The Sigmajig test was used to quantify the hot-cracking response of Type 316 stainless steel welded with the GTA W, EBWand LBWprocesses BY G. M. GOODWI N ABSTRACT. The Sigmajig cracking re- sponse of a single heat of 0.010-in.-thick (0.25-mm) Type 316 stainless steel was determined as a function of welding pro- cess and parameter variations within a process. It was found that changes in welding parameters resulted in changes in cracking response, even when gross energy input was maintained constant. Higher travel speeds and elongated wel d pool shapes increased the cracking response, as did increased heat input. Cracking responses for all three welding processes, gas tungsten arc (CTA), elec- tron beam (EB), and pulsed solid-state laser beam (LB), were found to follow a common relationship wi th net heat input. Both EB and LB welds showed higher cracking resistance than GTA welds. Introduction The initial application of the Sigmajig test (Ref. 1) used a standard set of wel d- ing parameters wi th the CTA process to evaluate the dependence of cracking sensitivity on heat-to-heat compositional KEY WORDS Sigmajig Test 316 Stainless Steel Hot Cracking Response Heat Input Electron Beam Welding Pulsed Laser Welding GTA Welding Cracking Resistance Threshold Stress Argon Shielding G M. GOODWIN is a member of the Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, Tenn. Paper presented at the 68th Annual AWS Meeting, held March 22-27, 1987. in Chicago, III. effects in stainless steels. In this work, a specific, relatively crack-sensitive heat of Type 316 stainless steel was used to study the effect of heat input variations on cracking sensitivity using three wel d- ing processes: gas tungsten arc, electron beam, and pulsed solid-state laser. There are numerous examples in the literature documenting the effects of heat input on hot cracking. Typically, an increase in heat input increases cracking sensitivity, but often, there are also com- plicating factors to be considered, includ- ing changes in solidification mode (Refs. 2-4) or compositional effects (Refs. 5-7). Solidification growth morphology has also been shown t o be an important factor (Refs. 8, 9). Few studies have attempted to isolate the effect of individ- ual parameters (Ref. 10), and fewer yet have addressed process-to-process varia- tions (Ref. 11). In this study, we have intentionally selected a single crack-sensi- tive heat of material known to solidify as primary austenite for the processes and parameters investigated. The variations in response are thus attributable as nearly as possible to changes in stress at the trailing edge of the wel d pool , which are deter- mined by heat input, heat fl ow, thermal gradients, and weld pool geometry. Experimental Conditions A single heat of 0.010-in.-thick (0.25- mm) Type 316 stainless steel was used throughout the experiment. Its composi- tion is given in Table 1. Previous work (Ref. 1) established a threshold cracking stress 1 of 18 ksi (124 MPa), as compared wi th other commercial heats ranging from 15 to 53 ksi (103 to 365 MPa). It has a reasonably high P 4- S content (0.042 wt-%), a Cr-Ni ratio of 1.47 (Ref. 12), a predicted ferrite number (FN) of 0 (Ref. 13), and, as typified in Fig. 1, invariably solidifies as primary austenite wi th a small amount of residual eutectic ferrite. Gas tungsten arc welds were pro- duced under argon wi th a constant arc length of 0.035 in. (0.88 mm). Using DCEN and a 1/16-in. (1.6 mm) diameter thoriated tungsten electrode wi th a 45- deg included angle and a 0.010-in. trun- cation, arc current and travel speed were