Hot Cracking Phenomena in Welds

BAM, Berlin, March 4-5, 2004


Testing for Susceptibility to Hot Cracking on GleebleTM Physical Simulator


Dr. Stan T. Mandziej, Advanced Materials Analysis,

P.O.Box 3751, 7500DT Enschede, The Netherlands.


Hot cracks appear when thermal shrinkage together with deformation caused by restraint cannot be accommodated by plastic deformation. This happens during welding to such alloys, which segregate on heating and cooling at near-solidus temperatures, in particular when low-melting and mechanically weak phases form and occur over a wide range of temperatures. To check for susceptibility to the liquation cracking caused by the low-melting, weak phases, hot tensile testing can be used in combination with a thermal cycle resembling that of real welding. This procedure, which can be executed on a GleebleTM thermal-mechanical simulator, comprises tensile testing of a number of cylindrical samples at the temperatures below solidus and determining their hot strength and ductility. For measuring of the brittle temperature range (BTR), the nil strength temperature (NST) is determined and used as the peak point, down from which the ductility recovery temperature (DRT) is searched for. The ductility is measured after the tensile test as a reduction in area at fracture. An alternative to the hot tensile test during the simulated welding cycle is the strain-induced crack opening (SICO) test, in which a rod-like sample mounted in “cold” copper jaws of the Gleeble is heated by electric current and then compressed till formation of a bulge in its uniformly heated central portion and appearance of cracks due to secondary tensile strain developed along the maximum perimeter of this bulge. Next to the studies of liquation cracking and ductility-dip cracking of the reheat-type, the Gleeble procedures can be also used to determine sensitivity to solidification cracking and for this the samples are melted and solidified in a controlled manner before the hot tensile testing or the SICO testing. The developed routines of the Gleeble testing allow accurate determination of temperatures at which the cracks occur as well as measurement of critical strains to fracture and strain rates which are associated with the hot cracking. At an appropriate geometry of samples and optimum setting of the experiments, the hot cracks generated in the samples have sizes comparable to those that occur in heat-affected zones of real welds or in weld metals.

TM - Gleeble is a registered trademark of Dynamic Systems Inc., Poestenkill, NY, USA.

1. Introduction

To date, over 150 weldability tests exist, many of which are designed to assess the susceptibility of welds to hot cracking. In general, they can be put into two categories as representative (self-restraint) and simulative (externally loaded) test techniques [1]. The representative test technique usually tells only ‘cracking’ or ‘no-cracking’ of a material when an actual welding situation is represented, which cannot quantify the cracking susceptibility of the material under different welding condition. The simulative test can follow a thermal-mechanical history of a material during welding, while an external strain is usually applied to produce cracks or to record material characteristics allowing quantification of cracking susceptibility.

Hot cracking of welds appears when shrinkages of the solidifying and thus stiffening weld metal and of the adjacent heat affected zone cannot be compensated or accommodated by at first the back-filling (liquid flow) of the molten material and afterwards by the visco-plastic flow of the solidified hot material of the welded joint. The process is dynamic and to avoid hot cracking the welded material must be able to plastically deform at critical sites of the welded joint with the strain rate exceeding that of the shrinkage during the relatively fast heat cycle of the welding.

Studies in the 1950's carried out in the USA by W.F. Savage at Rensselaer Polytechnic Institute and at Duffers Associates Inc., led to the development of the Gleeble thermal-mechanical simulator which allowed the hot ductility of welded materials during thermal cycles of the welding processes to be determined [2]. This study showed that the region most susceptible to hot cracking is the heat affected zone of the parent metal, in which contaminants entrapped at grain boundaries form liquid or low strength solid films while the grains become stiff and strong. It was also found that if such weak films exist over a large temperature range after solidification, the welded materials show hot cracks in the HAZ. To determine the range at which the weld HAZ is prone to hot cracking, a concept of nil strength temperature was introduced as the higher temperature of the brittle range, and appropriate attachments were designed to measure it. The lower temperature of the brittle range, so-called nil ductility temperature, was then taken as that at which 5% reduction in area on hot tensile samples appeared [3].

