, solid-state

cracking phenomenon that is observed in austenitic weld metals. In this
study, the DDC susceptibility of several high-chromium, nickel-base filler
metals was evaluated using the strain-to-fracture (STF) test technique.
These filler metals were of the Ni-30Cr type and included INCONEL®
Filler Metals 52 and 52M supplied by Special Metals Welding Products
Company, and Sanicro 68HP® and Sanicro 69HP® supplied by Sandvik
AB. In addition, two experimental Ni-30Cr filler metals were evaluated
which contained variations in other alloy additions, including niobium
additions up to 2.5 wt% and molybdenum additions of 4 wt%. A wide
range in DDC susceptibility was observed with these filler metals,
including large heat-to-heat variations in filler metals with similar
compositions. The experimental filler metals containing Nb and Mo were
found to be remarkably resistant to DDC. Cracking susceptibility is
primarily associated with the type and form of precipitate that forms along
the weld metal migrated grain boundaries. The formation of Nb-rich,
M(C,N) at the end of solidification has the most profound effect on DDC,
since these precipitates are most effective in pinning the boundaries. The
formation of M23C6 carbides during weld cooling or subsequent reheating
can also affect DDC susceptibility in these filler metals.

Introduction

Ductility-dip cracking (DDC) is an intermediate temperature, solid-state

grain boundary embrittlement phenomenon that has been observed in Ni-
base weld metals, particularly those with high-Cr levels conforming to
AWS A5.14, ERNiCrFe-7 and ERNiCrFe-7A. The precise mechanism for
410 J.C. Lippold, N.E. Nissley

DDC is unclear, although large grain size and long, straight grain
“migrated” grain boundaries in the weld metal have been shown to
increase susceptibility. The effect of filler metal composition and the
related nature of grain boundary precipitation phenomena are key elements
in the understanding of DDC, as recently described by Ramirez and
Lippold [1, 2]. The strain-to-fracture (STF) test has proved to be an
efficient and effective method for quantifying susceptibility to DDC [3]
and has been used here to compare a number of high-Cr, Ni-base filler
metals. This chapters is a continuation of the research that has been
reported previously at IIW annual assemblies [4, 5]. Readers are referred
to those chapter for the results of preliminary studies of the DDC
susceptibility of FM52 and FM52M. This chapter provides additional STF
data on other high-Cr, Ni-base filler metals and compares this data to the
FM52 and FM52M results reported previously.
Filler metals conforming to AWS A5.14, ERNiCrFe-7 (FM52, Sanicro
68HP) and ERNiCrFe-7A (FM52M) are frequently used in nuclear power
plant applications, often for welding Ni-base Alloy 690. These filler metals
have good resistance to stress-corrosion cracking and are used as a
replacement for lower-Cr, FM82 in environments where stress corrosion
cracking is a concern, such as in high-purity, primary water in nuclear
power plants. DDC has been observed when using these filler metals,
particularly in multipass welds made under medium to high restraint
conditions.

Experimental Procedures

Weld deposits of eight (8) different high-Cr filler metals were tested using
the strain-to-fracture (STF) test. This test was developed at The Ohio State
University for the evaluation of elevated temperature grain boundary
cracking in austenitic materials. Details of the test development and
procedures can be found in Ref. [3]. The compositions of the filler metals
tested in this investigation are provided in Table 1.
The FM52 and Sanicro 68HP filler metals conform to AWS
ERNiCrFe-7. The three FM52M filler metals were produced to conform to
ERNiCrFe-7A with Nb in the range from 0.5–1.0 wt%. Two Ni-30Cr
variants containing elemental variations and Nb and Mo (FM52X-D and
FM52X-H), and Sanicro 69HP with nominally 1.8 wt% Nb were also
tested.
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 411

The testing procedure used for the eight filler metals is essentially
identical to that used previously for FM52 and FM52M [5]. These
represent the FM52 and FM52M-A filler metals in Table 1. STF samples
were machined from multipass welds. An autogenous, gas tungsten arc
spot weld was made in the gage section of these samples under controlled
conditions to produce a radial pattern of grain boundaries in the spot weld.
Full temperature-strain plots were produced for the FM52, 52M, and
52X-H filler metals, and the Sanicro 68HP and 69HP filler metals. Filler
metals described as B, C, and D were only tested at 950qC. This
temperature was selected since it is normally the temperature that
represents the minimum threshold strain for cracking across the entire
DDC range and where the highest degree of cracking (number of cracks)
occurs as strain is increased in the range from 0–10%.

Table 1. Chemical composition (wt %) of filler metals

52 52M 52M 52M 52X 52X Sanicro Sanicro

Element
A B C D H 68HP 69HP
Ni 60.12 60.37 59.54 60.38 55.5 53.9 58.81 57.0
Cr 29.09 30.04 30.06 29.5 29.6 30.2 29.94 30.0
Fe 8.88 8.42 8.22 8.04 8.04 8.2 8.80 9.15
C 0.026 0.020 0.02 0.03 0.02 0.02 0.029 0.005
Nb 0.02 0.85 0.83 0.82 1.0 2.5 <0.01 1.83
Mn 0.25 0.81 0.80 0.77 0.72 0.8 0.90 1.21
Ti 0.50 0.21 0.22 0.19 0.2 0.2 0.68 0.39
Al 0.71 0.10 0.11 0.12 0.05 0.04 0.75 0.20
Si 0.17 0.03 0.09 0.12 0.03 0.2 0.13 0.11
Cu 0.011 0.02 0.02 0.01 <0.05 – <0.01 <0.01
Mo 0.05 0.02 – – 4.0 4.0 <0.01 <0.01
P 0.0044 0.004 0.003 0.004 0.015 0.004 <0.003 0.004
S 0.003 0.001 0.001 0.001 0.001 0.001 0.001 0.001
Co – 0.007 – – – – <0.005 <0.010
Zr – 0.015 0.01 0.008 0.006 0.007 – –
B – 0.004 0.003 0.001 0.004 0.001 0.0008 0.0014

The degree of DDC in a given sample is determined by counting the

number of cracks in the gage section of the sample using a stereo
microscope at magnifications up to 30X. Cracks that are only resolvable
above 30X are not included in the crack count. The number of cracks
reported for a given test sample represents the number of cracks per spot
weld. In samples where the spot welds on either side of the sample were of
the same composition, the total number of cracks in both welds was
divided by two.
412 J.C. Lippold, N.E. Nissley

Samples were prepared for metallographic analysis by electrolytic

etching with 10% chromic acid at 1.5-2 V for 30 s. Scanning electron
microscopy (SEM) and EDS analysis was performed on a Sirion
SEM/FEG and Phillips XL-30 ESEM/FEG at 15 kV.

Results and Discussion

Temperature-Strain Curves

STF test results for FM52 and FM52M that have been published
previously [5] are shown in Fig. 1. Each symbol represents a single test
with an “X” for a strain-temperature condition where cracking was not
observed and a circular symbol for samples where cracking occurred. The
number adjacent to each point represents the number of cracks that were
observed on the surface of the sample under a binocular microscope using
the procedure described previously. A curve separates samples that
cracked from those that did not and represents the minimum strain for
cracking as a function of temperature. The threshold strain (minimum
strain for cracking in the ductility-dip temperature range) was similar in
both materials, approximately 2% at 950°C. The temperature range where
the threshold strain occurred (900–1000qC) is consistent with other STF
testing, and the threshold strain is similar to results on the same heat of
IN52 previously reported by Collins et al. [6]. As a reference, the STF
curve previously developed by Collins et al. for Filler Metal 82 is
included.
Results of STF tests for Sanicro 68HP and 69HP are provided in Fig. 2.
These filler metals exhibit a slightly higher threshold strain (2.5%) than
FM52 and 52M and have less severe cracking at higher strains at 950qC.
No increase in the minimum strain for cracking could be identified at the
high temperature end of the DDC range (>1150qC), possibly due to the
overlap of the DDC and liquation cracking range.
STF results for the experimental 52X-H filler metal are presented in
Fig. 3. The threshold strain for cracking for this filler metal is considerably
higher than for the filler metals shown in Figs. 1 and 2. In the temperature
range from 750 to 950qC, 52X-H is extremely resistant to DDC as
indicated by both a high threshold strain for cracking and the small number
of cracks that occur above threshold.
In Fig. 4, The STF results for 52X-H are compared to STF data pre-
viously generated for Type 304L and Filler Metal 82 weld metals [6, 7, 8].
This shows that the DDC resistance of 52X-H is superior to FM-82 over
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 413

most of the DDC temperature range. FM-82 is relatively resistant to DDC,

except under very high restraint conditions. Austenitic stainless steel weld
metal that contains ferrite is known to be essentially immune to DDC.
Thus, the curve for Type 304L with Ferrite Number (FN) 6 represents STF
behavior characteristic of an extremely resistant weld metal. Based on
these comparative curves, it is anticipated that 52X-H weld metal will be
very resistant to DDC under actual fabrication conditions.

Comparison of DDC Susceptibility at 950qC

Development of a full STF curve for a single filler metal requires 30–40
samples and may not be appropriate for preliminary screening of filler
metal compositions. In this investigation, only filler metals 52, 52M-A,
52X-H, and Sanicro 68HP and 69HP were tested over the full DDC
temperature range. For filler metals 52M-B, 52M-C, and 52X-D, testing
was only conducted at 950qC. This temperature was selected because the
minimum in DDC behavior usually occurs at, or near, 950qC and that
cracking response at this temperature correlates well with actual
fabrication behavior.
A comparison of the eight filler metals at 950qC is provided in Fig. 5.
The horizontal bar in the column above each filler metal represents the
threshold strain for cracking for that material while the numbers represent
the number of cracks that were present in the sample as a function of
strain. All the FM52 and FM52M weld metals exhibit a threshold strain in
the range from 1–2%, while the Sanicro weld metals are at approximately
2.5%. There is a considerable difference in cracking susceptibility, based
on number of cracks at applied strains above 2%, for these filler metals.
As reported previously [5], FM52 and FM52M-A exhibit extensive
cracking above the threshold strain. FM52M-C exhibits similar behavior,
but FM52M-B does not show extensive cracking until the strain exceeds
6%. The Sanicro 68HP and 69HP weld metals do not show extensive
cracking until strains reach almost 8%.
The experimental weld metals containing Nb levels above 1 wt% and
4 wt% Mo exhibit both the highest threshold strains and the best resistance
to cracking above the threshold. FM52X-D has a threshold strain of 3%,
while that for FM52X-H is approximately 6%. It should be noted that all
the other filler metals showed extensive cracking above 6% strain, but the
experimental filler metals were very resistant to DDC at these strain levels.
 414

16%

14%

12%
FM-82

>100 >100
10%
J.C. Lippold, N.E. Nissley

>100 >100
8%
25 5

Strain (%)
>50 >50
6%
9 2
19 14 17 4
2 1 2
4% 3

FM-52M 7 5
2%

FM-52
0%
600 700 800 900 1000 1100 1200 1300
Temperature (°C)

Fig. 1. STF results for FM52 and FM52M. Numbers to the left of the open symbols are the number of cracks for the FM52 weld
metal, while the numbers to the right are for FM52M. A curve for FM-82 is included for comparison
 16.0

14.0

12.0
FM-82
10.0
69 HP 35 31
8.0
>50 >25
>25 >25
6.0 15 4

Calculated Strain %
7 5 >50
8 7 5 4 2
4.0 9 8
5 2 1 1 23 2 2 1
3 10 4
1 1
2.0

68 HP
0.0
600 700 800 900 1000 1100 1200 1300
Temperature (°C)

Fig. 2. STF results for Sanicro 68HP and 69HP filler metals. Numbers to the left of the open symbols are the number of cracks for
Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals

the 68HP weld metal, while the numbers to the right are for 69HP
415
416 J.C. Lippold, N.E. Nissley

18.00

16.00 Filler Metal 52X-H

(0)

14.00
(14)
12.00
Strain (%)

(2)
10.00
(0.5)
(0) (0.5)

8.00 (11) (1)

(0)
6.00 (0.5) (0) (2.5) (5)
(0.5)

