A Hot-Cracking Mitigation Technique for Welding High-Strength Aluminum Alloy

The mechanical strain associated with hot cracking is significantly reduced with the introduction of a heat sink trailing a gas tungsten arc welding torch

BY Y. P. YANG, P. DONG, J. ZHANG AND X. TIAN

ABSTRACT. A hot-cracking mitigation results confirmed the effectiveness of the ture range BTR and to increase its ductil- technique for gas tungsten arc welding trailing heat sink technique. With a ity, one documented method is to intro- (GTAW) of high-strength aluminum alloy proper implementation of the trailing duce alloy elements, such as Ti, Zr, V and 2024 is presented. The proposed welding heat sink method, hot cracking can be B in the aluminum alloy electrode. In technique incorporates a trailing heat completely eliminated in welding alu- doing so, the grain structure of the weld sink (an intense cooling source) with re- minum alloy 2024 (AA 2024). metal can be refined, resulting in an im- spect to the welding torch. The develop- proved ductility and resistance to hot ment of the mitigation technique was Introduction cracking (Ref. 1). A variation of this ap- based on both detailed welding process proach is to control the solidification mi- simulation using advanced finite element Hot cracking has been a subject of in- crostructure of the weld metal by using techniques and systematic laboratory ex- tensive studies over the last few decades. special solidification techniques. For in- periments. Hot cracking occurs during the solidifi- stance, magnetic arc oscillation and The finite element methods were used cation process due to a combination of electromagnetic stirring have been used to investigate the detailed thermo metallurgical behavior on cooling and to refine weld microstructure and to mechanical behavior of the weld metal the surrounding thermomechanical con- change solidification orientation (Refs. that undergoes the brittle temperature ditions. In general, two basic approaches 2–6). The alternating columnar grains re- range (BTR) during welding. As ex- are usually taken: 1) improving the weld sulting from transverse arc oscillations at pected, a tensile deformation zone and heat-affected zone (HAZ) material a low frequency can be effective in re- within the material BTR region was iden- ductility and 2) improving the thermo- ducing solidification cracking. tified behind the weld pool under con- mechanical conditions during welding. However, it has been realized that for ventional GTA welding process condi- Historically, the majority of the research some aluminum alloys, such as 2024, the tions for the aluminum alloy studied. To work has been focused on the former, aforementioned techniques may not offer mitigate hot cracking, the tensile zone i.e., improving the weld and HAZ mate- satisfactory results, especially for elimi- behind the weld pool must be eliminated rial ductility. nating liquation cracking. Consequently, or reduced to a satisfactory level if the To reduce the weld material tempera- thermomechanical-based techniques weld metal hot ductility cannot be further have received an increasing attention improved. With detailed computational over the recent years. Zacharia (Ref. 7) modeling, it was found that by the intro- studied the relationship between the dy- duction of a trailing heat sink at some dis- namic stress distribution and the ob- tance behind the welding arc, the tensile served cracking behavior in a Sigmajig KEY WORDS strain rate with respect to temperature in test specimen, particularly near the trail- the zone encompassing the BTR region ing edge of the weld pool. Feng (Ref. 8) Weld Hot Cracking can be significantly reduced. A series of analyzed thermal and mechanical condi- Aluminum Alloy 2024 parametric studies were also conducted tions associated with weld metal solidifi- High-Strength Aluminum to derive optimal process parameters for cation cracking. As a better understand- Trailing Heat Sink the trailing heat sink. The experimental ing of the thermomechanical conditions Welding Process Model associated with a hot crack was estab- Finite Element Modeling lished, Karlsson (Ref. 9) proposed a local Y. P. YANG, P. DONG and J. ZHANG are with Gas Tungsten Arc Welding

heating approach based on the charac- RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT the Center for Welded Structures Research, GTAW teristics of welding-induced stress distri- Battelle, Columbus, Ohio. X. TIAN is with bution in a large plate. It was postulated Harbin Institute of Technology, Harbin, P.R. that the local heating of the base metal on China.