The Gleeble testing procedure required a large number of samples to be hot tensile tested with strain rates representative of various welding methods (heat inputs), and this stimulated a study to develop another, simpler test, compatible with the deformation rates already known from the Gleeble testing. As a result, the Varestraint test was proposed in the early 1960's by W.F. Savage and C.D. Lundin [4], and applied to study the hot cracking susceptibility of welded alloys. The Varestraint test comprises bending a test plate while the weld bead is being made on the long axis of the plate. The original Varestraint test had some limitations, e.g. difficulty in controlling the real amount of strain at the outer bent surface due to the position of a neutral bending axis, which varied depending on the strength and strain partitioning between the hot and cold parts of the sample during bending.

The Gleeble testing method is dealing with the occurrence of the brittle temperature range (BTR) at high temperature during the solidification of the weld and the subsequent cooling. The BTR reflects the general appearance of the cracks occurring in heat-affected zones of welds, Fig.1, and can be illustrated on the graph proposed by Prokhorov [5], Fig.2. The graph shows that for any welded alloy there is some critical strain value, which can be accommodated at very low cooling rates by, for example, self-diffusion controlled visco-plastic flow of the hot metal. The crack in HAZ does not form exactly on the fusion line between weld metal and base metal; there is always certain space in which high diffusion rate and back-filling of the nucleating cracks or voids allow healing of the base metal. The hot cracks in HAZ occur in a small distance from the fusion line when the strain rate caused by shrinkage during cooling of the weld is sufficiently high for the solidifying material to "enter" the BTR. The wider the BTR of the alloy, the higher the hot crack susceptibility.

2. What is a Gleeble?

The thermal-mechanical simulator called Gleeble originated from the welding thermal cycle simulator built in 1948 at Rensselaer Polytechnic Institute in Troy, NY, USA, which soon after (1951) was equipped with pneumatic deformation system and more recently with dynamic servo-hydraulic system (1979) and computer controls (1980). Its developer, Dr. W.F. Savage, wrote in one of his review articles [6]: “In 1946 Nippes and I attempted to evaluate the notch toughness of various regions of the heat-affected zone of ship steel weldments by careful placement of the notches in Charpy V-Notch specimens machined from weldments. However, the results were biased by the sloping nature of the heat-affected zone, which caused the fracture to traverse more than one portion of the zone. Consequently, we decided to develop a thermal simulator and utilize temperature measurement data accumulated during the earlier cooling-rate studies at Rensselaer.”

By its design, Gleeble has been dedicated to reproduce weld thermal cycles characteristic of the heat-affected zones with (all) consequences) related to thermal-mechanical effects caused by thermal gradients, restraints and shrinkage deformations. For this the up-to-date Gleeble uses “bulk” samples of 10mm or more diameter and in appropriate parts of the thermal cycle can exert tensile or compressive deformations with strain rates adequate to those of the real welding process. The amounts of deformation and strain rates resulting from cooling rates and thermal gradients can be calculated using contemporary computer modeling techniques. The last should take into account apparent strengthening caused by thermal gradients and strain rates [7].

The thermal-mechanical situation in HAZ during welding comprises in the thin element of the base metal parallel to the fusion plane and longitudinal to the welding direction twice crosswise and lengthwise compressions on heating and on cooling and once on cooling a tension in the direction normal to the fusion plane, which is also the main direction of the heat flow. For the electric arc welding it also involves the presence of electric current and related electro-thermal effects. Gleeble accounts for these by using AC electric resistance heating uniform through the cross section of the specimen. This situation is schematically presented in Fig.3. To simulate the HAZ in Gleeble, rod-like specimens mounted in water-cooled copper grips are heated at a rate resembling the weld thermal cycle and a short span between the grips assures that the heated zone is about twice the width of a real weld HAZ, Fig.4. When necessary, controlled tensile or compressive strains can be added in any point of the thermal cycle. The thermal cycle in the Gleeble is controlled by a thermocouple, percussion welded to the surface usually in the middle of the specimen.