4.00 (0.5)
(0)
2.00

0.00
600 700 800 900 1000 1100 1200
Temperature (°C)

Fig. 3. STF results for Filler Metal 52X-H. No data was collected above 1050qC,
so a hypothetical extension of the curve is indicated by the dashed line

18.00

16.00 (0)

14.00
52X-H (14)

12.00
Type 304 SS, FN6
Strain (%)

(2)
10.00
(0.5)
(0) (0.5)

8.00 (11) (1)

(0)
6.00 (0.5)
Filler Metal 82 (0) (2.5) (5)
(0.5)

4.00 (0.5)
(0)
2.00

0.00
600 700 800 900 1000 1100 1200
Temperature (°C)

Fig. 4. Comparison of STF results for Filler Metal 52X-H with those of Type
304L (Ferrite Number 6) and Filler Metal 82 weld metal
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 417

14

Strain-to-Fracture Test 14
12 Results at 950qC

10
Applied strain (%)

35 20
8 >100 >100
11

6 32 2
0 15 4
1 7 5
19 14 3
4
42 2 9 8
2 0
7 5
2 1 24 0 0 0
0 0
0 1
0
IN52 IN52M-A IN52M-B IN52M-C IN52X-D IN52X-H San-68HP San-69HP

Fig. 5. Strain-to-fracture test results for high-Cr, Ni-base weld metals tested at
950qC. Horizontal bars represent the threshold strain for cracking. The numbers
represent the individual cracks counted on the sample surface

Effect of Composition and Weld Metal Microstructure on DDC

Metallographic examination was conducted on STF samples for FM52,

FM52M-A, B, and C and FM52X-H. As has been reported previously [5],
the FM52 filler metal exhibited large grain size and relatively long straight
migrated grain boundaries. A representative microstructure of both FM52
and FM52M weld metals containing DDC along migrated grain
boundaries is shown in Fig. 6. As can be seen, the grain size is large and
the boundaries are quite straight. When examined using electron
microscopy techniques (SEM and TEM), the migrated grain boundaries
were found to be decorated with small, Cr-rich carbides of the M23C6 type.
An SEM micrograph of a typical boundary in FM52 is shown in Fig. 7 and
a higher magnification TEM image in Fig. 8. These carbides form in the
solid-state as the weld metal cools to room temperature (or during
reheating) and seem to have little or no effect on boundary pinning in the
FM52 and FM52M filler metals. The microstructure and morphology of
the grain boundaries and carbides in FM52 and are consistent with weld
metals that have a high susceptibility to DDC, i.e. virtually no pinning of
the grain boundaries.
418 J.C. Lippold, N.E. Nissley

Fig. 6. Weld metal microstructure in a FM52 STF sample tested at 950qC and
8.5% strain. Ductility dip cracks are evident along the weld metal migrated grain
boundaries

Fig. 7. SEM secondary electron image of a migrated grain boundary in FM52

showing small (<0.5 ȝm) M23C6 carbides along the boundary
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 419

Fig. 8. TEM micrograph of a migrated grain boundary in FM52. The precipitates

were identified as M23C6 and exhibited a cube-on-cube orientation relationship
with the austenite matrix (Courtesy of Dr. Antonio Ramirez)

An SEM image of the FM52X-H weld metal microstructure is shown

in Fig. 9. The fraction of Nb-rich M(C,N) precipitates is much higher in
this weld metal compared to the FM52M weld metals. These precipitates
form at the end of solidification and are effective at pinning the migrated
grain boundaries, resulting in more tortuous grain boundaries in this weld
metal.
Further evidence of grain boundary tortuosity can be obtained by using
orientation imaging microscopy (OIM™) in the SEM. Using this
technique which analyzes electron backscattered diffraction (EBSD)
patterns, the nature of the migrated grain boundaries can be clearly
differentiated. Patterns for FM52M-C and 52X-H are shown in Fig. 10.
Note that the FM52M-C boundaries are generally straight, while the
FM52X-H boundaries are extremely jagged and tortuous. This clearly
demonstrates the effect of Nb-rich, M(C,N) precipitates that form at the
end of solidification on increasing the tortuosity of the weld metal
migrated grain boundaries.
420 J.C. Lippold, N.E. Nissley

M23C6

M(C,N)

Fig. 9. Weld metal microstructure of FM52X-H, SEM secondary electron image.

M(C,N) precipitates are visible in the interdendritic regions and M23C6 carbides
were found along the migrated grain boundaries. Grain boundary pinning by the
large M(C,N) precipitates is evident

Fig. 10. Electron backscattered diffraction (EBSD) patterns showing the nature of
the weld metal migrated grain boundaries in FM52X-H (left) and FM52M-C
(right)
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 421

Mechanism of Ductility-Dip Cracking

The behavior of the high-Cr, Ni-base filler metals again shows the
importance of the nature and morphology of precipitates on resistance to
DDC. Nb additions above the level prescribed in ERNiCrFe-7A (0.5–1.0
wt%) lead to the formation of a higher fraction of Nb-rich M(C,N) at the
end of weld solidification. Because these precipitates are present at the end
of solidification they are very effective in pinning the migrated grain
boundaries that form from the original solidification grain boundaries. The
fraction of these solidification- related precipitates is apparently quite
important, since when the fraction is low, such as with FM52M-A and
FM52M-C, grain boundary pinning is minimal and DDC susceptibility is
high. Increasing Nb content above 1 wt% has a profound effect on DDC
resistance as evidenced by the behavior of FM52X-D, FM52X-H, and
Sancro 69HP reported here and previous data for FM82 [6]. This behavior
is consistent with the model proposed by Ramirez and Lippold [2] and
shown in Fig. 11.

Fig. 11. Schematic showing grain boundary strain accumulation as a function of

boundary geometry and precipitation behavior. (From Ramirez and Lippold [2])
422 J.C. Lippold, N.E. Nissley

Precipitate formation and grain boundary pinning from Nb-rich

particles forming at the end of solidification does not completely explain
the DDC susceptibility that is summarized in Fig. 11. For example,
FM52M-A and FM52M-B have essentially identical compositions (Table 1)
yet their susceptibility to DDC is markedly different. FM52M-C shows a
very high susceptibility to DDC, yet (with the exception of boron) its
composition is nearly identical to A and B. Sanicro 68P which is
compositionally equivalent to FM52 (with the exception of higher Mn) has
better DDC resistance at 950qC. This suggests that grain boundary
segregation and precipitation that occur in the solid-state must also have
some influence on DDC susceptibility. It has previously been shown by
Collins et al. [6] that hydrogen can significantly reduce DDC resistance,
and Ramirez and Lippold [2] have speculated that other interstitial
elements (such as oxygen) may also be detrimental.
The addition of molybdenum to the experimental filler metals 52X-D
and 52X-H appears to provide additional DDC resistance. Based on
Scheil-Gulliver calculations of the solidification behavior of these filler
metals, Mo does not appear to affect the formation of M(C,N) at the end of
solidification [8]. Rather, it is hypothesized that the addition of Mo
changes the nature of M23C6 in the solid-state, thereby promoting a
distribution of carbides along the migrated grain boundaries that leads to
“microscopic locking” of the grain boundaries. Thus, the addition of Nb
above 1 wt% and Mo at 4 wt% results in a situation where both
macroscopic and microscopic locking of the grain boundaries occurs
within the DDC susceptibility temperature range. This situation is shown
schematically in Fig. 12.
The large difference in DDC susceptibility between FM52M-B and
FM52M-C appears to be associated with the small difference in boron. The
higher level of boron in FM52M-B resulted in more precipitation of M23C6
along the weld metal grain boundaries of this filler metal relative to
FM52M-C. Additional work is ongoing to understand the effect of both
Mo and B on grain boundary carbide formation in these filler metals.
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 423

MC Macroscopic GB Locking

Microscopic
GB Locking

M23C6

Fig. 12. Schematic showing the combination of grain boundary tortuosity that
provides macroscopic grain boundary locking (resistance to sliding) and
microscopic locking by the formation of grain boundary precipitates. This
combination appears to provide the best resistance to ductility dip cracking

Conclusions

1. Using the strain-to-fracture test, a wide variation in DDC susceptibility

was observed among Ni-base filler metals conforming to ERNiCrFe-7
and ERNiCrFe-7A.
2. Two experimental filler metals with Nb content exceeding 1 wt% and
Mo additions of 4 wt% were the most resistant to cracking, based on
STF test results.
3. DDC resistance is associated with macroscopic and microscopic
locking of the weld metal grain boundaries that prevents grain
boundary sliding at elevated temperatures.
4. Macroscopic boundary pinning by Nb-rich M(C,N) at the end of
solidification increases the boundary “tortuosity”. It appears that Nb
levels above 1 wt% are required to promote sufficient tortuosity to
424 J.C. Lippold, N.E. Nissley

prevent cracking by a boundary locking mechanism. When Nb is

increased to 2.5 wt%, boundary tortuosity due to pinning is significant.
5. Microscopic boundary locking appears to be promoted by additions of
Mo and B. Both of these elements affect the nature of grain boundary
M23C6 precipitation. Since only one level of Mo (4 wt%) was
investigated, it was not possible to determine if lower levels of Mo
may also be effective in reducing cracking.
6. Significant heat-to-heat variation is possible among filler metals that
meet the same specification. Based on our current understanding of the
DDC mechanism, there was no way to predict the susceptibility to
DDC in ERNiCrFe-7 and -7A filler metals based solely on
composition.

Acknowledgements
The authors wish to thank the Special Metals Welding Products Company
for providing the FM52, FM52M, and FM52X welded samples, and
Sandvik AB for providing the Sanicro 68HP and 69HP consumables. Our
thanks to Kenny Izor, an undergraduate research assistant at OSU, for
performing some of the STF testing. We also wish to thank Mr. Sam Kiser
of Specials Metals and Mr. Claes-Ove Pettersson from Sandvik Materials
Technology, R&D for useful discussions and insight.

References

1. A.J. Ramirez and J.C. Lippold, 2004. High Temperature Behavior of Ni-base
Weld Metal, Part I, Ductility and microstructural characterization, Materials
Science and Engineering A, 380, p. 259–271.
2. A.J Ramirez and J.C. Lippold, 2004. High Temperature Behavior of Ni-base
Weld Metal. Part II, Insight into the mechanism for ductility dip cracking,
Materials Science and Engineering A, 380, p. 245–258.
3. N.E. Nissley and J.C. Lippold, 2003. Development of the strain-to-fracture
test for evaluating ductility-dip cracking in austenitic alloys, Welding Journal,
82(12):355s–364s.
4. J.C. Lippold, N.E. Nissley, and A.J. Ramirez, 2005. Recent advances in the
understanding of ductility-dip cracking, IIW-IX-2161-05, Prague, July 11–14,
2005.
5. J.C. Lippold and N.E. Nissley, 2006. Further investigations of ductility-dip
cracking in high chromium, Ni-base Filler Metals, IIW-IIC-326-06, Quebec
City, August 27–31.
 Ductility-Dip Cracking in High Chromium, Ni-Base Filler Metals 425

6. M.G. Collins and J.C. Lippold. 2003. An investigation of ductility-dip

cracking in Ni-base filler metals-Part 1, Welding Journal, 82(10):288s–295s.
7. J.C. Lippold and A.J. Ramirez, 2004. Insight into the Mechanism of Ductility
Dip Cracking, IIW-IX-2123–04, July 2004, Osaka, Japan.
8. N. Nissley, 2006. Intermediate temperature grain boundary embrittlement in
Ni-base weld metal. PhD Dissertation, The Ohio State University.
Thermodynamic and Kinetic Approach
to Ductility-Dip Cracking Resistance
Improvement of Ni-base Alloy ERNiCrFe-7:
Effect of Ti and Nb Additions

A.J. Ramirez1, C. M. Garzón1-2

1
Brazilian Synchrotron Light Laboratory, Campinas-SP, Brazil
2
National University of Colombia, Physics Department, Bogotá,
Colombia

Abstract

Previous research has suggested that significant improvement in ductility-

dip cracking resistance of ERNiCrFe-7 weld metal can be obtained if sec-
ond phase precipitation during welding is optimized with some carbide and
nitride forming element additions. Therefore, a theoretical and experimen-
tal work has been conducted to address the effect of Nb and Ti additions to
the precipitation of second-phase particles during welding of a ERNiCrFe-
7-like alloy. The critical precipitation temperatures, phase fractions and
atomic partitioning in the microstructure were among the factors analyzed
in the light of DDC resistance improvement.