systematically varied above and below the standard conditions of 20 A and 35 ipm (14.8 mm/s) (Ref. 1). Five specimens were run at each combination of param- 1 Threshold cracking stress, a m in, is defined in Ref. 1 as the applied stress above which cracking first occurs. I'.aa y : -~~y^ Fig. 1 Typical microstructure of gas tungsten arc weld in Type 316 stainless steel. Heat 828013. Solidification mode is primary austenite with a small amount of eutectic ferrite. Murakami's etch 88-S | APRIL 1988 Table 1Composition of 0.25-mm-Thick Type 316 Stainless Steel (Heat 828013) Element C Mn P S Si Ni Cr Mo Cu N Cr / Ni ratio< a > Predicted FN< b > Primary solidification (c) Composition (wt-%) 0.018 1.70 0.032 0.010 0.34 12.16 17.04 1.98 0.05 0.047 1.47 0 Austenite (a) Calculated f r om Ref. 12. (b) Predicted f r om Ref. 13. (c) Based on metallographic observat i on; see Fig. 1. Fig. 2 Gas tungsten arc weld macrostruc- tures. A Standard conditions: 20 A, 35 ipm (14.8 mm/s), <r min = 18 ksi (124 MPa). B Low current: 16 A, 35 ipm, a min = 31ksi(214 MPa). C-High travel speed: 20 A, 49 ipm (20.7 mm/s), rT min = 31 ksi. D Round weld pool: 12 A, 21 ipm (8.9 mm/s), a mm = 37.5 ksi (259 MPa). EElongated weld pool: 28 A, 49 ipm, o-min = 7 ksi (48 MPa) WELDING RESEARCH SUPPLEMENT I 89-s 100 80 g 60 < g 40 20 HEAT 828013 207MPa (30ksi) TRAVEL SPEED = 14.8 mm/s (35in./min) AVERAGE OF FIVE TESTS 1 er 18 20 22 I (amps) 28 Fig. 3- a? o < o - Cracking response to changes in weld current TRAVEL SPEED (in./min) 21.0 24.5 28.0 31.5 35.0 38.5 42.0 45.5 80 60 4 0 20 0 8 A 1 1 1 1 1 1 \ HEAT 828013 \ T 207MPa(30ksi), I = 20amps \
AVERAGE OF FIVE TESTS 1o-
I I I I I 9 10.4 11.9 13.3 14.8 16.3 17.8 19.3 TRAVEL SPEED (mm/ s) 49.0
2C .7 Fig. 4 Cracking response to changes in travel speed I (amps)/TRAVEL SPEED (in./min) 12/21.0 14/24.5 16/28.0 18/31.5 2Q/35.0 22/38.5 24/42.0 26/45.5 28/49.0 80 g 60 < 40 20 1 1 1 1 i HEAT 828013 207MPa (30ksi) AVERAGE OF FIVE TESTS lo-
y
I I I I I 12/8.9 14/10.4 16/11.9 18/133 20/14.8 22/16.3 24/17.8 26/19.3 28/20.7 I ( amps) / TRAVEL SPEED (mm/s) Fig. 5 Cracking response at constant current-to-travel speed ratio eters to establish statistical variability. Electron beam welds were made in a 7.5-kW Hamilton Standard hard vacuum unit operating at 100 kV. Variations in travel speed were accompanied by com- pensating changes in beam current and focus. Pulsed solid-state laser beam welds were made under argon shielding using a 400-W Raytheon Model SS-500, wi th pulse rates, travel speeds, and power levels varied such that the weld pattern ranged from overlapping spots to a con- tinuous pool. Results and Discussion Gas Tungsten Arc Welds The standard conditions of 20 A and 35 ipm (14.8 mm/s) travel speed gave a Sigmajig response of about 50% cracking at 30 ksi (207 MPa). Figure 2 shows the various GTA wel d macrostructures. As plotted in Fig. 3, independently decreas- ing the current caused a reduction in cracking until at 16 A, which still results in a full-penetration wel d, cracking ceased. Increasing current increased cracking, with 100% cracking (specimen separa- tion) occurring at 26 A. The high standard deviation for high cracking percentages results from the fact that data between about 60 and 100% cracking do not occur, as noted in Ref. 1. Variations in travel speed at constant current caused comparable variations in cracking response, as detailed in Fig. 4. Decreasing travel speed increased crack- ing, wi th 100% cracking at 24.5 ipm (10.4 mm/s), and increasing travel speed decreased cracking, with zero cracking at 49 ipm (20.7 mm/s). Both of the above effects