Fig.2 Schematic presentation of the brittle temperature range (BTR) below solidus temperature (Ts) relating the hot crack susceptibility to critical strain (CS) and strain rate (CST)

Fig.1 Weld HAZ liquation crack appearing in a small distance from the fusion line i.e. nucleating at nil strength temperature (NST) lower than the solidus temperature Ts; marked ductility recovery temperature (DRT) near which the crack is closing

The ability of Gleeble to deform heated specimens can be effectively used for hot tensile testing. The length of the uniformly heated zone in the middle of the Gleeble’s specimen results from the balance between electric current heating the specimen and the heat flow towards the mounting grips. Depending on the span between the grips, as well as on conductivity of the grip material and on size of the contact area between the specimen and the grips, this length can be manipulated to a reasonable extent. The following two pictures show an example of the “hot” Gleeble jaws made of stainless steel, Fig.5, and the resulting substantial length of the uniformly heated zone on the specimen mounted in such “hot” jaws, Fig.6.

Fig.3 Schematic presentation of the heat affected zone formed during electric arc welding with heat flow from the weld metal to parent plate causing thermal gradient

Fig.4 Simulation of HAZ situation on a rod-like specimen mounted in water-cooled copper jaws of the Gleeble

Fig.6 Elongated uniformly heated zone on a plain carbon steel specimen mounted in the “hot” stainless steel jaws of the Gleeble

Fig.5 Stainless steel “hot” jaws of Gleeble for mounting of hot tensile specimens; note reduced contact portion between the jaws and the specimen

Hot ductility and hot cracking

As the hot cracks form on-cooling when tensile strains caused by shrinkage and assisted by restraint cannot be compensated by ductility of an alloy, then method of studying susceptibility of an alloy to hot cracking should involve tensile testing at the conditions simulating these of the real welding (or casting) process. The use of thermal-mechanical simulator like Gleeble, able to reproduce on a specimen the welding thermal cycle and impose a strain in a controlled manner, allows achieving this goal.

Testing on a Gleeble for susceptibility to weld liquation cracking / HAZ hot cracking, means hot tensile testing of a number of specimens on-heating and then on-cooling, and determining their hot ductility measured as a reduction in area at the specimen’s neck portion after the test. This procedure is schematically presented in Fig.7 [1].

The hot ductility of a weldable alloy increases gradually with increase of testing temperature from ambient towards melting point, however before reaching this point it drops abruptly from certain maximum to nil. Just above this nil ductility temperature (NDT) appears the nil strength temperature (NST) at which the alloy looses its strength due to formation of weak or liquid phases along grain boundaries. The real physically measurable melting temperature of such alloy –TL, is higher than NST, as shown in Fig.7. On-cooling from the melt or from the NST, the ductility does not recover exactly at NDT but below it at so-called ductility recovery temperature – DRT. The temperature span from NST to DRT is considered to be the brittle temperature range – BTR.

The extent of BTR can be used as a rough criterion of the susceptibility to hot cracking, however more exact criterion is the measure how fast does the ductility recover on-cooling as compared with its decrease on-heating. As the reference point for this measurement the maximum of ductility from the on-heating ductility curve is taken and the representative areas below the on-heating and on-cooling curves are compared, Fig.8 [8]. Arbitrarily considering 5% of reduction-in-area on-cooling as the ductility recovery point and comparing the hot ductility curves on-heating and on-cooling, the nil ductility range (NDR = BTR), ductility recovery rate (DRR) and ratio of ductility recovery (RDR) can be determined to exactly characterize the susceptibility of an alloy to hot cracking.

Fig.7 Schematic of Gleeble’s procedure for hot ductility testing, including hot tensile testing on-heating up to NST and then on-cooling after weld thermal cycle with NST as the peak temperature; note the NST being lower than the melting point TL

Fig.8 Evaluation of hot ductility curves for hot cracking susceptibility [8];

- RDR = ration of ductility recovery,
- DRR = ductility recovery rate,
- NDR = nil ductility range (or BTR),
- Point F = nil ductility temperature (NDT) on-heating,
- Point D = ductility recovery temperature on-cooling.

The reference point for the ductility measurements determining the susceptibility to liquation cracking is the nil strength temperature. This temperature has to be used as a peak of the welding thermal cycle, on-cooling after which the hot tensile tests should be run. To measure the NST on Gleeble an attachment is used, schematically presented in Fig.9. This NST attachment keeps the specimen under a constant tensile load of about 50N while allowing specimen heating with an initial heating rate the same as of the welding thermal cycle to be simulated up to a temperature 50oC below the solidus temperature of the material, then change to a heating rate of 2 - 5oC/s until the NST is reached, as shown in Fig.10.