Introduction

Ni-base filler metal AWS A5.14 ERNiCrFe-7 (NiCr30Fe9), also known as

filler metal 52 (FM-52), which is a nearly matching filler metal for alloy
690, is widely used for joining solid solution strengthened Ni-base alloys,
particularly alloys 600 and 690, and for dissimilar welding between several
Ni-base alloys, stainless and carbon steels. ERNiCrFe-7 exhibits high re-
sistance to corrosion and to stress corrosion cracking at moderate tempera-
tures. However, when highly restrained welded joints are performed, this
alloy is susceptible to ductility dip cracking (DDC), which is a solid state
failure characterized by intergranular cracking occurring at homologous
temperatures ranging from approximately 0.5 to 0.7. This type of failure
428 A.J. Ramirez, C.M. Garzón

plagues different alloy systems that present a FCC structure in these inter-
mediate temperatures, as Ni-alloys, Cu-alloys, steels and stainless steels
and has been the object of research for quite a long time [1, 2, 3, 4, 5].
Among the factors that have been reported to have influence on the DDC
phenomenon are: alloy chemical composition (with special influence of
impurities and interstitials), segregation to grain boundaries, precipitation
behavior, grain boundary migration and pinning, grain boundary orienta-
tion relative to the applied strain and recrystallization [6-12]. Depending
on the system, some of these factors become more important and eventu-
ally control the cracking mechanism. Therefore, as a result of this appar-
ently changing mechanism the technical and scientific literature on DDC
may become quite confusing and controversy involving these different
controlling factors has appeared. For example, the dominant factor on the
DDC of Ni-base alloys with moderate to high levels of impurities has
been proposed to be impurity segregation to the gain boundaries [11]. On
the other hand, the DDC of very “clean alloys’, as most commercial Ni-
base alloys, can not be directly linked to intergranular segregation of im-
purities [10]. However, S segregation has also been observed in these
“clean” alloys [13].
Based on microstructural analysis up to the nanoscale and the exhaus-
tive study of DDC behavior of as-welded Ni-base filler metals AWS A5.14
ERNiCrFe-7 (FM-52) and ERNiCr-3 (FM-82), it was verified that the dif-
ference on DDC resistance between these two alloys is due to the large
amount of primary intergranular precipitates found in ERNiCr-3
(NiCr20Mn3Nb), which is a near-matching filler metal for alloy 600 [8-
10]. These primary precipitates in as-welded ERNiCr-3 are (NbTi)(CN)
cabonitrides and their effect on DDC resistance is based on the capability
of such eutectic-like precipitates to preclude gain boundary (GB) migration
at the very end of solidification. The GB pinning results in tortuous GBs
that would make GB sliding more difficult and strain concentration at tri-
ple points less likely to happen, resulting in less severe void formation,
which improves DDC resistance [6, 8-10]. Figure 1 presents schematically
the proposed effect of GB tortuosity on GB sliding, strain concentration at
triple points and intergranular precipitates, and void formation.
In order to verify the Nb rich carbonitrides role on the DDC resistance
of ERNiCr-3, welds were prepared using the DDC susceptible filler metal
ERNiCrFe-7 with additions of Nb and Ti to induce a larger amount of pri-
mary precipitation. The remarkable DDC resistance improvement of these
modified ERNiCrFe-7 welds was verified using the strain-to-fracture test
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 429

[14-15]. Figure 2 presents the comparison of strain-to-fracture results ob-

tained for the unmodified ERNiCrFe-7 alloy (circles) and the Nb and Ti
additions modified alloys (inverted triangles). It is evident in this figure the
reduction in the number of cracks for the modified alloys.
Other researchers and welding consumable producers have been work-
ing on the development and testing of DDC resistant alloys based on ER-
NiCrFe-7. A patented alloy with 2,5% Nb and 4% Mo additions has shown
a remarkable performance with cracking starting at 10% strain, compared
to 2% strain in the original ERNiCrFe-7 [16]. La additions to alloy 690
have also been proposed with promising results controlling the damaging
effect of impurities segregation to GBs [17]. However, some of the pro-
posed modifications can solve the DDC problems at the cost of compro-
mising the high temperature resistance, the creep properties, the micro-
strutural stability at high temperatures, and the corrosion resistance of the
alloy 690. Therefore, exhaustive and expensive testing may be ahead be-
fore some of these improved alloys can be used in actual fabrication and/or
repair.
The advances on the formulation of DDC resistant Ni-based alloys re-
veal that some understanding has been obtained to date regarding the DDC
phenomena. However, there is still a long way to fully understand this
phenomenon and how to control it not just for Ni-based alloys, but other
important systems as steels and stainless steels, where it causes problems
during continuous casting and hot-rolling. Therefore, the authors have en-
gaged on a project to understand some of the factors controlling DDC in
FFC alloys. The present report concentrates on the thermodynamic and ki-
netic approach to optimize second phase precipitation on Ni-base alloy
AWS A5.14 ERNiCrFe-7 in order to improve DDC resistance[18-19]. The
precipitation of primary nitrides and carbonitrides is optimized through the
proposed addition of Ti and Nb. Secondary precipitation of M23C6 and γ´
were also analyzed. The addition of Ti and Nb up to 3%-wt has been
evaluated using CALPHAD-based numerical simulation of microstructural
evolution during weld metal solidification and cooling.
430 A.J. Ramirez, C.M. Garzón

a.

b.

c.

Fig. 1. Influence of intergranular precipitates on GB sliding, strain concentration

and void formation. (a) Straight GBs result in severe GB sliding and strain con-
centration at the triple points; (b) intergranular precipitates lock the GBs reducing
GB sliding; (c) tortuous GBs and intergranular precipitates lock efficiently the
GBs reducing GB sliding, causing further strain distribution, and reducing the
likeness of fracture initiation [10]
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 431

Actual weld metal samples have also been characterized and compared
with the simulations. Among the obtained results can be highlighted the
kinetic phase diagrams where the microstructure stability during cooling
after welding is related to Ti- and Nb-content. Atomic partitioning between
the austenitic matrix and the precipitates was also evaluated.

a. b.

c.
Fig. 2. Comparison of Strain-to-fracture results on unmodified (circles) and modi-
fied (inverted triangles) ERNiCrFe-7 alloy with small (a) Ti, (b) Nb, and (c) Nb+Ti
additions [15]
432 A.J. Ramirez, C.M. Garzón

Materials and Experimental Procedures

Materials

The nominal chemical composition of the studied alloy AWS A5.14 ER-
NiCrFe-7 used as a master alloy is presented in Table 1. The base alloy
was in the form of welding filler metal wire of 1.2 mm diameter.
Samples were obtained by arc melting using inert gas atmosphere the
base alloy with additions of high purity Ti and Nb powder. Melted buttons
were cold forged at 25°C using a uniaxial hydraulic press to obtain flat-
like shapes of proximately 15 mm diameter and 3 mm thick. As-welded
microstructure with a radial distribution of columnar grains was obtained
by autogenously spot welding on top of forged samples. 5 mm diameter
spot welds were performed with manual GTAW using 115 A (ampere) for
3–5 s. Table 2 shows the targeted and measured chemical composition of
the melted samples, measured using energy-dispersive X-ray spectroscopy
(XEDS) associated with the scanning electron microscope (SEM).

Table 1. Chemical composition of welding filler metal wire AWS A5.14 ER-
NiCrFe-7 used as master alloy (wt.%)

Fe Cr Al Ti Mn Si C N S P Nb+Ta Ni
10.1 29.1 0.71 0.51 0.25 0.13 0.027* 0.015* <0.001 0.003 <0.01 Bal.
Notes: * N-content was between 0.008 and 0.0170 and carbon content was
between 0.025 and 0.029 wt.%.

Chemical and Microstructural Characterization

The produced as-welded microstructures were studied using optical micro-

scope (OM), scanning electron microscopes (SEM) JEOL® JSM-6330F
and JEOL® JSM-5900LV operating between 5 and 20 kV, and transmis-
sion electron microscope (TEM) JEOL® JEM-3010 operating at 300 kV.
The chemical analyses were performed using XEDS associated with the
SEM and TEM. Crystallographic analysis was performed on the SEM us-
ing a backscattered electron diffraction (EBSD) system by HKL® and on
the TEM by selected area electron diffraction.
Conventional grinding and polishing practices were used for OM and
SEM sample preparation. Final polishing for EBSD analyses was achieved
using colloidal silica on a vibrating polishing system. Electrolytic etching
for OM and SEM analyses was performed using a 40%-vol HNO3 aque-
ous solution at 25°C and applying between 1.3 and 1.7 V during 60-120 s.
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 433

TEM sample preparation was performed using carbon replica extraction of

second phases.

Table 2. Chemical analysis of Ti and Nb content of studied samples (wt.%)

Sample Designation Ti % -wt Nb % -wt

ERNiCrFe-7 0.52 -*
0.74/0.95/1.45/2.7
0.7Ti/1Ti/1.5Ti/3Ti -*
respectively
0.6/1.2/1.6/2.9
0.5Nb/1Nb/1.5Nb/3Nb 0.52
respectively
1Nb-1Ti 0.95 1.1
*
Notes: : Measured content lower than the SEM-XEDS detection limit for the
used procedure.
For a confidence interval of 67%: Ti-content, between ± 0.05 and ± 0.15,
and Nb-content, between ± 0.1 and ± 0.2, increasing standard deviations
corresponding to increasing alloying contents in the table.

Simulation Methodology

Calphad-based numerical work consisted of simulating the following phe-

nomena:

• Simultaneous growth during weld pool solidification, against a liquid

region, for both austenite dendrite and primary MX precipitate;
• Dendritic micro-segregation of alloying elements;
• Solid state precipitation of second phases during cooling inside the aus-
tenitic dendrites.

The geometry of the modeled system, which is schematically shown in

Fig. 3, with a size of 5 μm, was defined based on the experimentally ob-
served half the interdendritic spacing. Simulations started from a fully liq-
uid system, which was cooled at the reasonable welding cooling rate of
100°C⋅s–1 down to 327°C (600 K). As usual on some of these simplified
models was assumed a planar and sharp phase interfaces, as well as the es-
tablishment of a local equilibrium condition at these interfaces. Diffusion
in each region was computed by solving the multicomponent non-steady
434 A.J. Ramirez, C.M. Garzón

state atomic diffusion Equation 1 [20]. This equation was solved with
Dictra® software [21-22], which uses the finite-difference method and
assesses thermodynamic and kinetic parameters from specific databases.
The commercially available databases Ni-data® and Mob2® were used for
the calculations performed. The migration rate for each phase interface
was computed by assuming that interface migration is controlled by the
mass balance obtained from fluxes of diffusing elements across the interface
(Eqs. 2 and 3).