are consis- tent wi th results observed in actual prac- tice, and both are explained by noting that increasing heat input results in a wider zone of plastic yielding and causes a larger volume of metal t o expand and subsequently contract, increasing shrink- age stresses. Maintaining essentially constant heat input by varying both current and travel speed proportionally resulted in the data plotted in Fig. 5. At 12 A and 21 ipm (8.9 mm/s) and wi th a nearly circular wel d pool , no cracking was observed. Crack- ing increased as both current and travel speed increased, and 100% cracking occurred at 28 A and 49 ipm and wi th an elongated mucronate wel d pool shape. Although the change in wel d pool shape itself clearly affects the observed cracking response through geometric stress concentration and alteration of solidification growt h morphology at the trailing edge of the pool , t wo other possible contributions should be noted. First, examination of Fig. 2 shows that the weld wi dth increases roughly 30% going 90-s I APRIL 1988 from the circular to the elongated wel d pool shape. This is undoubtedly due t o the effect of the copper chill bar on which the specimens are wel ded. Although the arc heat input is maintained constant, slower travel speeds allow a higher fraction of this total heat input to be extracted through the copper, result- ing in a lower effective heat input to the specimen (Refs. 14, 15). As noted above, lower heat input reduces cracking. Secondly, as pointed out by Chihoski (Ref. 16), changes in wel d speed strongly affect the stress distribution behind the wel d pool. Wi thout an extensive stress analysis, which is beyond the scope of this work, and clearly beyond the capa- bilities of the author, it should be noted that the local transverse stress at which hot cracking initiates behind a wel d pool must be exceedingly l ow, limited by the composite strength of liquid plus solid near its melting point. Thus, when the normally predicted compressive stress in that location is offset by the preapplied test stress, the onset of cracking should be expected to be sensitive to shifts in the stress distribution, as detailed by Chi- hoski. The data shown in Figs. 3-5 are com- bined in Fig. 6 to give an overall view of the effect of current and/or travel speed on cracking response. The constant heat input (Fig. 5) data are a traverse from lower right t o upper left of the figure. The standard welding conditions of Ref. 1 are at the intersection of the t wo bold lines in the center of the plot. The choice of a test stress other than 30 ksi, or a different heat of material, woul d, of course, shift the results. The effect of variations in current and/ or travel speed on cracking response can also be reported by determination of the threshold stress value, <T m j n . Table 2 tabu- lates these values for the extreme GTA conditions and compares them wi th val- ues, to be discussed later, obtained with 20.7JA91 1001 [jn^.n.'.n0s 100 % CRACKING % CRACKING Fig. 6 Cracking response to changes in current and/or travel speed the EB and pulsed LB processes. For the GTA process, the standard conditions of 20 A and 35 ipm yield a threshold stress of 18 ksi (124 MPa). The elongated wel d pool formed at 28 A and 49 ipm drops the threshold stress to 7 ksi (48 MPa), whereas the lower heat input conditions, achieved by either high travel speed (49 ipm) or low current (16 A), raise the threshold stress to 31 ksi (214 MPa). The highest cracking resistance (37.5 ksi/259 MPa) is obtained with the round wel d pool (12 A and 21 ipm), at the same nominal heat input as the