To avoid mixing of phenomena related to liquation cracking with those of solidification cracking, in the weld thermal cycles simulated to measure hot ductility for the liquation cracking the NST must not be exceeded.

To test for the susceptibility to solidification cracking, controlled melting and solidification of a rod-like specimen is carried out on Gleeble, and after the solidification the hot ductility is determined in the manner as described above. Here the ability of Gleeble to melt electrically conductive specimens is used. In this test the central portion of the specimen protected by a crucible / quartz sleeve is brought to a temperature above solidus and then this crucible contains the molten / semi-liquid metal, as shown in Fig.11. The thermal gradient between the molten portion and mounting jaws prevent the metal from flowing out of the crucible while controlled thermal cycle allows conducting the solidification in a manner similar to that of real casting or welding. Schematic of the assembly used for this purpose in Gleeble is given in Fig.12.

More details of the Gleeble testing for liquation and solidification cracking can be found in a Gleeble application note [9] available upon request from DSI in USA ( The Gleeble hot tensile testing procedures for liquation and solidification cracking are also mentioned in the technical report CEN ISO/TR 17641-3:2003, under the chapter “Hot Tensile Test” [10].

Fig.9 Nil strength attachment of Gleeble using trapezoidal spring grip maintaining constant small tensile load on specimen during heating

Fig.10 Schematic time–temperature graph of Gleeble test for nil strength temperature measurement

An alternative procedure to study susceptibility to solidification cracking on Gleeble is the strain-induced crack opening test - SICO, developed at Dynamic Systems Inc., in USA, for studying hot deformability of alloys. In this test the central hot portion of the Gleeble specimen is compressed to form a bulge, Fig.13, on outer perimeter of which cracks appear at the critical secondary tensile strain, Fig.14. The critical strain to fracture in a SICO test is defined as the hoop strain at onset of cracking in the bulge zone:

εc = ln (Df / Do)

where Do is the initial diameter of the specimen, while Df is the final maximum diameter in the bulge zone.

As during the controlled melting and solidification in the Gleeble the dendritic crystals grow mainly in the direction of the heat flow i.e. in the axial direction, the mid-span segregation may occur in the specimen causing deep and partly hidden cracking of the SICO specimen along the central plane perpendicular to the compression axis, Fig.15. In such situation it is advised to check for the true critical diameter on the cross-section of SICO sample, like it is shown in Fig.16.

As the hot tensile testing on Gleeble gives the adequate characteristics of an alloy regarding its hot cracking susceptibility, the SICO test appears to be more accurate for measuring of critical strains to fracture and related critical strain rates.

Fig.12 Schematic of Fig.11, with end nuts for hot tensile testing

Fig.11 Gleeble setup for controlled melting and solidification study

Fig.14 Schematic of SICO test showing crack formed on bulge portion of sample

Fig.13 SICO sample tested in Gleeble

3. Hot Cracking of Austenitic Steels and Alloys

Austenitic stainless steels, Ni-base alloys and other metal alloys having large thermal expansion coefficients are susceptible to hot cracking during welding. In most of these alloys the high content of alloying elements extends their solidification temperature range thus increasing segregation of impurities along grain boundaries. These alloys are susceptible to solidification cracking in weld metal and to liquation cracking in the HAZ, depending on the relative strength of the HAZ and weld metal in the hot range, and to avoid cracking both the HAZ and weld must be able to plastically accommodate the shrinkage strains.

The hot strength of austenitic stainless steels and Ni-base alloys below solidus is almost zero in a certain range of temperature till NST, and below the NST the metal becomes rigid and strong but its ductility is still be very low. During cooling through the low ductility range, the thermal shrinkage may cause hot cracks, and the susceptibility of steel to this cracking can be expressed as a difference between the NST and the ductility recovery temperature - DRT. Below the DRT, the steel is able to plastically accommodate the shrinkage thus avoiding cracks. To determine the hot ductility for some austenitic alloys and their susceptibility to hot cracking, Gleeble hot tensile tests were carried out at DSI for a collaborative program of IIW – Commission II, of which some results are presented here. The full report of this testing [11] is available from DSI upon request at Four austenitic steels and alloys of the composition given in Table 1 were examined. Test specimens were 6mm diameter by 100mm long. They were heated to fracture in a nil-strength test fixture under a load of ~8kgf. The measured NST for each alloy, as well as the determined BTR and RDR, are listed in Table 2.