WELDING RESEARCH SUPPLEMENT | 9-s each side of the weld could be attributed to the fact the transverse and produce additional thermal longitudinal strain fields were altered by stresses to neutralize tensile the use of local heating. For instance, the stresses in the weld zone. The results obtained by Wu (Ref. 14) indicated local heating techniques were the tensile strain rate within the brittle also used by Sekiguchi and temperature range in the weld zone was Mayake (Refs. 10, 11) to over- reduced with local heating. However, the come end-cracking problems resulting large HAZ and coarser grain mi- in welding high-strength steel. crostructure as a result of local heating The end cracks were success- have been of concern over its applicabil- fully prevented by preheating ity in practice. It was also worth noting the panel to 120°C (248°F), local mechanical rolling could also be maintaining at a specified used to improve the thermomechanical inter-pass temperature of conditions in welding of some simple 200°C (392°F), and postweld specimens, as recently illustrated by Liu heating of 700°C (1292°F) with and Tian (Ref. 15). a local heating torch. Hernan- In the following, detailed finite ele- dez and North (Refs. 12, 13) ment results were presented to charac- used a pair of oxyacetylene terize the detailed evolution of instanta- heating torches to heat the base neous stress and strain behavior metal on either side to prevent associated with hot cracking. A novel Fig. 1— Thermomechanical conditions associated with solidification cracking of H30 welding technique with a trailing heat hot cracking (Ref. 15). aluminum alloy sheets. sink is then proposed to mitigate hot The mechanisms associated cracking during welding of high-strength with the local heating methods aluminum alloys. Parametric studies in preventing hot cracking can were conducted to investigate optimal

A B

Fig. 2 — Illustration of the effects of a trailing heat sink on isotherms. A — Conventional welding; B — welding with a trailing heat sink.

A B

RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT Fig. 3 — Welding conditions and specimen configuration. A — Conventional GTA welding; B — welding with a trailing heat sink.

10-s | JANUARY 2000 parameters for the trailing sink. Experi- mental studies were also carried out to validate the proposed welding technique on high-strength aluminum alloy 2024 panel specimens.

Analysis Procedures

Thermomechanical Conditions for Hot Cracking

The necessary condition for hot cracking is the presence of tensile strains in the region that undergoes the brittle temperature range. Note in this study, mechanical strains were used as a measure of the driving force for hot cracking instead of stresses since at the BTR region the transient stress level is usually low due to significantly reduced material yield strength at high temperature. As depicted in Fig. 1 (Ref. 15), if the tensile strain rate (with respect to temperature) exerting on the BTR region becomes smaller than “the critical Fig. 4 — Finite element model. strain rate for temperature drop (CST), as depicted by the tangent line (Curve B) to the ductility curve, or dε < CST dT (1) or in theory, hot cracking can be avoided, as depicted by line C. Decomposing dε/dT, one obtains the following: dεε∂∂/ t = dT ∂∂Tt/ (2) The first term ∂ε/∂t in Equation 2 becomes the standard expression of strain rate de- termined by the thermomechanical re- sponse during welding. The second term ∂T/∂t represents cooling rate typically controlled by the heat flow characteristics of the workpiece under considera- tion. For a given material, both the strain rate and the cooling rate can be altered by either modifying welding parameters or introducing local heating/cooling Fig. 5 — Temperature dependence of thermal physical properties for aluminum alloy 2024. mechanisms, or applying mechanical means. For instance, by introducing a trailing heat sink (Fig. 2), an auxiliary compression zone can be generated to temperature) should be obvious. As- moment in time becomes, within the trailing BTR region between suming a simple additive relationship be- RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT the heating source and the heat sink. The tween welding-induced strain (εw) and ε = εw – εc (3) effectiveness of the additional compres- the strain (εc) generated by the heat sink sive strains on the strain rate (with respect source, the total mechanical strain at any and Equation 1 takes the form of

WELDING RESEARCH SUPPLEMENT | 11-s AB

Fig. 6 — Temperature dependence of mechanical properties for AA 2024. A — Elastic modulus E, thermal expansion α and Poisson’s ratio ν; B —

Yield stress σs and hardening modulus H.