standard and elongated pool. Electron Beam Welds All three sets of welding parameters used wi th the EB process gave higher threshold stress values than any of the GTA conditions. Figure 7 shows the EB wel d macrostructures. Using the same travel speed, 35 ipm, as the standard GTA condition, with beam current and focus adjusted t o give approximately the same wel d wi dth, the measured thresh- old stress value was 57.5 ksi (396 MPa). Increases in travel speed to 70 and 105 ipm (29.6 and 44.5 mm/s), wi th compen- sating changes in beam current and focus, reduced the threshold stress val- ues to 55 and 50 ksi (379 and 345 MPa), respectively. As noted for the GTA welds, this effect is attributed to a reduc- tion in heat input (see Table 2), change in weld pool shape, and change in stress distribution at the trailing edge of the wel d pool . It is interesting to note that although travel speed was varied by a factor of 3, changes in the pool shape were such that substructure size remained essentially constant; this observation was confirmed by both optical metallography and scan- ning electron microscopy of crack sur- faces. Pulsed Laser Beam Welds As with the EB welds, all sets of weld- ing parameters used in the pulsed LB process gave higher threshold stress val- ues than any of the GTA conditions. The laser beam wel d macrostructures are shown in Fig. 8. Table 2Sigmajig Threshold Stress (o mi ) Values for Various Welding Conditions on O.OIO-in.-Thick (0.25-mm) Type 316 Stainless Steel, Heat 828013 Process CTAW EBW LBW Conditions Elongated pool, 28 A, 49 ipm (20.7 mm/s) Standard, 20 A, 35 ipm (14.8 mm/s) Low current, 16 A, 35 ipm (14.8 mm/s) High travel, 20 A, 49 ipm (20.7 mm/s) Round pool, 12 A, 21 ipm (8.9 mm/s) 35 ipm (14.8 mm/s), 1.0 mA 70 ipm (29.6 mm/s), 1.5 mA 105 ipm (44.5 mm/s), 2.0 mA 35 ipm (14.8 mm/s), 40 Hz, 64 W, 3 ms 35 ipm (14.8 mm/s), 100 Hz, 110 W, 3 ms 35 ipm (14.8 mm/s), 200 Hz, 107 W, 1 ms 52.5 ipm (22.2 mm/s), 200 Hz, 274 W, 1 ms 70 ipm (29.6 mm/s), 200 Hz, 314 W, 1 ms Heat Input, J/in. ()/mm) Cross 343 (13.5) 343 (13.5) 206 (8.1) 245 (9.6) 343 (13.5) 171 (6.7) 129(5.1) 114 (4.5) 110(4.3) 189 (7.4) 183 (7.2) 313(12.3) 269 (10.6) Net (a > 120 (4.7) 120 (4.7) 72 (2.8) 86 (3.4) 120 (4.7) 86 (3.4) 64 (2.5) 57 (2.2) 22 (0.9) 38(1.5) 37 (1.4) 63 (2.5) 54 (2.1) o- mJn , ksi (MPa) 7.0 (48) 18.0 (124) 31.0(214) 31.0(214) 37.5 (259) 57.5 (396) 55.0 (379) 50.0 (345) 80.0 (552) 70.0 (483) 65.0 (448) 60.0 (414) 55.0 (379) (a) Assumed efficiencies: CTAW, 35% (Ref. 15); EBW, 50% (Ref. 9); LBW, 20% (Ref. 17). WELDI NG RESEARCH SUPPLEMENT | 91-s ^^- ^^^I Lt ^^^J S^v ^ SH* ~ aa_ . I K - ^- - - ^ i > W - r^-a ' V JCa^^ffrta a ^ \ V . a ; . 4 ^ ,TalfcII*a*a-- . ^ a a ^ - ^ - ._- 4?S _ ': ^Ay*$ " . A -<- """ . Fig. 7 Electron beam weld macrostructures. A35 ipm (14.8 mm/s), 1.0 mA, cr min = 57.5 ksi (396 MPa); B-70 ipm (29.6 mm/s), 1.5 mA, cr min = 55 ksi (379 MPa); C- 105 ipm (44.5 mm/s), 2.0 mA, a mjn = 50 ksi (345 MPa). 400/um, 50X 92-s | APRIL 1988 A 7 Fig. 8 - Pulsed solid-state laser beam weld macrostructures. A-35 ipm (14.8 mm/s), 40 Hz, 64 W, 3 ms, <r mn = 80 ksi (552 MPa); B-35 ipm, 100 Hz, 110 W, 3 ms, (r m in = 70 ksi (483 MPa); C-35 ipm, 200 Hz, 107 W, 1 ms, <r min = 65 ksi (448 MPa); D-52.5 ipm (22.2 mm/s), 200 Hz, 274 W, 1 ms, <rmm = 60 ksi (414 MPa); E-70 ipm (29.6 mm/1.5), 200 Hz, 314 W, 1 ms, (7 min = 55 ksi (379 MPa) At a travel speed of 35 ipm (14.8 mm/s), a pulse rate of 40 Hz wi th 3 ms duration at a 64-W average power resulted in a series of spots wi th about 75% overlap. Since this particular heat of stainless steel did not crater