The results, given in Table 2, have shown a substantial difference in the behaviour of Alloy 926 and 800H, and they also showed a good agreement with the hot cracking susceptibility data available from the modified Varestraint test and from practical observations [12].

Fig.16 Method of measuring critical diameter Dcr when central plane crack appears

Fig.15 Central plane crack in SICO often appearing after melting and solidification

Table 1 Chemical composition of tested alloys, Fe is balance to 100%


chemical composition (wt%)


































Nb=.83, Ce=.085












Cu=.91, N=.196













Table 2 Nil-strength temperature (NST), brittle temperature range (BTR) and ratio of ductility recovery (RDR) for each alloy




Alloy 926

Alloy 825






BTR (oC)





RDR (%)





Selected examples from these hot strength / hot ductility tests are given below on the graphs of Alloy 926 and Alloy 800H (Figs 17 & 18), and one may see from them that the real hot ductility curves are not exactly so “smooth” like these in Figs 7 & 8. The overlapping and competing effects of liquation and homogenization may result in a solid-state diffusion healing dependent on applied time / temperature / strain rate of the executed hot tensile test. Thus, for the applied strain rate representative to the shrinkage rate of the weld thermal cycle ductility dips were observed at 1240oC and 1170oC for alloys 800H and 926 respectively, while for slower strain rates these dips were absent. This last observation calls for metallographic verification of the tests, in particular for comparing the microstructures of test specimens with these of real welds containing hot cracks and for identifying the micromechanisms of cracks formation at the conditions of the hot tensile tests.

Fig.18 Hot strength and hot ductility curves of alloy 926

Fig.17 Hot strength and hot ductility curves of alloy 800H

What follows, is a part of metallographic verification of the specimens from the mentioned IIW testing programme, which revealed various microstructural features coinciding with the microstructures of the real welds as well as some discrepancies further used to modify the Gleeble testing conditions.

In Alloy 926, very well resistant to hot liquation cracking, substantial grain growth appeared near to NST however no grain-boundary segregation / liquation could be noticed. Samples tested at 1250oC showed existence of long segregation bands filled with weak / low-melting eutectics, Fig.19, however when tested at 1200oC these bands almost entirely disappeared due to diffusion healing, Fig.20.

In a more susceptible to hot cracking alloy 800H, a rapid grain growth occurred near to the NST, Fig.21, which resulted in formation of thin films as well as voids along grain boundaries, Fig.22.

Also chains of equiaxial voids appeared along segregation bands, however at the highest testing temperatures these voids did not easily coalesce to cause the failure – the weakest link of the microstructure appeared to be the grain boundaries, Fig.23. At temperatures lower than NST, i.e. at 1250oC and below, elongated cracks appeared along the segregation bands and from them intergranular cracks extended in transverse directions, Fig.24. The behaviour of alloy 800H indicated that various micromechanisms could operate at different testing temperatures and that the hot crack susceptibility is strain rate dependent.

Fig.20 Isolated longitudinal cracks near to fracture zone of alloy 926 tensile specimen tested at 1200oC

Fig.19 Longitudinal cavities containing eutectic near to fracture zone of alloy 926 tensile specimen tested at 1250oC

Fig.22 Formation of grain boundary films and cavities in alloy 800H near to NST

Fig.21 Abnormal grain growth near to surface of specimen in alloy 800H at nil strength temperature (NST)

It was possible to find on the longitudinal sections of the hot tensile specimens the regions at which the highest density of intergranular transverse cracks appeared and using the formula:

ε = 2 ln (D/d)

to determine the local critical strains as well as strain rates for the hot cracking at particular temperatures. An example below, Fig.25, shows the section through neck portion of hot tensile specimen, on which this relation is drawn and the following two pictures, Figs 26 & 27, present cracks, which appeared in alloy 800H at temperature 1200oC and strain rate 1.20/sec while no cracks formed at the same temperature and strain rate 0.30/sec.