line were clamped out-of-plane) were imposed at clamp tightly by a hydraulic positions (Fig. 4) to simulate the clamp- keyboard jig to ening conditions during welding. sure uniformity of pressure in order to Moving Arc and Trailing Heat Sink simulate severe re- straint conditions in A moving heat source model was used production situa- as an ABAQUS user-subroutine tions, under which (*DFLUX) in the form of an equivalent hot cracking was typ- body heat flux generation. The energy- ically observed with density distribution of the moving arc the conventional was assumed as a Gaussian distribution GTA process. (Ref. 19) expressed by The commercial 2 22 3ηEI ()xvty− ++ z finite element pack- q = 2 exp 2 Fig. 7 — Comparison between FE and experimental results — con- age ABAQUS (Ref. πa −a / 3 (5) ventional welding. 16) was used in this Where η is arc efficiency, ν welding investigation. As dis- speed, E arc voltage, I welding current, a cussed in Refs. 17 effective radius of the heat source, and t welding time. The corresponding weld- dε w dεc and 18, almost all commercial finite el- − < CST ing parameters considered in this study dT dT (4) ement codes are not suited for analyzing the detailed thermomechanical behav- are summarized in Table 1. Equation 5 As a result, hot cracking can be prevented was used to approximate an equivalent if the reduction of the strain rate with re- ior associated with welding. Additional capabilities must be added in the form 3-D power density distribution of the spect to temperature is sufficient within weld region within an effective heat the BTR region. Note the actual reduction of user-developed subroutines. In the present study, a unified constitutive law source radius, a. The final values of the of the strain rate (with respect to temper- effective heat source radius, a and arc ef- ature) is dictated by the complex interac- (Ref. 27) was adopted in the form of user-material routine interfaced with ficiency η as shown in Table 1, were tions between the instantaneous strain based on a series of numerical experi- rate (∂ε/∂t) and cooling rate (∂T/∂t) during ABAQUS to simulate the detailed melting and subsequent “annealing” effects. A ments to achieve a good estimation of welding, as indicated in Equation 2. In the both temperature histories at measured following, advanced finite element pro- three-dimensional (3-D) shell element model as shown in Fig. 4 was adopted locations and weld zone profile, as dis- cedures were used to quantify and opti- cussed later. A similar approach has been mize the effects of the trailing heat sink from an early study (Ref. 17) where its application for modeling multipass used in a number of early studies (Refs. method on the dynamic thermomechani- 18, 31) for 3-D weld residual stress pre- cal interactions during welding. welds was presented. With the shell element model, the surface convection dictions in aluminum alloys. The trailing heat sink was modeled by Finite Element Model heat loss from the top and bottom sur- faces of the coupon sheet and the developing an ABAQUS user-subroutine As shown in Fig. 3, two types of weld- out-of-plane (in z-direction) deforma- (*FILM) to simulate the intense cooling effects due to the presence of a heat sink. ing conditions were analyzed. Full pene- tion mode were considered. The The heat sink was assumed to follow the tration bead-on-plate welds were made through-thickness temperature and de- welding arc with a distance D. The re- in the test specimens of aluminum alloy formation gradients were resolved by sulting energy density within the heat 2024 with a length of 100 mm, width of using five integration points along shell sink area was assumed to be uniform and 150 mm and thickness of 2.5 mm with thickness directions based on an early was expressed as

RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT both conventional GTA welding process study (Ref. 17). Symmetry conditions (Fig. 3A) and the proposed trailing heat were assumed with respect to the cen- ηcQVl c ql = 2 sink technique — Fig. 3B. Both sides of terline of the weld. Fixed displacement πb (6) the test specimens parallel to the weld boundary conditions (both in-plane and where ηc is cooling efficiency, Vc flow

12-s | JANUARY 2000 A B

Fig. 8 — Predicted temperature distributions on top surface. A — Conventional welding; B — welding with a trailing heat sink.

Fig. 9 — Effects of cooling distance, D, on temperature histories. Fig. 10 — Transient strain behaviors under various cooling conditions