crack under these conditions, the result was that each spot was essentially an independent event, and the threshold stress value of 80 ksi (552 MPa) approached the ultimate tensile strength of the base material (80.2 ksi/553 MPa in the annealed condition). Increasing the pulse rate to 100 Hz (with 3 ms duration) and 200 Hz (with 1 ms duration) required approximately twice the power level (110 W versus 64 W) to maintain the same wel d wi dth. Threshold stress dropped to 70 and 65 ksi (483 and 448 MPa), respectively. Increasing travel speed t o 52.5 and 70 ipm (22.2 mm/s and 29.6 mm/s) at 200 Hz, wi th concurrent increases in power level, caused further reductions in thresh- old stress to 60 and 55 ksi (414 and 379 MPa), respectively. The higher speed LB welds resembled the EB welds, both in physical appearance and in threshold stress values. Threshold Stress versus Heat Input The threshold cracking stress values for each of the welding process/parame- ter combinations investigated versus net heat input are plotted in Fig. 9. In deter- mining net heat input, the following effi- ciencies were assumed: GTA, 35% (Ref. 15); EB, 50% (Ref. 9); and LB, 20% (Ref. 17). It is debatable whether these effi- ciency values are precisely correct for the conditions of this experiment, but modest changes in assumed efficiency do not grossly affect the trend of the data. Wi th the above noted exceptions where changes in travel speed and wel d pool shape affect threshold stress indepen- dent of heat input, there is an overall relationship for all three processes between threshold stress and net heat input. Although a straight line can be fitted to the data with a correlation coef- WELDI NG RESEARCH SUPPLEMENT 193-s 700 600 500 400 b 300 200 100
"
I -if I I I (cold worked) UTS (annealed) A A A D I I I A D I I O I I I O GTA- 35% D EB - 5 0 % A LASER- 20%
O I I
o o o 100 75 50 25 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 NET HEAT INPUT (J/mm) 0 5.0 Fig. 9 Threshold cracking stress versus heat input. UTS = ultimate tensile strength fi ci ent R = 0. 87, i nt ui t i on wo u l d suggest t hat an exponent i al cur ve begi nni ng at t he ul t i mat e tensi l e st r engt h at zer o heat i nput and decayi ng t o zer o stress at ver y hi gh val ues of heat i nput mi ght be mor e appr opr i at e. Concl usi ons Based on t he above obser vat i ons, we can concl ude f or this part i cul ar heat of Type 316 stainless st eel , and i mpl y f or ot her materi al s, t hat : 1) For each of t he t hr ee wel di ng pr o- cesses, wel di ng paramet ers have a signif- i cant ef f ect on cr acki ng response. 2) In general , i ncreasi ng heat i nput i ncreases cracki ng response. 3) At least f or t he GTA process, mani pul at i on of wel di ng par amet er s at essentially const ant heat i nput alters t he cracki ng response due t o changes in wel d pool shape and stress di st ri but i on at t he trai l i ng edge of t he wel d pool . 4) Compar ed wi t h t he GTA process, bot h t he EB and pul sed LB processes pr ovi de i mpr oved cracki ng resi stance. These obser vat i ons are all in qual i t at i ve agr eement wi t h results obt ai ned in actual pract i ce. This not onl y conf i r ms t he abi l i ty of t he Si gmaj i g test t o r espond appr opr i - at el y t o changes in t he var i ous wel di ng par amet er s, but f ur t her suggests that t he quant i t at i ve nat ure of t he test wi l l pr ove useful in sel ecti ng par amet er s (or pr o- cesses) t o avoi d cracki ng, whi l e opt i mi z- i ng t r avel speed, heat i nput or ot her vari abl es of i nterest. Acknowledgments The aut hor woul d like