Fig.23 Cavities in segregation bands close to fracture in alloy 800H tensile tested at NST

Fig.24 Longitudinal cracks along segregation bands and transverse intergranular cracks in alloy 800H tested at 1200oC

Fig.25 Method of determining strain from the neck portion of tensile test specimen:

ε = 2 ln (D/d)

Fig.27 No cracks visible in tensile test specimen of alloy 800H tested at 1200oC, at region deformed with strain rate ~ 0.30/sec

Fig.26 Cracks revealed in tensile test specimen of alloy 800H tested at 1200oC, at region deformed with strain rate ~ 1.20/sec

In another case, of alloy AC66, the loss of ductility due to liquation was quite severe and the recovery of ductility on-cooling delayed. Here, the liquation caused a permanent precipitation of a carbide eutectic, which evidently hampered deformability of the alloy. The precipitates appeared in segregation bands, Fig.28, assisting in the hot tensile test formation of transverse cracks nucleating from voids at these precipitates. In Fig.29 carbide eutectic is shown, which separated during the test forming a deep transverse void. These precipitates in alloy AC66 visibly assisted the hot cracking for more than 100oC down from the NST, mainly by the formation of voids around them.

Fracture of the hot tensile specimens near NST occurred mainly by joining of these voids in the transverse direction thus indicating additional weakness of grain boundaries at this temperature, Fig.30. The eutectics appeared elongated in the rolling direction of the tested material, Fig.31, suggesting its metallurgical low quality. Nevertheless, at lower temperatures a substantial recovery of ductility appeared, due to thermal-mechanical grain refinement, Fig.32, most probably resulting from a dynamic recrystallization in the regions of strain localization and concentration between the elongated fields of the eutectics. This last phenomenon “successfully” competed with the formation of voids near to the eutectics, giving a high value of reduction-in-area and simultaneously generating in the neck portion of the specimen a large amount of voids surrounded by fine recrystallized grains, Fig.33.

Fig.29 An example of a cavity formed during the hot tensile test across the carbide eutectic in alloy AC66

Fig.28 Carbide eutectic near to fracture and on the fracture surface of hot tensile specimen of alloy AC66 tested at 1225oC

Fig.31 Elongated carbide eutectics in alloy AC66 after heating up to NST and cooling down; total strain ~0.05

Fig.30 Extended transverse crack near to the fracture portion and numerous voids in alloy AC66 tensile tested at NST

In conclusion it may be said that the metallurgical quality of an alloy is important as regards its susceptibility to hot cracking and that sometimes the “standard” test conditions may need to be adjusted in order to reach correct results.

4. Microfissuring in Multi-Bead Welds

Microfissures are reheat-type fine cracks of length about 1mm or less when visible on transverse sections of welds and up to a few millimeters in the length direction of the weld. They are often of a ductility-dip origin and form in inter-bead heat affected zones of multi-layer, multi-bead welds, almost exclusively in the upper layers of these welds, i.e. when underlying portion of the weld is stiff enough to provide adequate restraint.

The microfissures are related to thermal-mechanical history of the weld manufacturing and assisted by the primary segregation of weld metal solidification and the secondary reheating liquation as well as by solid-state embrittling processes resulting in ductility dips. They form in the multi-pass welds in zones of adjacent (underlying) weld beads, which constitute the heat-affected zone of a subsequent pass; an example is given in Fig.34.