rate of cooling media, Ql energy required ness, its relevant formulations are briefly vanished in the context of solid mechan- by vaporization of cooling media and b summarized below, particularly on the ics analysis procedures as the material effective radius of the heat sink area. The definitions of some of the parameters transforms from solid to liquid states. To heat sink parameters for using liquid ni- used in this study. restore such a “virgin” state , the associ- trogen are given in Table 1. The energy By assuming additive decomposition . A loss density q due to the presence of the of the strains, the rate form of the strains ε l ated annealing strain can be ex- trailing heat sink was considered as an can be written ~ pressed as equivalent convection loss effect with the Tot e pθ A ph ...... AA following expression: εεεεεε=++++ . .  ⋅  ~ ~ ~ ~ ~ ~ (8) εεθθε= −  ,, ~ ~  ~ (9) h (θ – θ ) = q (7) l l l . Tot. e. p. θ . A. ph . A εεεε,,,,,εε ε The equivalent convection coefficient, where ~ ~ ~ ~ ~ ~ are the where~ is a function of temperature, its total, elastic, plastic, thermal, “anneal- rate and strain. Neglecting solid-state hl, for the heat sink area can be calcu- lated for a given equivalent ambient tem- ing” and phase change strain rate tensors, phase change effects for the present dis- respectively. The elastic, plastic, and cussions, the stress strain relation may be perature θl. A series of numerical experiments showed the equivalent ambient thermal strains are defined in the classi- written as: cal sense. As required in welding process temperature within the heat sink area .  . Tot. p. θ . Ap  could be assumed at –10°C to achieve simulation, we postulate the existence of σεεεε= E: −−− ~  ~ ~ ~ ~  satisfactory temperature predictions. The an “annealing strain,” which is important ≈   (10) thermophysical properties for AA 2024 only at near-melting temperatures. The E used in the thermal analysis are listed in “annealing strain” eliminates the history where ≈ is the temperature dependent Table 2 and Fig. 5 (Refs. 23, 25, 26). of prior straining above a reference tem- elastic stiffness tensor, and we distinguish perature (such as at melting), θ . For in- . Ap . Ap A ε ε Weld Constitutive Model between plastic and elastic an- stance, once reaching melting tempera- nealing strains. By adopting Von Mises RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT A unified constitutive model for weld- ture, all accumulated elastic and plastic yield criterion within a framework of ⋅⋅ep ing process simulation was adopted from εε+ combined kinematic and isotropic hard- an early work (Ref. 27). For complete- strains ~~ and stresses should be ening rule, as required in modeling

WELDING RESEARCH SUPPLEMENT | 13-s A B

Fig. 11 — Effects of cooling distance, D, on the development of transverse strain at two selected positions. A — X = 29.5 mm, Y= 0.43 mm; B — X = 5 mm, Y = 0.43 mm (near weld start position).

Fig. 12 — Effects of cooling intensity (cooling liquid flow rate) on transverse strain. Fig. 13 — Experimental set up for GTA welding with trailing cooling.

cyclic efforts in multipass welds, the Results and Discussions hind the welding torch (Fig. 3B), a severe yield function can be formulated, as de- isotherm depression behind the weld scribed in detail (Ref. 27). Thermal Analysis pool can be seen in Fig. 8B. Such an For the annealing strains, we assume isotherm depression induced by the trail- . A Under the given set of welding ing sink was intended to change the tran- εθθ=<0, for A sient stress and strain behavior within the ~ process parameters measured (welding m current, voltage and travel speed), the BTR region in a favorable manner. The ef-  .  . A θ . A final selection of arc efficiency, η and fects of the distance, D between the ε =   εθθθ, for Am<< heat source radius, a were based on a se- welding torch and the trailing heat sink ~  θθm −  ~   (11) ries of numerical experiments until a are shown in Fig. 9, where the tempera- where θm is the material melting temper- good agreement with the measured tem- ture histories at a material point (x = 29.5 ature given in Table 2, θA is the anneal- perature history and fusion profile was mm and y = 0.43 mm) were obtained ing temperature and m can be assumed achieved. For instance, the predicted with D = 18, 23 and 28 mm, respectively. to fit various material annealing behav- temperature history at the center of the It is important to note the maximum tem- iors. The detailed numerical implemen- specimen with x = 50 mm and y = 0 is perature reached remained the same tation was discussed in Ref. 27. In the compared with experimental results in among the three cases, but their cooling current study, it is assumed that θA = Fig. 7 under conventional gas tungsten rates were altered drastically. As ex- 450°C and m = 1. Other material prop- arc welding conditions. It can be seen a pected, the shortest distance (D = 18 mm) erties used in the present analysis are good agreement was obtained between generated the highest cooling rate. shown in Fig. 6 (Refs. 24, 25). Young’s the finite element predictions and mea-

RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT modulus E, Poisson’s ratio ν and thermal surements. The predicted temperature Thermomechanical Analysis expansion coefficient α are given in Fig. distribution on the top surface of the 6A and yield stress σs and linear harden- specimen at t = 10.4 s is shown in Fig. 8A. With the temperature histories ob- ing modulus H in Fig. 6B. As a trailing heat sink was introduced be- tained above, sequentially-coupled ther-