Fig. 3. Schematic representation of the simulated system. (a) OM top image of the
welding spot; (b) SEM micrograph showing alloying micro-segregation at inter-
dendritic regions; (c) scheme of the one dimensional system used for numerical
simulations and its relevant dimensions
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 435

n
∂μ i
J K = −¦ L'ki (1)
i =1 ∂z

where JK is the diffusional flux of the atomic alloy component k, L'ki is a

proportionality factor that depends on the mobility of the atomic species,ҏ μi
is the species chemical potential in the alloy, and z is the spatial coordi-
nate.

v MX / L1 (ckMX − ckL1 ) = J kMX − J kL1 , k = 1,2, … , n (2)

v L 2 / γ (c kL 2 − c kγ ) = J kL 2 − J kγ , k = 1,2, … , n (3)

where v MX / L1 and v L 2 / γ are the precipitate–liquid and liquid-austenite

phase interface migration rates, respectively, c kMX , c kL1 , c kL 2 and c kγ are
the concentrations of the atomic component k close the interface in the
precipitate, liquid near to the precipitate, liquid near to the austenite, and
austenite near to the precipitate, respectively, and J kMX , J kL1 , J kL 2 and J kγ are
the corresponding diffusional fluxes.
Within the dendrite, occurrence of diffusion-controlled precipitation of
second phases was accounted by allowing the growth of spherical precipi-
tates at different distances from the dendrite surface. Thus, the dendrite
was divided in several sub-systems; each one consisting of a 0.5 μm di-
ameter spherical cell (1/20 of measured dendrite width), which contains a
spherical precipitate at the center and matrix phase in the outer case. Dur-
ing the cooling the precipitation of both M23C6 and γ’ phases was arbitrar-
ily allowed to start within these cells when driving force was greater than
10–3. Further complication with the simulation of the nucleation stage of
these precipitation reactions was avoided by assuming the prior existence
of 1 nm diameter second-phase nucleus. The chemical composition of this
nucleus is such that leads to equal chemical potential at both sides of the
precipitate–matrix phase interface. Diffusion inside γ’ was neglected due
to lack of diffusion data for this phase in the Mob2® database. The average
chemical composition of each cell was set equal to the previously com-
puted average chemical composition at this point of the dendrite in the sys-
tem containing primary MX precipitates and dendritic austenite.
436 A.J. Ramirez, C.M. Garzón

Results and Discussion

Effect of Ti and Nd Additions on the Thermodynamics and Kinetics

Figures 4 and 5 show pseudo-binary phase diagrams relating phase do-

mains with Ti- and Nb-content for temperatures between 327 and
1727°C for the alloy ERNiCrFe-7 (10.1Fe–29.1Cr–0.71Al–0.25Mn–0.13Si–
0.027C–0.015N wt-%). The liquidus temperature is strongly increased
when Ti is added to ERNiCrFe-7 alloy but not when Nb is added, whereas
the solidus temperature slightly decreases with both Ti and Nb additions.
Therefore, solidification cracking problems are expected when large
amounts of these two elements are added to ERNiCrFe-7, as experimen-
tally verified [15]. Precipitation of primary MX carbonitrides is expected
in this alloy. In the unmodified ERNiCrFe-7 alloy, primary precipitates are
expected to be Ti(N,C). In Nb-modified alloys is expected the substitution
of metallic specie Ti for Nb in MX precipitates, whereas in both Ti- and
Nb-modified alloys substitution of non-metallic specie N for C is also ex-
pected. For example, at 977 °C, the content of Nb in MX precipitates
changes from 2.4 to 50.0 Nb wt.%, with additions between 0.5 and 3 Nb
wt.% to the alloy, respectively. While, the ratio between N-content wt-%
and C-content wt-% in MX precipitates changes from 1.7 in the unmodi-
fied ERNiCrFe-7 alloy to (i) 0.6 in an alloy containing 3.0 Ti wt-% and
(ii) 0.6 in an alloy containing 3.0 Nb wt.%. The liquidus temperature in-
crease with the additions of Ti is due to the precipitation within the liquid
of MX precipitates, which at high temperatures are predominately TiN.
At 700°C the equilibrium presence of secondary phases M23C6, gamma-
prime (γ´), and Cr-rich BCC is expected in the unmodified alloy. On the
other hand, the addition of Ti or Nb induces the stabilization of eta (Ș) and
delta (į) phases, respectively. No significant chemical composition
changes of secondary precipitates as a consequence of alloying additions
are expected.
Computed kinetic phase diagrams that relate phase domains with the Ti-
and Nb-content for temperatures between 727°C and 1727°C are presented
in Fig. 6. The strong influence of Ti and Nb additions on critical phase
transformation temperatures during cooling following welding can be seen.
Solidification temperature is strongly reduced by both Ti and Nb additions.
However, the Nb effect becomes more pronounced for small additions. For
example, according to the calculations 3%-wt of Ti in this alloy will re-
duce the kinetic solidification temperature to ~1150°C, while 3%-wt of Nb
in this alloy could bring the solidification temperature down to ~900°C.
Thus, Nb additions will have a more severe effect on solidification crack-
ing than Ti additions.
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 437

Fig. 4. Equilibrium phase domains as a function of temperature and addition of Ti

to the Ni-base alloy ERNiCrFe-7

Fig. 5. Equilibrium phase domains as a function of temperature and addition of Nb

to the Ni-base alloy ERNiCrFe-7
438 A.J. Ramirez, C.M. Garzón

Fig. 6. Kinetic phase domains as a function of temperature and addition of Ti and

Nb. L: liquid, γ: austenite, γ’: gamma-prime, -Inter: M23C6 or γ’ precipitated near
the dendrite surface, -Intra: M23C6 or γ’ precipitated at dendrite core

Alloying additions and the resultant segregation also induce a decrease

on M23C6 precipitation temperature and strong increase on γ’ precipitation
temperature, with a stronger effect caused by Ti. As can be seen in the pre-
dicted kinetic diagrams the model accounts for intra-dendritic and inter-
dendritic second phase precipitation.
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 439

Micro-Segregation During Solidification

Figure 7 shows SEM micrographs of three representative samples, namely,

master alloy ERNiCrFe-7, 3Ti and 3Nb. The low magnification images re-
veal the cellular solidification pattern. Also some details from the inter-
dendritic (or intercellular) regions and large primary precipitates are pre-
sented. It should be highlighted in these images that not all the contrast
observed at the inter-dendritic regions on the low magnification images is
due to MX precipitates. In fact part of this contrast is due to segregation. In
fact, EBSD analysis (not shown) revealed that these regions have the same
crystallographic structure and orientation as the surrounding matrix. How-
ever, these highly segregated inter-dendritic regions probably have large
amounts of nanometer-sized secondary precipitates, as will be presented
later. Stereological analysis showed that the distance between dendrite (or
cell) triple intersections follows a normal distribution, with average inter-
dendrite spacing of ~7–12 μm.
Figure 8 shows computed average alloying content in the region near
the dendrite surface, this region being defined as the 0.5 μm tick outmost
layer of the dendrite (5% of the dendrite width). It is evident the accentu-
ated tendency of Ti and Nb species to segregate towards the inter-dendritic
regions, as well as to induce strong and moderate negative inter-dendritic
segregation of Fe and Cr, respectively. On the other hand, C and N segre-
gation towards the austenite inter-dendritic regions increases when ER-
NiCrFe-7 is modified with minor additions of Nb and/or Ti (up to ~1
wt.%), meanwhile this segregation decreases when Nb and/or Ti additions
are intensified. Ti additions are more efficient inducing N depletion from
austenite inter-dendritic regions than Nb additions. However, both Ti and
Nb additions display a similar effect with regard to C depletion from aus-
tenite interdendritic regions. Notice that the C and N depletion from aus-
tenite inter-dendritic regions is a consequence of primary carbonitrides
precipitation at this region.
This segregation has a strong effect on the inter- and intra-dendritic pre-
cipitation of MX carbonitrides, M23C6 and γ´ and these precipitates will
have an effect on the DDC resistance of the microstructure and on the al-
loy performance. Therefore, a good understanding of the microstructure,
chemistry, and kinetics of these alloys will provide important insights on
the DDC phenomenon. The following sections describe the second phase
precipitation based on the simulation and characterization work.
440 A.J. Ramirez, C.M. Garzón

Fig. 7. SEM micrographs of three representative samples ERNiCrFe-7, 3 Ti, and

3 Nb. Detail of inter-dendritic regions

Fig. 8. Computed average alloying content at the 0.5 μm tick outmost layer of the
austenite dendrite as a function of Ti and Nb additions

Primary Carbonitride Precipitation

Figure 9 shows TEM micrographs and XEDS spectra of a MX precipitate

observed in a carbon replica of unmodified ERNiCrFe-7 alloy. Figure 9d
shows precipitate morphology; Fig. 9c shows the selected area diffraction
(SAD) pattern at the axis zone [011]MX from the region arrowed in Fig. 9d;
Fig. 4b shows a lattice resolution image of region arrowed in Fig. 9d;
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 441

Fig. 9a shows an XEDS spectrum colleted focusing a 10 nm diameter elec-

tron probe at the middle region of the shown MX precipitate. The SAD
pattern reveals the FCC crystallographic structure with ~0.424 nm lattice
parameter of the observed precipitate, which corresponds to the previously
reported NaCl type-structure of MX carbonitride [23]. The Cu peak ob-
served in the XEDS spectrum is due to the TEM Cu mesh upon which the
carbon replica is supported.

Fig. 9. MX precipitate observed in carbon replica of unmodified ERNiCrFe-7

sample. (a) XEDS spectrum colleted with a 10 nm diameter electron probe at the
middle region of the particle; (b) lattice resolution image detail; (c) SAD pattern;
(d) general view of precipitate morphology

Table 3 presents XEDS measurements and numerical simulations of aver-

age metallic fraction chemical composition of MX particles. The analysis
of this chemical composition data reveals a minimal solubility of Ni, Fe,
442 A.J. Ramirez, C.M. Garzón

Cr, and Al in the MX precipitates. The measured high Cr and Ni content

on the MX precipitates was verified to come from a spurious X-ray signal
generated in the surrounding carbon film containing un-dissolved matrix.

Table 3. TEM-XEDS Chemical analysis and numerical calculations (NC) of me-

tallic elements fractions in MX primary precipitates (wt.%)

Phase Sample Ti Nb Cr Ni Fe Al
EDS NC EDS NC EDS NC EDS NC EDS NC EDS NC
MA/1Ti /3Ti 90.0 99.8 - - - - 4.5 0.2 4.5 <0.001 1.0 <0.001 NM <10-8
MX 1Nb 25 14.8 65 85 5.0 0.2 3.5 <0.01 1.5 <0.001 NM <10-8
1Nb-1Ti 45* 38 42* 62 6 0.2 5 <0.01 1.5 <0.001 NM <10-8
Notes: EDS: Experimentally measured by TEM-XEDS on extracted carbon rep-
lica;
NC: Numerical calculation;
MA: Master Alloy ERNiCrFe-7;
NM: Non measurable value;
* : MX particles with average Nb-content varying between 20 and 70
wt.% and Ti-content varying between 20 and 70 wt.% were identified.

An heterogeneous chemical composition pattern inside the MX precipi-

tates is predicted by Scheil and diffusion calculations; a core consisting of
titanium nitride TiN and a shell consisting of titanium and/or niobium car-
bide (Ti, Nb)C. This gradual core-shell chemical structure does not mean a
change in the crystallographic structure since both phases have the same
crystalline structure and very similar lattice parameter. Therefore, what ac-
tually happens is the gradual exchange of metallic specie Ti in MX
stoichiometry for Nb and of non-metallic specie N for C as temperature
decreases during cooling. Similar behavior on the MX precipitation has
been previously reported on Ni-base alloys [24] and high strength low-
alloy steels [25].
Table 4 summarizes the main stereological features of primary MX pre-
cipitates observed on the SEM. Primary MX precipitates were evenly dis-
persed within the microstructure, although minor presence of clusters with
up to four precipitates was also observed (not shown). Preference for grain
boundary precipitation was not observed. MX precipitates with near equi-
axial, elongated and ramified morphologies were observed (Fig. 7). Nb ad-
ditions induced a prevalence of both near equiaxial and elongated primary
MX morphologies and Ti additions induced a prevalence of elongated and
ramified morphologies. In general terms, in the samples alloyed with Nb
the MX precipitates displayed irregular holes or imperfections (Fig. 7).
The effect of alloying additions on observed MX morphologies can to a
certain extent be explained by the relationship between the MX transformed
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 443

fraction and temperature. Figure 10 presents the measured and calculated

total volumetric fraction of primary MX precipitates present in the micro-
structure for different alloying additions.