To explain the formation of microfissures the following model [13] can be used, Fig.35, in which the inter-bead heat-affected zones are divided into two portions: the first marked (a) representative to higher temperature and coarse grain microstructure and the second marked (b) representative to fine-grained microstructure and incomplete recrystallization. While welding with the bead sequence: (1) (2) (3) (4) (5) (6) and so on, after laying down the bead #(6) the most potential site for microfissure formation will be reheated by the bead #(6) coarse-grained zone of the HAZ below the bead #(5) at the site where it overlaps with the HAZ of the bead #(6). The (a) and (b) zones of the interbead HAZs can be also treated as the upper one (a) in which the diffusion healing and the nil-ductility appear, and the lower one (b) in which the ductility has recovered substantially however strain hardening may occur. Thus at the bottom of the lower layer the thermal cycling may results in a slight strengthening due to the generation of dislocation and their incomplete annihilation / recovery, without any substantial annihilation of the primary crystalline lattice defects being mainly vacancies in the “as-frozen” weld metal. In general, the microstructure of the as-solidified weld metal beads is far from thermodynamic equilibrium by containing large amounts of crystalline lattice defects, such as vacancies and dislocations as well as planar defects, which are formed to compensate shrinkage effects and which tend to annihilate during subsequent thermal cycling. Thus, when during the laying down of the bead #(6) such microstructure in the bead #(5) is heated up to a temperature near to solidus, a large number of defects annihilate and this process is affected at first by thermal expansion of the whole interbead HAZ, and then by shrinkage restrained in some particular directions. An intensive annihilation of vacancies and grain growth causing local reduction of volume is expected to occur during heating in the upper layer (a6) of the HAZ by the overlying bead (6), i.e. close to the fusion line, due to restrained expansion. Then on cooling, this zone shrinks and may be particularly prone to nucleation of cracks if its ductility is low. However, the simultaneous shrinkage of the already solidified weld metal in the bead #(6) acts to compensate this effect. The expansion and its resulting compressive strain may have a positive effect that is the acceleration of homogenization and partial healing in the upper layer (a6) of the HAZ, and these may reduce the sensitivity to liquation cracking. Additionally, the shrinkage after solidification of the weld metal bead #(6) may close partly the liquation cracks in the upper layer (a6) of the HAZ. Depending on which effect prevails, the sensitization occurs or the microfissures nucleate in the upper or in the lower layer of the interbead HAZ. The critical sites for their formation are intersections with the coarse grained part of the HAZ (a5) of the previous bead #(5) where laying down of this bead caused the liquation. In such site which is relatively remote from the centre ‘C’ of the weld bead #(6) as is visualized in Fig.35, the shrinkage of the weld bead #(6) after its solidification and “anchoring” between points ‘A’ and ‘B’, can result in localization of a substantial tensile strain, thus opening the microfissure in the a5-b6 area. The location of microfissures at the intersection with the HAZ of previous beads and frequent propagation along boundaries of primary columnar grains into the underlying bead, indicate that the segregation and liquation are important factors for their formation.

Fig.33 Refined grains and numerous voids formed in the neck portion of hot tensile specimen of alloy AC66 deformed with strain rate ~ 2.50/sec at 1175oC

Fig.32 Refined grains and coarse primary grains formed next to the precipitates of eutectics in alloy AC66 at the portion of tensile specimen deformed with strain rate ~ 0.50/sec at 1175oC

To study the susceptibility of the multi-bead weld joints to microfissuring as well as the influence of microfissures on hot strength and ductility of the weld metals, SICO testing can be used on Gleeble. Samples for such testing are cylindrical bars of 10mm diameter by 90mm long, and are clamped from both sides in the Gleeble’s cold jaws. The simulated weld heating cycle is then applied and the sample can be compressed immediately at the peak temperature or at any temperature on cooling from the peak. To check for the presence of microfissures and their influence on hot ductility SICO tests were run on samples taken from the upper half and lower half of an austenitic stainless steel multi-bead weld, and the cylindrical bar samples were machined transversely to the weld length, like schematically presented in Fig.36. Then in the Gleeble the samples were heated at a heating rate of 150oC/sec to a test temperature, and compressed at a speed of 50mm/sec to different reductions. The mean strain rate was approximately 3/sec at a free span of 30mm.

Fig.34 An example of microfissure(s) appearing in inter-bead heat-affected zone of a multi-bead weld, visible on a cross-section of the weld

Fig.35 Schematic presentation of the multi-bead weld with sites where the microfissures do form;

(1)÷(6) - sequence of weld beads,

(a1+b1) ÷(a6+b6) - inter-bead heat-affected zones

For the SICO test specimens cut from the top layer of the multi-pass weld metal, critical strains were lower at each temperature at the same testing condition than that from the bottom layer with rare microfissures, as shown in Fig.37. Moreover, it was found that all the cracks appeared at the specimen surface directed to the top of the weld, which indicates the presence of microfissures in the top layers of the multi-pass weld joint. The microfissures simply act as nuclei or microcracks for crack opening during compression under secondary tensile stresses. This conforms with the applied physical model of microfissuring. It has been concluded that the strains at which the weld metal cracked at the temperature and strain rate of the SICO test, or the microfissures extended to the surface of the SICO sample and appeared visible, can be treated as characterizing the susceptibility of weld metal to microfissuring.