14-s | JANUARY 2000 momechanical analyses were performed significantly reduced under conditions with and without a trail- as a trailing heat sink ing heat sink. Predicted transverse me- was introduced. chanical strain (as a measure of hot crack- To illustrate the ing driving force) histories at the same effects of the trailing material point as the one described in Fig. heat sink on the ther- 9 (x = 29.5 mm and y = 0.43 mm) are momechanical con- shown in Fig. 10. The solid line shows the ditions associated results obtained under conventional GTA with hot cracking, welding conditions. With the approach of the finite element re- the welding torch, the material point was sults are plotted in rapidly heating up and subjected to com- the same format as pression. After reaching its melting tem- the material’s hot- perature, a zero mechanical strain state ductility curve pre- was restored at about t = 4.8 s. Note the sented by Tamura restoration to the material’s virgin state (Ref. 28) in Fig. 11. after melting would not be possible with- As it is correctly Fig. 14 — Effects of cooling distance, D, on hot crack length — ex- out the use of the unified constitutive law pointed out by one of perimental results described earlier. As the welding torch the reviewers (Ref. was moving further away from this posi- 29) for this manu- tion, the material started solidifying, with script, the ductility curve (solid line) was finite element results in the same format a sudden increase in the transverse me- obtained by measuring averaged strains (strain vs. temperature) as the ductility chanical strain being predicted. It is this (thermal and mechanical) over a gauge data by Tamura (Ref. 28). tensile strain increase within the BTR re- length of 20-mm distance (Ref. 28). The In Fig. 11A, the computed transverse gion that acts as the driving force for hot transverse strain results from the current strains as a function of temperature are cracking. The effects of the distance, D, study are local mechanical strain values shown under conditions with and with- between the welding torch and trailing at the selected positions. With this in out trailing heat sink conditions for the heat sink are also shown in Fig. 10. It can mind, however, some general observa- same material point (x = 29.5 mm, y = be seen the increase in tensile strain was tions can still be made by presenting the 0.43 mm) as that in Figs. 9 and 10. The

Fig. 15 — Photographs of weld samples. A — Conventional welding; B — welding with a trailing heat sink. RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT

Fig. 16 — Weld crystallization orientation. A — Conventional welding; B — welding with a trailing heat sink.

WELDING RESEARCH SUPPLEMENT | 15-s As indicated in Fig. 11, the distance, D, between the welding torch and the trailing heat sink was identified as an important parameter in reducing the propensity for hot cracking. With the optimal process conditions being approxi- mately established from the detailed finite element simulations discussed earlier, welding trials were performed for various values of D. The results are summarized in Fig. 14. It can be seen the smaller the cooling distance D, the shorter the hot cracking length. As the cooling distance became shorter than 18.5 mm, hot cracking was completely eliminated. Fig. 15 shows the photographs of two typical welding trial specimens with and without using the trailing heat sink method, respectively. The hot crack length under conventional GTA welding conditions was measured at 35 mm, while there was no indications of any cracks for specimens welded with the trailing heat sink (D = 18 mm). Since the heat flow characteristics were significantly altered with the presence of a trailing heat sink, the corresponding weld microstructure evolution process was expected to be changed as well, such as solidification orientation, grain size, etc. As shown in Fig. 16A, grains grow from the weld interface and interlock at the weld centerline under conventional GTA welding conditions that can facilitate the forming of suscep- tible sites for hot cracking initiation (Ref. 30). With a trailing heat sink, as shown in Fig. 16B, grains grow from the area Fig. 17 — Microstructure of weld samples. A — Conventional welding; B —welding with a trail- subjected to intensive cooling. Such a ing heat sink; C — weld center. solidification pattern may have additional beneficial effects on creating favorable flow conditions for low-melting transverse strain level under conven- rate) of the trailing heat sink on the trans- species and impurities back into the tional GTA welding conditions (solid verse strain evolution. In this case, the molten weld pool. A detailed investiga- line) without trailing heat sink was the distance, D, was kept constant at 28 mm. tion is underway to examine these met- highest throughout the temperature The results indicate that the transverse allurgical effects and will be reported range among the four cases presented. strain further decreases with an increas- later. In addition, the overall microstruc- When a trailing heat sink was intro- ing cooling intensity (cooling flow rate). ture becomes finer than the one without duced, the transverse strain was notice- This is desirable since the intense cooling the trailing heat sink — Fig. 17. These are ably reduced, depending on the dis- may be used in exchange for a longer dis- additional metallurgical benefits result- tance, D, between the welding torch and tance between the welding torch and the ing from the trailing heat sink. the heat sink. It is known that the detailed heat sink so the technique can be more stress/strain evolution can be a strong readily implemented in practice. Conclusions function of location and that hot cracking often initiates near the weld start position Experimental Verification The detailed mechanisms associated (Refs. 7, 8). The computed transverse with weld hot cracking in aluminum strain results at a position (x = 5 mm, y = To implement the proposed welding alloy 2024 have been discussed in light 0.43 mm) near the weld start position are technique with a trailing heat sink, an ex- of finite element results. A hot cracking summarized in Fig. 11B. The results for perimental set up was developed, as mitigation technique was developed by the four cases share a similar trend to the shown in Fig. 13. Liquid nitrogen was se- introducing a trailing heat sink during one shown in Fig. 11A. However, the lected as a cooling media behind the GTA welding. Advanced finite element computed transverse strain levels near welding torch. This was accomplished by techniques were used to establish the de- the weld start position were higher than using a spray nozzle connected with a tailed understanding of the thermome- RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT those in Fig. 11A for the position well in- pressurized liquid nitrogen supply sys- chanical process during welding. The rel- side the specimen, as one would expect. tem. The distance between the welding ative importance of various cooling Figure 12 shows the effects of the torch and the cooling nozzle can be ad- parameters were then derived from the fi- cooling intensity (cooling liquid flow justed for optimal performance. nite element results. Welding trials were