Table 4. SEM-measured size and volumetric fraction of MX precipitates. Numeri-

cally computed volumetric fraction of MX particles are also shown for compari-
son

Average size Vol.-fraction [vol.-%] Morphologic Character Distribution

Sample
[μm2] SEM NC Near equiaxial Elongated Ramified
ERNiCrFe-7 0.035 ± 0.01 0.08 ± 0.04 0.10 High Low Low
0.7Ti 0.05 ± 0.02 0.1 ± 0.05 0.12 High Low Low
1Ti 0.05 ± 0.02 0.1 ± 0.05 0.13 High Low Low
Ti
1.5Ti 0.07 ± 0.03 0.18 ± 0.07 0.19 High Moderate Low
3Ti 0.11 ± 0.05 0.33 ± 0.10 0.24 Moderate Moderate Moderate
0.5Nb 0.08 ± 0.03 0.16 ± 0.07 0.10 High Moderate Low
1Nb 0.11 ± 0.05 0.22 ± 0.08 0.15 High Moderate Low
Nb
1.5Nb 0.14 ± 0.06 0.28 ± 0.11 0.19 Moderate Moderate Low
3Nb 0.18 ± 0.06 0.43 ± 0.14 0.24 Low High High
Notes: NC: Numerically computed.

As previously discussed, according to phase diagrams in Figs. 4 and 5,

growth of MX during cooling takes place initially directly as precipitation
inside the liquid phase, i.e. above γ solvus temperature, then cooperative
growth of both MX and γ take place from the liquid phase. The primary
MX formed above the γ solvus temperature will be called pro-eutectic MX
and the primary MX formed below γ solvus will be called eutectic MX.
Figure 11 shows the calculated volumetric faction ratio of pro-eutectic MX
to total MX as a function of alloying. On the other hand, Figures 12 and 13
show the computed MX growth kinetics as a function of Nb and Ti con-
tent. The fraction of pro-eutectic MX with regard to the total MX smoothly
increases with Ti additions and steeply decreases with Nb additions. On
the other hand, the higher the alloying content the slower the proportional
kinetics of MX growth. Thus, for high additions of Ti and Nb a significant
fraction of the total MX grows at very low temperatures (~1100–1200°C).
However, the total fraction of MX also increases with alloying content and
consequently the reduction on the proportional kinetics should be carefully
evaluated. Nevertheless, the proportional precipitation kinetics during the
weld cooling is definitely slowed down by the higher Nb and Ti additions.
Nb additions displayed a more accentuated effect than Ti additions on de-
creasing the fraction of pro-eutectic MX and retarding the proportional
MX growth kinetics. The Ti addition effect on promoting higher pro-
eutectic MX fractions can be understood in light of the predominant phase
444 A.J. Ramirez, C.M. Garzón

formed, which in this case is Ti nitride TiN and/or Ti carbonitride Ti(CN),

different from the Nb additions that predominantly promote Nb Carbide
NbC and/or Nb carbonitride Nb(CN). Therefore, TiN being a phase that
becomes stable at higher temperatures than NbC, it is expected that higher
fractions of pro-eutectic MX are formed when Ti instead of Nb is added to
ERNiCrFe-7 alloy. However, Ti stabilizes γ´, which in principle is a unde-
sired phase in this type of solid solution strengthen Ni-base alloy.

Fig. 10. (a) Computed MX fraction as a function of Ti and Nb additions; (b) com-
puted and SEM-measured MX fraction as a function of Ti or Nb additions

In general terms, both the MX precipitate size and volumetric fraction

were increased when the alloying addition was larger (Table 4). However,
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 445

the increase in volumetric fraction was higher than the increase in average
size. This is due to solute segregation at interdendritic regions that simul-
taneously increases driving force for MX precipitation and the number of
cell-triple intersections, the density of nucleation sites and the concomitant
number of particles per unit area being consequently increased. For modi-
fied alloys with high alloying contents (3Nb or 3Ti wt-%), the experimen-
tally verified intensification of MX precipitation was significantly higher
than the one numerically predicted, which most probably arises from errors
associated with neglecting nucleation events in the numerical algorithm as
well as with simplifying the analyzed 3D micro-segregation phenomena to
a 1D system. However, an actually wider solubility range of either carbon
or nitrogen inside the MX phase with regard to the one described by the
Ni-data® database can also be a factor affecting the relationship predicted
between alloying additions and primary carbonitride precipitation, i.e. the
actual carbon and/or nitrogen content of MX precipitates can be lower than
the one numerically predicted.
It is a reasonable approximation to relate the MX precipitates displaying
irregular morphologies with eutectic particles precipitated at lower tem-
peratures, where diffusion distances are shorter due to limited diffusivity
of substitutional atomic species in austenite. On the other hand, it is a good
approximation to relate the MX particles displaying regular, almost po-
lygonal or equiaxial morphologies with pro-eutectic particles precipitated
at high temperatures, where minimization of interfacial energy prevails
over minimization of diffusion distances. Thus, it can be inferred that Nb
and Ti additions induce an increase in MX particles with irregular mor-
phologies due to MX particles in modified alloys having more eutectic
character and they are precipitated at lower temperatures when compared
to MX particles in the unmodified alloy. Irregular primary precipitates are
expected to produce less cracking resistance microstructures due to the fa-
cilitated nucleation of cavities at such intergranular particles when com-
pared to smooth round-like precipitates. Therefore, Nb and/or Ti additions
should be optimized in order to have the right particle temperature forma-
tion, the right number and distribution of particles within the macrostruc-
ture and finally the right particle morphology. From the point of view of
morphology, the lower the addition of these elements would produce
needed round-like primary MX precipitates.
446 A.J. Ramirez, C.M. Garzón

Fig. 11. Computed ratio between vol.-fraction of MX precipitated above γ-solvus

temperature (MXPro-Eutectic) and total vol.-fraction of primary MX (MXTotal).

Fig. 12. Computed growth kinetics of primary MX. The start and end of MX pre-
cipitation are shown as a function of Nb and Ti content of the alloy and tempera-
ture. Notice the reverse order of alloying content axis.
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 447

Fig. 13. Computed growth kinetics of primary MX. The 0%, 20%, 80% and 100%
transformed fractions are shown as a function of temperature, and Nb and Ti con-
tent of the alloy. Notice the reverse order of alloying content axis

Secondary Precipitation of M23C6 and γ’

As an example, Figs. 14 and 15 show some of the results from the TEM
analyses of M23C6 and γ’ precipitates observed in carbon replicas extracted
from the master alloy ERNiCrFe-7 and the sample 1Nb-1Ti. There are
shown in these figures the overall precipitate morphology, enlarged detail
with lattice resolution image, SAD patterns of the region shown, and
XEDS spectra colleted focusing an 5 nm diameter electron beam at the
centre of the correspondingly enlarged second-phase particles.
There were not detected significant changes in M23C6 or γ’ morphology
as a consequence of alloying additions. M23C6 carbides precipitated in
448 A.J. Ramirez, C.M. Garzón

densely agglomerated regions with their individual particle shape slightly

deferring from equiaxial. Minor precipitation of this phase was observed
with a vol. fraction lower than ~0.01%. Individual particles were smaller
than ~30 nm diameter and most of them were around 5 nm diameter.
Gamma-prime particles also precipitated in agglomerated regions, al-
though not as densely packed as the M23C6 carbides. A significant disper-
sion in γ´ size and volumetric fraction was observed. γ” has not been iden-
tified; however it is expected in high Nb alloys.

Fig. 14. M23C6 precipitates observed in carbon replicas from master alloy ER-
NiCrFe-7. XEDS spectra of precipitate colleted using 5 nm diameter electron
beam at the center of precipitate shown on the lattice resolution image detail; SAD
pattern
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 449

Fig. 15. γ’ or γ” precipitates observed in carbon replicas from alloy 1Ti-1Nb.

XEDS spectra of precipitate colleted using 5 nm diameter electron beam at the
center of precipitate shown on the lattice resolution image detail; SAD pattern

Table 5 presents a summary of the semi-quantitative description of

M23C6 and γ’ particles analyzed in five samples. It is also presented the
computed driving force for precipitation of secondary phases averaged at
the region near the dendrite shell, i.e. the outmost layer equivalent to 5%
of dendrite width. Care should be taken because the database used appar-
ently does not differentiate between γ’ and γ”; therefore, both phases may
be mixed and will be called here γ’. Work is ongoing in this matter. Driv-
ing force for γ’ precipitation steadily increases with alloying additions,
whereas the driving force for M23C6 precipitation is minimum when high
450 A.J. Ramirez, C.M. Garzón

or non alloying additions are made. These two behaviors can qualitatively
be explained by the average Ti- or Nb- (constituents of γ’ and γ”) and C-
content (constituent of M23C6) segregated to the inter-dendritic region (see
Fig. 8). The higher the alloying segregation, the higher the driving force
for secondary precipitation. Although, simultaneous presence of M23C6 and
γ’was numerically predicted, it was not observed. However, it is expected
that a microstructure containing γ’ precipitation may hinder the observa-
tion and identification of small fractions of M23C6 carbides. Nevertheless,
when the numerical and experimental results are analyzed together it is
found that M23C6 precipitation can be inhibited by the faster γ’ stability in-
crease compared to the M23C6 stability rise (samples 1Nb, and 1Ti1Nb).
This is probably due to the competition between these two phases for the
nucleation sites, with M23C6 nucleation being overrun by γ’ nucleation.
The prevalence of γ´ nucleation is due to the lower volumetric atomic dif-
fusion required and the lower energy interface formed when compared
with M23C6 nucleation. It is not clear to the date if intergranular M23C6 are
damaging for the DDC resistance. However, is believed that when these
precipitates appear in the microstructure the ductility dip cracks have al-
ready had the chance to form.

Table 5. Semi-quantitative description of M23C6 and γ’ particles experimentally

observed and driving force for their precipitation near the dendrite surface at
1027°C and 727°C. Driving force, ΔG/RT, is reported as a dimensionless magnitude
Dr iving For ce, ΔG/RT
Sample TEM Char acter ization M 23C 6 γ’
1027°C 727°C 1027°C 727°C
FM52 Minor precipitation of M23C6 0.13 0.96 -0.12 0.23
1Ti Minor precipitation of M23C6 0.18 1.01 -0.12 0.24
1Nb Minor precipitation of γ’ 0.22 1.12 -0.04 0.30
1Nb-1Ti Moderate precipitation of γ’ 0.37 1.32 -0.004 0.33
3Ti Profuse precipitation of γ’ -0.23 0.79 0.09 0.45
Note: Minor: aprox. between 0.001 and 0.01 vol.%.
Moderate: aprox. between 0.05 and 0.5 vol.%.
Profuse: aprox. between 1 and 10 vol.%.

XEDS spectra (Figs. 14 and 15) showed the occurrence of minimal solu-
bility of Ni, Fe, Al, Nb and Ti in M23C6 and Fe and Cr in γ’. In a similar
way as presented above for XEDS analysis of MX particles, it is expected
that Ni and Fe content of M23C6 particles and Fe and Cr content of γ’
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 451

particles are lower than revealed by the XEDS analyses, which are de-
graded by spurious X-ray signal coming form regions away from the ana-
lyzed particles. Minor changes of chemical composition inside second-
phase precipitates were predicted by kinetic simulations, and therefore
these results are not shown. These numerical results show that atomic ele-
ment stoichiometry in second-phase precipitates can be described as Cr26C6
and Ni3(Ti,Nb,Al). XEDS analyses and numerical simulations of average
chemical composition of M23C6 and γ’ revealed that Ti additions did not
caused significant composition changes of M23C6 or γ’. In contrast, in Nb-
modified alloys, the here so-called γ’ particles displayed substitution of Ti
and Al for Nb. Therefore, the new formed phase may be γ”.