5. Summary and Conclusions

The examples of Gleeble thermal-mechanical simulation procedures presented in this paper, show the results of simulation that can be achieved if the fundamental physical phenomena occurring during welding is well understood and obeyed. They highlight an important role of thermal gradients, which occur in all known industrial thermal processes and resulting from these gradients, thermal-mechanical strains and strain accommodation phenomena which in turn affect strain hardening and recrystallization, phase transformations and/or precipitation processes. The combinations of different factors, and their importance, may vary substantially, and the correct physical simulation of welding must take into account a number of interacting phenomena appearing in the real application process, such as:

Fig.37 Results of SICO testing for microfissure susceptibility of multi-bead austenitic stainless steel weld metal [13]

Fig.36 Schematic presentation of SICO specimens’ extraction from lower and upper portions of multi-bead weld

1. The balance between the heat input and electric current flow during heating, and controlled by thermal gradient heat flux, cooling rate and micro-deformation rate during the short thermal cycle of welding, in the case of heat-affected zone during arc welding.

2. The changes of hot ductility and of hot strength of welded material on heating and on cooling and its susceptibility to form liquid phase along grain boundaries at elevated temperatures below solidus, in the case of hot cracking.

3. The accommodation of strains occurring due to multiple welding thermal cycles and due to annealing microstructural restraints in multi-bead austenitic weld metals, in the case of micro-fissuring.


[1] Lin W, Lippold JC, Baeslack III WA (1993) ‘An Evaluation of Heat-Affected Zone Liquation Cracking Susceptibility, Part I: Development of a Method for Quantification’, Welding Journal 72, Res Suppl, pp 135-153.

[2] Savage WF (1955) ‘An Investigation of the Hot Ductility of High-Temperature Alloys’, Welding Journal 34, Res Suppl, pp 183-196.

[3] Nippes EF, Savage WF, Grotke G (1957) ‘Further Studies of the Hot-Ductility of High-Temperature Alloys’, Welding Research Council, Bulletin 33.

[4] Savage WF, Lundin CD (1965) ‘The Varestraint Test’, Welding Journal 44, Res Suppl, pp 433-442.

[5] Prokhorov NN (1968), ‘Theorie und Verfahren zum Bestimmen der Technologischen

Festigkeit von Metallen beim Schweissen’, Schweisstechnik 1, pp 8-11.

[6] Savage WF (1987), ‘A Historical View of Weldability’, in Welding Metallurgy of Structural Steels, TMS, Warrendale PA, USA, pp 3-19.

[7] Ferguson HS (1992), ‘Fundamentals of Physical Simulation’, in Proceedings of the International Symposium on Physical Simulation, TU Delft, The Netherlands, pp.1-21.

[8] Lundin CD, Qiao CYP, Gill TPS, Goodwin GM (1993), ‘Hot ductility and hot cracking behavior of Modified 316 Stainless Steels designed for high temperature service’, Welding Journal 72, Res Suppl, pp 189-200.

[9] Chen WC (1995), ‘A Standard Procedure for Hot Cracking Test’, Gleeble Application Note, Dynamic Systems Inc, Poestenkill, NY, USA, (IIW Doc. II-C-042A/95).

[10] (2003), CEN ISO/TR 17641-3:2003, ‘Destructive tests on welds in metallic materials — Hot cracking tests for weldments - Arc welding processes — Part 3: Externally loaded tests’.

[11] Chen WC (1995), ‘Hot Tensile Ductility Test for HAZ Liquation Cracking Susceptibility Study’, Report #950426, Dynamic Systems Inc, Poestenkill, NY, USA, (IIW-Doc. II-C-042/95).

[12] Wilken K (1994), ‘Investigation to Compare Hot Cracking Tests - external loaded specimens’, BAM, Berlin, Germany, IIW Doc. IX-H-305/94.

[13] Mandziej ST (1997), ‘Physical Simulation of Welding’, in Welding and Joining Science and Technology, ASM International Europe, Brussels, pp 253-268.