16-s | JANUARY 2000 also conducted to verify the effectiveness T. 1984. Effect of addition element on weld so- Reactor Technology. of the proposed technique. The major lidification cracking susceptibility of Al-Zn- 18. Dong, P., Hong, J. K., and Rogers, P. findings are summarized as follows: Mg alloy (Report III). Transactions of JWRI 1998. Analysis of residual stresses in Al-Li 13(1): 57–66. alloy repair welds and mitigation techniques. 1) With the introduction of a trailing 6. David, S. A., and Vitek, J. M. 1989. Cor- Welding Journal 77(11): 439-s–445-s. heat sink, the mechanical strain associ- relation between solidification parameters 19. Kou, S. 1987. Welding Metallurgy. pp. ated with hot cracking can be signifi- and weld microstructures. International Mate- 393–395. John Wiley & Sons. cantly reduced particularly within the re- rials Review 34 (5): 213–245. 20. Rose-Innes, A. G. 1965. Low Temper- gion undergoing BTR between the weld 7. Zacharia, T. 1994. Dynamic stresses in ature Technique, p. 140. D. Van Nostrand pool and the heat sink. weld metal hot cracking. Welding Journal Company, Inc. 2) With a proper combination of the 73(7): 164-s–172-s. 21. Necati, O. M. 1993. Heat Conduction. 8. Feng, Z. 1994. A computational analy- p. 14. John Wiley & Sons, Inc. distance D (between the welding torch sis of thermal and mechanical condition for 22. Sparrow, E. M. 1978. Radiation Heat and trailing heat sink) and cooling inten- weld metal solidification cracking. Welding in Transfer, p. 45. Mcgraw Hill Book Company. sity, hot crack can then be eliminated. the World 33(5): 340–347. 23. Pehlke, R. D., Jeyarajan, A., and Wada, 3) Welding trials on aluminum alloy 9. Karlsson, L. 1976. Stress Fields Induced H. 1982. Summary of thermal properties for 2024 confirmed the effectiveness of the by Moving Heat Source in Butt-Welding of casting alloys and mold materials, pp. 82–93. trailing heat sink techniques. Plates. Ph.D. dissertation. Chalmers University National Technical Information Service. 4) Additional benefits of microstruc- of Technology, Goteborg, Sweden. 24. ASM Handbook Committee. 1979. 10. Sekiguchi, H., and Miyake, H. 1975. Metals Handbook Ninth Edition, Vol. 2, Prop- ture refinement and more favorable grain Prevention of welding cracks through a local erties and selection: nonferrous alloys and orientation were also observed from heating process. Transactions of the Japan pure metals, p. 68, 72–74. ASM International, weld samples produced by using the Welding Society 6(1): 59–64. Materials Park, Ohio. trailing heat sink techniques. 11. Sekiguchi, H., and Miyake, H. 1975. 25. Davis, J. R. 1994. ASM Specialty Hand- Postheating of the welding part of steel book, pp. 653–654. The Materials Information through a local heating process. Transactions Society. Acknowledgments of the Japan Welding Society 6(1): 53–58. 