Summary

Based on the exposed results and previously reported data on the weldabil-
ity and performance of alloy ERNiCrFe-7, it is possible to establish appro-
priate composition ranges for the modification of such alloy with Nb and
Ti additions. Thus, the kinetic phase diagrams presented indicate that alloy
ERNiCrFe-7 modified with Ti-content up to ~1.0 wt-% and Nb-content be-
tween ~0.5 and ~1.0 wt-% will present optimized as-welded microstruc-
ture with regard to DDC resistance and in some degree to corrosion and
SCC resistances because of (i) M23C6 precipitation can be slightly inhibited
with regard unmodified ERNiCrFe-7, (ii) although some γ´ precipitation is
induced, still low driving force for γ’ precipitation is established, and (iii)
profuse MX precipitation above the ductility-dip temperature range (i.e.
~0.2–0.25 vol.% of MX above ~1150 °C) takes place. Intense precipitation
of primary carbonitrides above the ductility-dip temperature range and mi-
nor precipitation of secondary chromium carbides during the whole weld-
ing cooling cycle are desired because [1-5] primary carbonitride precipita-
tion results on increased DDC resistance due to the formed tortuous grain
boundaries, whereas chromium rich carbide precipitation at low tempera-
ture induces chromium depletion of the matrix phase near grain boundaries
and, consequently, may reduce corrosion resistance as well as SCC resis-
tance of the alloy.
It was observed that Ti or Nb additions are useful to induce increasing
precipitation of primary carbonitrides and simultaneously to inhibit secon-
dary precipitation of chromium rich carbides. This is due to alloying addi-
tions increasing the fraction of C partitioned in the (Ti,Nb)C phase with
regard to C partitioned in the γ phase. However, these Ti or Nb additions
lead to strong atomic segregation near the interdendritic regions and, as a
452 A.J. Ramirez, C.M. Garzón

result, to both intense decreasing of solidification temperature and precipi-

tation of N- and C-free second phase, namely γ’, at the interdendritic re-
gions in as-welded microstructures. However, high Ti additions compro-
mise the thermal stability of this alloy due to γ’-induced precipitation
hardening during following heat treatments and service.

Acknowledgements

The authors would like to acknowledge the Brazilian Synchrotron Light

Laboratory-LNLS, CNPq (contract 150523/2005-0), and FAPESP (con-
tract 04/04526-8) for funding. Arc melting of the alloys was performed at
State Univ. of Campinas with the invaluable help of Prof. S. Gama and Dr.
A. Coelho. Special thanks to Sam D. Kiser from Special Metals Welding
Products for providing the Ni-base alloys.

References

1. Rhines FN, Wray PJ (1961) Investigation of the Intermediate Temperature

Ductility Minimum in Metals. Trans. ASM 54: 117–128.
2. Yeniscavich W (1966) A Correlation of Ni-Cr-Fe Alloy Weld Metal Fissuring
with Hot Ductility Behavior. Weld. J. 45(8): 344s–355s.
3. Ouchi C, Matsumoto K (1982) Hot ductility in Nb-bearing High-strength Low
Alloy Steels. Trans. Iron Steel Inst. Japan 22(3): 181–189.
4. Mintz B, Yue S, Jonas JJ (1991) Hot Ductility of Steels and its Relationship to
the Problem of Transverse Cracking During Continuous Casting. Inter. Mater.
Rev. 36(5): 187–217.
5. Duvall DS, Owczarski WA (1967) Further Heat-Affected-Zone Studies in
Heat-Resistant Nickel Alloys. Weld. J. 46(9): 423s–432s.
6. Collins MG, Lippold JC (2003) An Investigation of ductility Dip Cracking in
Nickel-Based Filler Materials – Part I. Weld. J. 82(10): 288s–295s.
7. Collins MG, Ramirez AJ, Lippold JC (2003) An Investigation of Ductility-
Dip Cracking in Nickel Based Weld Metals – Part II. Weld. J. 82(12):
348s–354s.
8. Collins MG, Ramirez AJ, Lippold JC (2004) An Investigation of Ductility-
Dip Cracking in Nickel Based Weld Metals – Part III, Weld. J. 83(2): 39s–49s.
9. Ramirez AJ, Lippold JC (2004) High Temperature Behavior of Ni-base Weld
Metal – Part I. Ductility and Microestructural Characterization. Mater. Sci.
Eng. A. 380: 259–271.
 Ductility-Dip Cracking Resistance Improvement of ERNiCrFe-7 453

10. Ramirez AJ, Lippold JC (2004) High Temperature Behavior of Ni-base Weld
Metal – Part II. Insight Into the Mechanism for Ductility Dip Cracking. Mater
Sci. Eng. A 380: 245–258.
11. Nishimoto K, Saida K, Okauchi H (2006) Microcracking in Multipass Weld
Metal of Alloy 690 Part 1 – Microcraking Susceptibility in Reheated Weld
Metal. Sci. Tech. Weld. Joining 11(4): 455–461.
12. Nishimoto K, Saida K, Okauchi H, Ohta K (2006) Microcracking in Multipass
Weld Metal of Alloy 690 Part 1 – Microcraking Mechanism in Reheated
Weld Metal. Sci. Tech. Weld. Joining 11(4): 462–470.
13. Capobianco TE, Hanson ME (2005) Auger Spectroscopy Results from Ductil-
ity Dip Cracks Opened under Ultra-High Vacuum. In: Proc. 7th Int. Conf. on
‘Trends in Welding Research’, Pine Mountain, GA, USA, May 2005, ASM
International: 767–772.
14. Nissley NE, Lippold JC (2003) Development of the Strain-to-Fracture Test.
Weld. J. 82(12): 355s–364s.
15. Ramirez AJ, Sowards JW, Lippold JC (2006) Improving the Ductility-Dip
Cracking Resistance of Ni-Base Alloys. J. Mat. Proc. Tech. 179: 212–218.
16. Lippold JC, Nissley NE (2006) Further Investigations of Ductility-Dip Crack-
ing in High Chromium, Ni-base Filler Metals. Document IIW-IIC-326-06:
1–10.
17. Nishimoto K, Saida K, Okauchi H, Otha K (2006) Microcracking in Multipass
Weld Metal of Alloy 690 Part 1 – Prevention of Microcraking in Reheated
Weld Metal by Addition of La to Filler Metal. Sci. Tech. Weld. Joining 11(4):
471–479.
18. Garzon CM, Ramirez AJ (2007) Second Phase Precipitation in As-Welded
Microstructures of ERNiCrFe-7 Ni Alloy Modified with Ti and Nb Additions.
Submitted to Mater. Sci. Eng. A.
19. Garzon CM, Ramirez AJ (2007) Kinetic phase diagrams for assessing second-
phase precipitation during welding a Ni-30Cr-10Fe alloy modified with Ti
and Nb additions. Submitted to Scripta Materialia.
20. Andersson JO, Ågren J (1992) Models for numerical treatment of. multicom-
ponent diffusion in simple phases. J. Appl. Phys. 72: 1350–1355.
21. Borgenstam A, Engström A, Hoglund L, Ågren J (2000) J. Phase Equilibria
21: 269–280
22. Andersson JO, Helander T, Höglund L, Shi P, Sundman B (2002) Thermo-
Calc & DICTRA, Computational Tools for Materials Science. Calphad 26:
273–312
23. Christensen AN (1978) Acta Chem. Scand. Ser. A 32(1): 89–90.
24. Huang X, Zhang Y, Liu Y, Hu Z (1997) Effect of Small Amount of Nitroten
on Carbide Characteristics in unidirectional Ni-Base Superalloy, Metall.
Mater. Trans. A 28A(10): 2143–2147.
25. Craven AJ, He K, Garvie LAJ, Baker TN (2000) Complex Heterogeneous
Precipitation in Titanium–Niobium Microalloyed Al-killed HSLA Steels—1.
(Ti,Nb)(C,N) Particles, Acta Mater. 48: 3857–3868.
Index

6xxx Al alloys during direct-chill (DC) casting

hot tearing during laser butt welding, of, 4
241–254 hot tearing, 3–7, See also Hot
2D/3D characterisation, 241–254 tearing
experimental procedure, 242–244 initiation mechanism, 12
fastening system influence, nucleation, 5, 13
246, 247 prediction, 3–16
filler wire quantity influence, 245 propagation mechanism, 5, 12, 13
process optimization, 241–254 tensile stress and, 13
solidification rate influence, 244 solidification crack initiation in,
weld pool chemical composition 277–304, See also individual
influence, 241 entry
weld pool silicon content See also 6xxx Al alloys
influence, 241–245, 247 Ashby maps, 375, 376
welding parameters influence,
244–247 Backfilling effect on cracking
welding speed influence, 244, 245 susceptibility, 166–169
hot tearing phenomenon in, Bay, N., 319
characterisation, 247–253 Bead shape induced cracking, 116, 117
fracture surface, 248 Beckerman, C., 30
shape of hot tear, 250 Bend testing, Ni-base SAW weld
X-ray picture, 251 metals, 331, 333, 336–338
3D X-ray tomography, 252, 253 Billets, hot rolled
resistance spot welding, hot crack appearance in, 371–386
cracking during Ashby maps, 375
Mg content influence, 313 central bursts, 372, 373, 377–380
Si content influence, 313 ductile fracture, 374–377
simulation investigations, mechanical factors, 371
311–324, See also under metallographic examination,
Simulation investigations 380–384
strategies for avoiding, 313 microstructural factors, 371
Aluminium alloys plastic strains causing, 371
hot cracking in strain rate, 372
cause, 5 stress rate, 372
composition of alloy, 4 Bochvar, A.A., 5
conditions, 5 Bouchard, M., 5
456 Index

Braccini, M., 5, 9 major factors, 385

Bridging undercooling, 32, 33 segregation of inclusions, 385
Brittleness temperature range (BTR), tensile stress/strain situation, 385
66, 264 SICO tests for hot/reheat crack
welding cracks, 95–113 susceptibility, 378
See also under Integral approach Centreline cracking
bead shape induced cracking, 116
Calphad calculations during submerged arc welding,
in Ni-base filler metals weld 115–125, See also Submerged
solidification cracking, 153, 154 arc welding
calculated phase diagrams, formation, 116
154–157 segregation induced cracking, 116
equilibrium solidification surface profile induced cracking, 116
temperature ranges and Charpy V-notch testing, 351, 357–361
solidification sequence, 157 Chihoski, R.A., 129, 138, 143, 283, 295
Scheil-Gulliver solidification Cicala, E., 244
simulation, 154–160 Circular patch test (CPT), 49
Campbell, J., 5, 20, 44, 66 Classic fracture mechanics, 64, 65
Cast pin tear test (CPTT) Clyne, T.W., 5, 8, 9, 66
arc melting technique, 194 Coalescence, 32, 33
for circumferential cracks, 194, 199 Coherency point, 19
high performance Ni-base Cold cracks, 11
superalloys, testing, 203–210 Collins, M.G., 412, 422
dendritic arm spacing, Constitutional solidification, criteria
evaluation, 205–207 determined by, 66, 67
fracture morphology, 208–210 Cont. tensile weldability test
shrinkage porosity, 207 (CTW), 49
solidification cracking Control, cracking, 59–73
susceptibility, 204, 205 See also under Thermo-physical
solidification microstructure, mechanism
205–207 Controlled Tensile Weldability (CTW)
for Ni-base superalloys weld test, 55, 127, 128, 131–133
cracking evaluation, 193–213 CTW weldability test, sensitivity,
origin, 194 301–303
other hot cracking tests versus, position of extensometer,
200–202 301–303
sensitivity of, 199 solidification crack initiation in Al,
See also Second-generation CPTT 277, 278, 285
Cavitation in hot rolled billets, alloy heat/temper effect, 294, 295
372–384 hydrogen contamination, 283,
during metal forming, 385 284, 301
microscopic cavities, 385 insert control test, 297
See also Billets, hot rolled local strain measurement, 287,
Central burst-susceptible steels 288, 294
characteristics, 377–380 oxygen contamination, 283, 284,
hot tensile tests, 380 299, 300
 Index 457