t o t hank several i ndi vi dual s f or t hei r cont r i but i ons t o this wor k . ). D. Hudson was responsi bl e f or t he GTA wel di ng, R. W. Reed pr oduced t he laser beam wel ds, and B. G. Cross per f or med t he EB wel di ng. The pr ogr am was managed by t he Y-12 Plant Devel op- ment Di vi si on, and R. A. Huber , T. M. Must al eski , Jr., and M. H. Ri chey of t hat or gani zat i on deser ve speci al t hanks f or thei r pat i ence and counsel . The manu- script was t yped by Kat hy Gar dner and r evi ewed by J. F. Ki ng and T. M. Must al - eski . The fi nal r epor t was edi t ed by Geor ge Battle and pr epar ed by Al ma McDonal d. This wor k was sponsor ed by t he U.S. Depar t ment of Energy under cont r act DE- AC05- 84OR21400 wi t h Mar - t i n Mar i et t a Energy Systems, Inc. References 1. Goodwi n, G. M. 1987. Development of a new hot-cracking test - the Sigmajig. Weld- ing Journal 66(1):33-s t o 38-s. 2. Cieslak, M. J., Ritter, A. M., and Savage, W. F. 1982. Solidification cracking and analyti- cal electron microscopy of austenitic stainless steel wel d metals. Welding Journal 61(1):1-s t o 8-s. 3. Lippold, J. C. 1985. Centerline cracking in deep penetration electron beam welds in Type 304L stainless steel. Welding Journal 64(5):127-s t o 136-s. 4. Kujanpaa, V. P., Suutala, IM., Takalo, T., and Moisio, T. 1979. Correlation between solidification cracking and microstructure in austenitic and austenitic-ferritic stainless steel welds. Weld. Res. Int. 9(2):55-76. 5. Cieslak, M. J., and Savage, W. F. 1985. Hot-cracking studies of Alloy CN-7M. Welding lournal 64(5): 119-s t o 126-s. 6. Patterson, R. A., and Milewski, J. O. 1985. GTA wel d cracki ng-Al l oy 625 t o 304L. Welding Journal 64(8):227-s t o 231-s. 7. Golderer, W. , Schwab, R., and Stytzle, R. 1985. Avoidance of hot cracks when welding a material combination consisting of tool steel and a magnetically soft material wi th a solid state laser. Welding and Cutting 37(12):E211- E213. 8. Davi d, S. A., and Liu, C. T. 1982. High- power laser and arc welding of thori um- doped iridium alloys. Welding Journal 61(5):157-s t o 163-s. 9. Komizo, V., Pungshon, C. S., Gooch, T. G., and Blakeley, P. J. 1982. Effects of process parameters on centerline solidification in EB welds. Met. Constr. Br. Weld. J. 18(2):104R- 111R. 10. Koren, A., Roman, M., Weisshaus, I., and Kaufman, A. 1982. Improving the wel d- ability of Ni-base superalloy 713C. Welding M/ma/61(11):348-s t o 351-s. 11. Grigor 'Yants, A. G., Fishkis, M. M., and Andriyakhin, V. M. 1980. A comparison of the capacities of laser, electron-beam and arc- wel ded joints for resisting hot cracking. Avtom. Svarka, No. 10, pp. 11-14. 12. Suutala, N. 1982. Effect of manganese and nitrogen on the solidification mode in austenitic stainless steel welds. Met. Trans. 13A(12):2121-2130. 13. Delong, W. T., Ostrom, G., and Szuma- chowski, E. 1956. Measurement and calcula- tion of ferrite in stainless steel wel d metal. Welding Journal 35(11):526-s to 533-s. 14. Niles, R. W., and Jackson, C. E. 1975. Wel d thermal efficiency of the GTAW pro- cess. Welding Journal 54(1):25-s to 32-s. 15. Key, J. F., Smartt, H. B., Chan, J. W. , and Mcllwain, M. E. 1982. Process parameter effects on arc physics and heat f l ow in GTAW. Proceedings of international Conference on Welding Technology for Energy Applications. Gatlinburg, Tenn., May 16-19. 16. Chihoski, R. A. 1979. Expansion and stress around aluminum wel d puddles. Weld- ing Journal 58(9):263-s to 276-s. 17. Personal communication, Prof. T. DebRoy, Pennsylvania State University, State College, Pa. 94-s | APRIL 1988