26. Victor, J. J. 1990. A Compendium of the 12. Hernandez, L. E., and North, T. H. Properties of Materials at Low Temperature The authors would like to express sin- 1984. The influence of external local heating (Phase I), pp. 2.132, 3.132, 4.132. Wright Air cere thanks to Profs. H. Zhu, M. Li and G. in preventing cracking during welding of alu- Development Division. Zhong of Harbin Institute of Technology minum alloy sheet. Welding Journal 63(3): 27. Brust, F. W., Dong, P., Zhang, J. 1997. for their valuable suggestions and discus- 84-s–90-s. A constitutive model for welding process sim- 13. Hernandez, I. G. 1977. Nature and ulation using finite element methods. Ad- sions during the experimental phase of Prevention of Solidification Cracking in vances in Computational Engineering Sci- this investigation. Welds. Ph.D. dissertation. Strathclyde Univer- ence, ed. by S. N. Atluri and G. Yagawa, pp. sity, Glasgow, Scotland. 51–56. Tech Science Press. References 14. Wu, E. 1990. A Study of Solidification 28. Tamura, H. 1977. Cracking study of Cracking of Cr-Ni Austenitic Stainless Steel. aluminum alloys by the variable tensile strain Ph.D. dissertation. Tsing Hua University, Bei- hot cracking test. Transaction of the Japan 1. Dvornak, M. J., Frost, R. H., and Olson, jing, China. Welding Society 8(2): 63–69. D. L. 1991. Influence of solidification kinetics 15. Liu, W., Tian, X., and Zhang, X. 1996. 29. Principal Reviewer’s Comments, Nov. on aluminum weld grain refinement. Welding Preventing weld hot cracking by synchronous 1998. Journal 70(10): 271-s–276-s. rolling during welding. Welding Journal 75(9): 30. ASM International Handbook Com- 2. Tseng, C., and Savage, W. F. 1971. The 297-s–304-s. mittee. 1997. ASM Handbook, Vol. 6, Weld- effect of arc oscillation. Welding Journal 16. ABAQUS/Standard User’s Manual. ing, brazing, and soldering, p. 88. ASM Inter- 50(11): 711–786. 1995. Hibbitt, Karlsson & Sorensen, Inc., Ver- national, Materials Park, Ohio. 3. Kou, S., and Le, Y. 1985. Alternating sion 5.5. 31. Dong, P., Hong, J. K., Zhang, J., Rogers, grain orientation and welding solidification 17. Zhang, J., Dong, P., Brust, F. W. 1997. P., Bynum, J., and Shah, S. 1998. Effects of re- cracking. Metallurgical Transactions 16A(10): Composite shell element model for residual pair weld residual stresses on wide-panel 1887–1896. stress analysis of multi-pass welds. Transaction specimens loaded in tension. ASME Journal of 4. Kou, S., and Le, Y. 1985. Improving weld of the 14th International Conference on Struc- Pressure Vessel Technology 120(2): 122–128. quality by low frequency arc oscillation. tural Mechanics in Reactor Technology, ed. Welding Journal 64(3): 51–55. Michel Livolant, (1): 335–344. The Interna- 5. Matsuda, F., Nakata, K., and Uchiyama, tional Association for Structural Mechanics in

Call for Papers

EPRI is sponsoring the Aging Workforce and Educational Infrastructure Conference and Exhibition, June 12–15, 2000, in Charlotte, N.C. The conference will address the methods and needs of maintaining capabilities of an aging workforce and declining education infrastructure. Some areas of interest are computer-based training, distance learning, proficiency testing, cross training, simulators and application of the Internet for education.

Individuals or organizations interested in making a presentation should submit a 1–2-page abstract to Ken Brittain, RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT conference manager, at [email protected]. Abstracts are due March 1, 2000.

WELDING RESEARCH SUPPLEMENT | 17-s