test sequence, 287 lambda-curve, 8

weld travel speed, 283, 295, 296 sensitivity, 9
Crack appearance, in hot rolled billets, thermal history, 8
371–386 Dislocation creep, 107
See also Billets, hot rolled Dobatkin, V.I., 5
Crack initiation, weld solidification Drezet, J.-M., 19, 23, 30, 44, 67, 71
cracking, 40–45 Ductile fracture
Crack length parametrics, 49–52 in, hot rolled billets, 374–377
Cracking susceptibility coefficient, 66 stages, 375
Critical strain void growth stage, 375, 376
rate determination for solidification void initiation stage, 375, 376
cracking, 77–91, See also under void linking, 375, 376
Numerical simulation Ductility curves, 22, 23, 40–42
weld solidification cracking, 40–43 Ductility-dip cracking (DDC),
critical strain rate, 43–45 95, 96, 99
ductility curves, 40–42 in high chromium, Ni-base filler
slow bend trans-varestraint metals, 409–424
test, 44 composition effect, 417–420
Λ Curve, 24 grain boundaries, 421, 422
Cyclic tensile loading tests mechanism, 421–423
high nitrogen stainless steels, precipitate formation, 422
362–364 susceptibility at 950°C,
413–417
Dahle, A.K., 281, 298 weld metal microstructure effect,
Darcy’s equation, 21 417–420
David, S.A., 71 high nitrogen stainless steel, 350,
Davies, G.J., 5, 8, 9 359, 363
Deformations, during high- in multi pass laser overlay welds
temperature creep, 107–109 of alloy 690, 389
DELTA SPOT™, resistance spot grain boundary segregation,
welding process, 311–313 398, 399
simulation components for, 317 microsegregation to grain
Dendritic arm spacing (DAS) boundary, 400, 401
evaluation microsegregation, theoretical
high performance Ni-base super model, 399, 400
alloys, 205–207 prediction, 403–405
Dickhaus, C.H., 5 stress and strain, analysed models
Diffusion creep, 107 for, 402, 403
Digital image correlation (DIC) theoretical approaches, 398–405
technique, laser based, 47 thermal strain analysis, 401, 402
Dilution maps, 284, 285, 291, 292 See also ERNiCrFe-7 alloy
Direct-chill (DC) casting of Embrittlement, 104
aluminium alloys, 4 Environment-assisted cracking
hot tearing applicability to, 8, 9 (EAC), 171
billet centre, 9 ERNiCrFe-7 alloy
casting speed, 9 DDC resistance improvement in
458 Index

Calphad-based numerical Fracture surface, analysis, 98, 99,

work, 433 181–186, 202, 208–210, 248
chemical and microstructural Charpy test, 357–361
characterization, 432, 433 hot tearing in laser welding, 248
influencing factors, 428 transcrystalline fatigue fracture
intergranular precipitates surface, 364
influence, 430 Fredriksson, H., 5
materials, 432 Fully austenitic stainless steels
micro-segregation during crack initiation and growth,
solidification, 439, 440 134–136
Nb addition effect, 436–438 CTW test, 127, 131–133
primary carbonitride precipitation, high speed recording, 134
440–447 hot cracking in, 349–368, See also
secondary precipitation of Heat affected zone (HAZ) hot
M23C6 and γ’, 447–451 cracking
simulation methodology, 433–435 local thermo-mechanics
thermodynamic and kinetic compressive cells, 129
approach, 427–452 design specific influences,
Ti addition effect, 436–438 128–130
Eskin, D.G., 4, 20 tensile cells, 129
Extensometer measurements, 47 MVT-Test, 131
Externally loaded tests solidification cracking susceptibility
Ni-base weld metal hot cracking local weld deformation influence
susceptibility, evaluation, 331 on, 127–144
Extrinsic weldability testing, 49 strain and strain rate during
welding, 136–143
Farup, I., 5 external load effect, 138–140
Feeding factor, hot tearing and, 6, 13 local thermo-mechanics, external
Feng, Z., 43, 71, 129 load effect on, 141–143
Feurer, U., 5, 6, 9, 44, 66 restraint free, 136–138
Filler dilution amount, importance solidification crack accompanying,
solidification crack initiation 140, 141
in Al, 278 strain monitoring, 133, 134
Filler metals, see High chromium Fuselage structures, welded, 257
filler metals; La additions; Fusion welding process, 59
Ni-base filler metals
Finite Element Analysis (FEA), 62, 63 Gas-tungsten arc welding (GTAW)
Finite element method (FEM), 311 process, 205
FEM mesh for resistance spot Geometrical parameters
welding with process tape, 318 welding influencing, 258–275, See
See also SORPAS® FEM software also Integral approach
Fractography, 63, 110 Goldak, J., 77
Fracture mechanics Grain boundary segregation, 53
in hot tearing and, 6, 10, 12 Granular model for hot tearing
role in weldment cracking, 64–66 criterion, 32–35
classic fracture mechanics, 64 coalescence, 32
 Index 459

Voronoï tessellation, 34 types, 96

feeding KPL model, 34 See also Liquation cracking
microsegregation model, 34 Heine, R.W., 5
Gremaud, M., 19, 30, 44, 67 High chromium filler metals
Griffith model for brittle crack DDC in, 409–424, See also under
growth, 6, 7 Ductility-dip cracking
Growth, crack experimental procedures,
weld solidification cracking, 410–412
45, 46 STF test, 409–413
in a liquid film, 45 temperature-strain curves,
modeling approaches, 45 412, 413
Gruzleski, J.E., 5 High nitrogen stainless steel
Guven, Y.F., 5 HAZ hot cracking in, assessment,
349–368, See also Heat
Hansen, P.N., 66 affected zone (HAZ) hot
Heat affected zone (HAZ) hot cracking
cracking High-alloy steels, welding cracks
ductility dip cracks, 96 in brittle temperature ranges,
in high nitrogen stainless steel, 95–113
assessment, 349–368 classification, 95–113
Charpy V-notch tests, 349, 351, mechanisms, 95–113
357–361 Hilbinger, R M., 258
cyclic tensile loading tests, Hood, B., 172
362–364 Hot crack sensitive nickel-base alloys,
ductility dip cracks, 350, joining, 215–236
359, 363 See also SAW cold wire technology
fracture surface studies, 361, 362 Hot cracking susceptibility (HCS),
intercrystalline hot crack with a 27, 28
freely solidified surface, 361 distribution of, 29
intercrystalline hot crack with a of Ni-base SAW consumables,
smooth surface structure, 361 329–345, See also Ni-base
mechanical testing, 349, 355–364 SAW weld metals
microscopic examination, Hot cracking tests using MVT-tests,
353–355 234–236
necked weld pool shape Hot rolled billets, crack appearance in,
influence, 351 371–386, See also Billets, hot
tensile testing, 351, 355–357 rolled
transcrystalline fracture with Hot tearing
dimple structure, 361 during laser butt welding of 6xxx Al
transcrystalline fracture without alloys, 241–254, See also 6xxx
dimples, 361 Al alloys
weld pool shape influence, 351 initiation model, 67
X-ray cone beam computer susceptibility, 66
tomography, 349, 364–367 Hot tearing, in Al alloys
liquation or segregation hot applicability to DC casting, 8, 9
cracks, 96 cold cracks and, 11
460 Index

equiaxed dendritic solidification, 7 slot influence on weldability,

feeding factor and, 6, 7, 13 264–273
fracture mechanics and, 6 active principle, 264–266
initiation, 4 BTR, 264
liquid-metal-embrittlement determinations, 266, 267
mechanism, 7 influencing parameters,
mechanical criteria, 8, 10 267, 268
mechanisms, 3–7, 15 residual stress, 264, 265
as a base for new hot tearing slot height, 267, 268
criteria, 10–15 slot width, 267, 268
fractures, 10, 12, 15 stringer configuration, 267, 268
nucleation, 10, 15 tack welding regime, 267, 268
propagation, 10, 15 weld seam length, 267, 268
temperature range, 14, 15 welding steps order, 267, 268
non-mechanical criteria , 8, 10 for structural behavior and
porosity formation and, 6 weldability optimization,
micro-porosity, 6 257–275
solidification and, 4 skin and stringer welding, 259
strain causing, 6 stringer geometry
stress causing, 6 modification, 260
ultimate hot cracking criteria structural behavior and weldability,
requirement, 8 interaction between, 260
See also RDG (Rappaz Drezet Intrinsic weldability testing, 49
Gremaud) criteria Iron role, solidification crack initiation
Houldcroft test, 49, 72 in Al, 281, 282, 298
Hull, F.C., 194
Hunt, J.D., 5, 66 Jennings, P.H., 42, 279, 280
Hybrid welding, 351–353 Joint welding, SAW cold wire
technology, 231–233
Impurities role, in weld solidification
cracking, 54 Kannengießer, Th., 130
Integral approach Katgerman, L., 5, 8, 9
displacements within BTR, Kazutoshi Ichikawa, 122
evaluation, 268–273 Kirchhoff’s law, 34
length of secondary part, 271, 272
length of tack-free primary La additions effect on microcracking
part, 272 in multi pass laser overlay welds of
reference model, 268, 269 alloy 690, 389–407
slot height, 270, 271 ductility-dip cracks, 389
slot width, 270 experimental procedures,
welding direction, 273 392, 393
slot influence on structural materials, 391, 392
behavior, 261–264 microcracking behaviour,
stress concentration analysis, 393–395, See also Overlay
261–264 weld metal
 Index 461

primary water stress corrosion Macro test of welds, 122–124

cracking, 390 Magnin, B., 5, 9
strain, 403, 405 Mathier, V., 33
stress, 403, 405 Matsuda, F., 46, 67
temperature change, 403, 405 Measurement by means of in-sito
Lahaie, D.J., 5 observation (MISO), 48
Langlais, J., 5 Metallographic evaluation
Laser butt welding of 6xxx Al alloys Ni-base SAW weld metals, 329,
hot tearing during, 241–254, See 332–336
also 6xxx Al alloys Microcracking, in multi pass laser
Laser cladding technique, 390 overlay welds of alloy 690,
in-air welding, 392 389–407
underwater welding, 392 Mo, A., 5
Lees, D.C.J., 5 Mock-up metal welds, Ni-base alloy,
Lin, W., 172 175–189
Lippold, J.C., 410, 421, 422 Modified varestraint test (MVT), 49,
Liquation cracking, 350 80, 81
assessment, 351 hot cracking tests using, 234–236
cyclic tensile loading, 351 Ni-base SAW weld metals, 329,
definition, 350 331–342
See also Heat affected zone (HAZ) Monroe, C., 30
hot cracking Multi pass laser overlay welds of
Local strain measurement alloy 690
discontinuities on weld microcracking in
surface/MISO La additions effect, 389–407
strain measurements using, 48
weld solidification cracking, 46–48 Neste Oil mock-up weld, Ni-base
DIC technique, 47 alloy, 175–189
extensometers use, 47 Ni-base alloy dissimilar metal welds
laser based DIC system, 47 hot cracking susceptibility, 171–190
Local thermo-mechanics base materials, 174
fully austenitic stainless steel experimental methods, 173–178
design specific influences, filler metals, 174
128–130 Groove geometry, 175, 176
Local weld deformation influence materials, 173–178
on the solidification cracking maximum crack lengths (MCL),
susceptibility 186–188
of a fully austenitic stainless microcracking behavior, 172
steel, 127–144, See also Fully mock-up metal welds, 175–189
austenitic stainless steel Nb and Nb (C, N) content
Longitudinal varestraint test (LVT), influence, 179–189
49, 51 Neste Oil mock-up weld, 175–189
Lu, L., 281, 298 oxide layer on hot crack fracture
Lundin, C.D., 148 surface, 181, 183
Lyons, J.V., 5 Si content influence, 179–189
462 Index

STF DDC susceptibility weld metal characterization,

curves, 172 158–165
total crack lengths (TCL), Ni-base SAW weld metals
186–188 hot cracking susceptibility in,
varestraint tests, 173–178 evaluation, 329–345
Ni-base alloys bend testing, 331, 333,
joining hot crack sensitive, 336–338
215–236, See also SAW cold experimental, 331–338
wire technology externally loaded tests, 331
welding cracks mechanical testing, 329
in brittle temperature ranges, metallographic evaluation, 329,
95–113 332–336
See also ERNiCrFe-7 alloy; MVT, 329, 331–342
individual entries ranking, 342–344
Ni-base filler metals self restrained tests, 331
DDC in, 409–424, See also under test methods and cracking
Ductility-dip cracking mechanisms, 344, 345
experimental procedures, 410–412 Ni-base superalloys
STF test, 409–413 weld solidification cracking
temperature-strain curves, evaluation
412, 413 cast pin tear test, 193–213, See
Ni-base filler metals, solid-solution also under Cast pin tear test
strengthened (CPTT)
weld solidification cracking in, Nishimoto, K., 172
147–170, See also Varestraint Nissley, N., 172
(VAriable RESTRAINT) test Nitrogen alloyed stainless steels, 350
backfilling effect on cracking See also Heat affected zone (HAZ)
susceptibility, 166–169 hot cracking
Calphad calculations, 153, 154, Niyama, E., 5, 6, 21
See also individual entry Non-destructive testing by liquid
characterization, 153 penetrant, 120
equilibrium and non-equilibrium Non-equilibrium solidification and
solidification behavior, SCTR, comparison, 165, 166
comparison, 156, 157 Novik, F.S., 5
experimental procedures, 151–154 Novikov, I.I., 4, 5, 9
materials, 151 Nuclear power plants (NPP)
non-equilibrium solidification hot cracking susceptibility in,
and SCTR, comparison, 171–190, See also Ni-base
165, 166 alloy dissimilar metal welds
Scheil-Gulliver solidification Nucleation, in Al alloys hot cracking,
simulation, 154–160 5, 10, 13
SCTR, 150, 153 Numerical simulation
susceptibility quantification, critical strain rate determination
149, 152 for solidification cracking,
transverse Varestraint test, 77–91, See also under Critical
147–150 strain rate determination
 Index 463

for critical strain rate determination, Prokhorov, N.N., 5, 6, 9, 40, 67, 68,
77–91 70, 82, 128, 278
experimental procedure, 78–80 Propagation, in Al alloys hot cracking,
local strain, 89, 90 5, 10, 13
MVT test, 80, 81 ProPHASE program, 24, 25
numerical approach, 79, 80 Pshennikov, A., 264
strain rates, 87–90 Pumphrey, W.I., 5, 42
temperature field, 82–87
test materials, 78, 79 Ramirez, A.J., 410, 419, 421, 422
weld pool geometry, 82–87 Rappaz, M., 5, 6, 9, 19, 30, 32, 44,
welding parameters, 78, 79 67, 128
Rappaz-Drezet-Gremaud hot tearing
Overlay weld metal criterion, see RDG
microcracking behaviour in, RDG (Rappaz Drezet Gremaud)
393–395 criteria, 44, 45
microcracking mechanism in, in Al alloys welding, 19–36
395–398 alloy hot tearing susceptibility,
improvement mechanism, 23–27
396–398 coalescence, 20
microcracks, characterization, deformation, 20–22
393–394 filler content influence, 27
microstructural analysis with added granular model, 32–35
La, 395, 396 macroscopic approach, 20
microporosity, 20
Patterson, W., 5 microscopic approach, 20
Pellini, W.S., 5, 40 percolation of grains, 20
Pellini, W.S.F., 66 solidification paths, 25–27
Ploshikhin, V., 43, 72 thermal hot tearing criterion,
Porosity formation, and hot tearing in 27–29
Al alloys, 6 thermo-mechanical approach,
Prediction, crack 29–32
weld solidification cracking, 55 coherency point, 19
alloy composition, 55 limitations, 22
composition-strain rate maps, 55 Resistance spot welding of 6xxx Al
numerical simulation, 55 alloys, 311–324
pertinent welding information, 55 See also under Simulation
restraining conditions, 55 investigations
Preheating, to avoid hot cracking, 313 Rindler, W., 23, 67
Primary water stress corrosion Robino, C.V., 51
cracking (PWSCC), 390
Process tape Savage, W.F., 148
resistance spot welding simulation Saveiko, V.N., 5
with a tape, 317–322 SAW cold wire technology, 115–125,
welding with and without, difference 215–236, 329, 330
between, 312 chemical composition of weld
spot welding of Al, 314–316 metal, 120, 121
464 Index

cold wire addition influence on Self restrained tests

welded joint, 218–233 Ni-base weld metal hot cracking
chemical composition, 228–230 susceptibility, evaluation, 331
on the dilution with the base Semi-continuous casting, defects
metal, 219 in, 20
joint welding, 231–233 Senda, T., 40, 278
penetration area, 221 Shape castings, defects in, 20
penetration depth, 219 Shrinkage porosity, high performance
reinforcement area, 221 Ni-base superalloys, 207
weld area, 221 Sigma-jig test, 49
weld cross-section, 221 Sigworth, G.K., 4, 5
weld geometry, 218–222 Silicon role, solidification crack
weld width, 219 initiation in Al, 279–281
welds reinforcement, 219 Simulation investigations
experiment, 117–119 of hot cracking in resistance spot
hardness test, 124, 125 welding of 6xxx Al alloys,
hot cracking tests using MVT-tests, 311–324
234–236 finite element method (FEM), 311
industrial implementation, 236 high strength Al alloys, 313–317,
macro test of welds, 122–124 322–324
microstructure of welds, 124 mathematical models, 319, 320
non-destructive testing by liquid parameter optimization process,
penetrant, 120 321, 322
structure examination, 222–228 resistance spot welding simulation
grain boundaries, 223, 224 with a tape, 317–322
macro-sections, 225 solver strategy, 320
volume change, 226 SORPAS®, 311, 316, 318
testing equipment, 216–218 virtual simulation experiments,
welding procedures, 120 311, 321
wire positioning, parameters, Wanheim and Bay’s friction
217, 218 theory, 319
See also Ni-base SAW weld metals Singer, A.R.E., 5, 281
Scheil equation, 53 Slow bend trans-varestraint test, 44
Scheil-Gulliver solidification Solidification crack initiation in Al
simulation, 154–160 weld parameter and minor element
equilibrium solidification effects on, 277–304
temperature ranges and CTW test, 277, 278
solidification sequence, 157 alloy heat/temper effect, 294, 295
Second-generation CPTT, 195–199 crack-no crack boundary
hearth design, 196 formation, 278, 291, 292, 303
mold, 196, 197 critical strain rate, 278, 284, 285
test apparatus, 195 critical strain rate–dilution,
test procedure, 196 284, 285, 291, 292
Segregation hot cracks, 96 extensometer position, 301–303
Segregation induced cracking, 116 extrusions used, 288, 289
 Index 465

filler dilution amount, hot tearing and, 4

importance, 278 initiation, 39, 40, See also Weld
grain refinement, 282, 283, solidification cracking
298, 299 Solidification cracking susceptibility
hydrogen contamination, high performance Ni-base
283, 284, 301 superalloys, 204, 205
insert control test, 297 local weld deformation influence on
inserts used, 289, 290 fully austenitic stainless steel,
interacting factors, 303 127–144, See also Fully
iron role, 281, 282, 298 austenitic stainless steel
local strain extensometer Solidification cracking temperature
measurements, 277 range (SCTR), 50, 150, 153
local strain measurement, in Ni-base filler metals weld
287, 288, 294 solidification cracking
negative strain rate, 303 Calphad calculations and,
oxygen contamination, 283, 284, comparison, 153–157
299, 300 non-equilibrium solidification
shielding gas used, 291 and, comparison, 165, 166
silicon role, 279–281 Scheil-Gulliver solidification
strain and stress fields around simulation, 154, 157, 158
weld pool, 283 Solidification microstructure, high
temperature measurement, 288 performance Ni-base
tensile stress and strain, 278 superalloys, 205–207
weld pool shape characterization, Solidification path, weld solidification
292, 293 cracking, 54
weld travel speed, 283, Solid-solution strengthened Ni-base
295, 296 filler metals
Solidification cracking weld solidification cracking in,
assessment, 66–72 147–170, See also Ni-base filler
computational procedure, 71 metals
concepts, 68–72 SORPAS® FEM software, 311, 316, 318
criteria determined by parameter optimization process,
constitutional solidification, 321, 322
66, 67 solver strategy, 320
criteria determined by structural Spot welding
mechanisms, 67 of Al with and without a tape,
modelling mechanisms and 314–316
criteria, 66 to avoid hot cracking, 313, See also
characterization tools, 62–66, See under 6xxx Al alloys
also under Weldment hot cracks in 6xxx alloys during,
cracking 316, 317
critical strain rate determination Stableaustenitic welds, 95
by numerical simulation, 77–91, Steady-state creep
See also under Critical strain types, 107
rate determination diffusion, 107
in the high-temperature region, 97 dislocation, 107
466 Index

Strain critical strain, 40–45

causing hot tearing in Al alloys, 6 ductility curve concept in, 40–42
monitoring numerical simulations, 43
fully austenitic stainless steel, Thermo-physical mechanism
133, 134 in weldments cracking control,
strain rate partitioning, 53 59–73, See also Weldment
strain-to-fracture (STF) technique, cracking
172, 409–413 Total accumulated crack length
See also Critical strain (TLC), 49
Stress, causing hot tearing in Al Transverse varestraint test (TVT), 49,
alloys, 6 51, 147–150
Stringer geometry, 257–262 Tsai, C., 172, 189
modification, 260
Structural behavior and weldability van Haaften, W.-M., 14
optimization, 257–275 Varestraint (VAriable RESTRAINT)
See also Integral approach test, 52, 148, 152, 153,
Structural mechanisms, criteria 173–178, 188
determined by, 67 sample preparation, 151, 152
Submerged arc welding spot-Varestraint tests, 172
see SAW cold wire technology transverse Varestraint test, 147, 148,
Surface profile induced cracking, 161–165
116, 117 maximum crack distance
Suyitno, 5, 9 (MCD), 161
test conditions, 153
Temperature dependent interface total number of cracks
element, 71 (TNC), 161
Temperature induced weldment Variable deformation rate test
cracking, 60 (VDR), 49
Thermal cycles, and weldment Variable speed varestraint, 49
cracking, 60 Variable tensile strain test, 49
Thermal hot tearing criterion Vernède, S., 33
in Al alloys welding, 27–29 Verö, J., 5
filler content influence, 27 Virtual simulation experiments,
incompressible model, 29 311, 321
Thermal strain analysis Voronoï tessellation, 34
during multipass laser overlay feeding KPL model, 34
welding, 401, 402 microsegregation model, 34
Thermal-deformation processes, 111
ThermocalcTM software, 153, 154 Wanheim and Bay’s friction
Thermo-mechanical analysis theory, 319
in Al alloys welding, 29–32 Wanheim, T., 319
Von Mises equivalent plastic Weld solidification cracking
strain rate, 31 crack initiation and growth
weld solidification cracking, 40–46 critical conditions, 39–56
crack growth, 45, 46 thermo-mechanical analysis,
crack initiation, 40–45 40–46, See also individual entry
 Index 467

crack prediction, 55 strain rates, 51

grain boundary segregation, 53 transverse varestraint test (TVT),
local strain, measuring, 46–48, See 49, 51
also Local strain measurement varestraint tests, 52
weldability testing, 49–52, See variable deformation rate test
also individual entry (VDR), 49
in Ni-base filler metals, 147–170, variable speed varestraint, 49
See also Ni-base filler metals variable tensile strain test, 49
role of impurities, 54 Weldment cracking, 60–66
solidification path, 54 classification, 61, 62
strain rate partitioning, 53 fracture mechanics role in,
Weldability testing, 49–52, 63, 64 64–66
circular patch test (CPT), 49 solidification cracking,
cont. tensile weldability test characterization, 62–66
(CTW), 49 FEA, 62, 63
crack length measurements fractography, 63
evaluation, 49 microscopy, 63
crack length parametrics, 49–52 weldability tests, 63, 64
extrinsic, 49 temperature-induced, 60
houldcroft test, 49 Williams, J.A., 5
intrinsic, 49 Wu, W., 172, 189
longitudinal varestraint test (LVT),
49, 51 X-ray cone beam computer
modified varestraint test (MVT), 49 tomography, 349, 364–367
program. verformungsrisstest (PVR),
49, 63, 64 Zacharia, T., 129, 138, 143
sigma-jig test, 49 Zaharia, T., 71