WELDING RESEARCH

SUPPLEMENT TO THE WELDING JOURNAL, APRIL 1993 Sponsored by the American Welding Society and the Welding Research Council

An Evaluation of Heat-Affected Zone Liquation Cracking Susceptibility, Part I: Development of a Method for Quantification

A material-specific quantification of liquation cracking susceptibility was developed using the hot-ductility, spot- and longitudinal-Varestraint tests

BY W. LIN, J. C. LIPPOLD AND W. A. BAESLACK III

ABSTRACT. A new methodology has liquation cracking susceptibility; 4) elim­ and experimental procedure involved in been developed for quantifying heat-af­ inated the inconsistency among certain the development of the methodology. A fected zone (HAZ) liquation cracking weldability test results; 5) added impor­ subsequent paper will describe the met­ susceptibility. This methodology char­ tant insight regarding the mechanics of allurgical basis for HAZ liquation crack­ acterizes a thermal crack-susceptible re­ HAZ liquation cracking, and 6) con­ ing as it relates to this methodology. gion (CSR) in the HAZ. The thermal CSR firmed the criterion assumed in the de­ was theoretically derived based on the velopment of liquation cracking theo­ Introduction ductility of a material during welding as ries, which states that cracking results obtained from the Gleeble hot-ductility from the localized loss of grain bound­ HAZ liquation cracking is a type of test and the criteria assumed in the de­ ary ductility due to liquation. This paper high-temperature weld cracking that oc­ velopment of liquation cracking theo­ addresses the theoretical background curs in the HAZ adjacent to the fusion ries. This CSR was experimentally veri­ boundary (Ref.1). This type of cracking fied using longitudinal- and spot-Vare­ is often encountered during the welding straint tests performed on A-286 and of a variety of engineering alloys. It is Type 310 stainless steels. The thermal particularly prevalent in nickel- and alu­ CSR is material-specific and represents KEY WORDS minum-based alloys, and in fully a true quantification of liquation crack­ austenitic stainless steels. The metallur­ ing susceptibility. The development of Brittle Temperature gical basis for cracking involves the pres­ this methodology has 1) elucidated the Range ence and persistence of liquid films at physical relationship among weldability Crack-Susceptible grain boundaries and the inability of test results, liquation cracking theories Region these films to accommodate the ther­ and material properties; 2) provided a Heat-Affected Zone mally and/or mechanically induced more precise interpretation of hot-duc­ HAZ Liquation strain experienced during weld cooling. tility, spot- and longitudinal-Varestraint Cracking Although the precise mechanisms for li­ tests; 3) defined a method for determin­ Hot-Ductility Test quation cracking are not fully under­ ing a material-specific measure of HAZ Liquation Cracking stood, it is recognized that the simulta­ Stainless Steels neous presence of a crack-susceptible Varestraint Test microstructure and a critical level of re­ W. LIN and). C. LIPPOLD are with Edison Weldability straint is necessary to promote cracking. Welding Institute, Columbus, Ohio. W. A. Si nee the control of weld restraint is often BAESLACK III is with The Ohio State Univer­ Weldability Test sity, Columbus, Ohio. Technique difficult, reduction in liquation cracking susceptibility is normally achieved by Presented at the 73rd Annual AWS Meeting, adjusting the composition and mi­ held March 22-27, 1992, in Chicago, III. crostructure.

WELDING RESEARCH SUPPLEMENT I 135-s rial/process/restraint combination. They are ineffective in quantifying weldabil­ welding direction ity among different materials because of o o the difficulties in isolating the material A C factor from the test results. Simulative test techniques, such as the Varestraint test, attempt to simulate . some aspect of the thermomechanical response of the material to the welding process. These tests normally involve the application of an external aug­ mented stress or strain whose magnitude can be easily quantified. The thermally induced restraint of the specimen and/or fixturing is usually negligible compared with the relatively large amount of ex­ ssss ternally applied stress or strain. This ap­ proach allows the metallurgical and J block compositional factors associated with cracking to be isolated from the me­ chanical factors, and permits their ef­ Front View ££t/(2R+t) fects to be studied and quantified. As a result, simulative tests have been suc­ cessful in providing a comprehensive Fig. 1 — Schematic illustration of the longitudinal-Varestraint test. order ranking of families of alloys or heats of a given alloy. However, the test Since the 1950s, the effects of com­ mented restraint) test techniques (Refs. conditions, especially the combination position and microstructure on liqua­ 3,4). of thermal and mechanical history, are inherently different from actual welding tion cracking susceptibility have been Representative test techniques, such conditions. According to the technolog­ studied extensively using weldability as the circular patch test, seek to repro­ ical strength theory (Refs. 5, 6), the duc­ testing techniques. To date, over 150 duce the actual welding condition as tility of a material varies in a weld ther­ separate and distinct techniques have closely as possible in an effort to accu­ mal cycle. The temperature at which a been developed to quantify weld solid­ rately "represent" the situation of inter­ critical level of strain is applied on a ma­ ification and HAZ liquation cracking est. These tests depend on self-restraint terial during weldability testing may be susceptibility (Ref. 2). These tests vary induced by the specimen design and/or different from that under actual welding widely in their approach and utility, but fixturing. In most cases, these test tech­ conditions. This difference may result in can generally be classified as represen­ niques only provide a simple "crack or disparate cracking behavior in the sim­ tative (self-restraint) or simulative (aug­ no crack" solution for a specific mate- ulative test relative to actual practice.

The approach used to discern the ef­ HAZ cracks fects of test conditions on the cracking behavior relies on developing a physi­ cal link between weldability test results and the metallurgical response of the *""l material during testing. The absence of such a link in essentially all weldability CO OO test techniques, however, has created difficulties in standardizing the test pro­ spot weld cedures and resulted in poor repro­ Top View ducibility and correlation. Discrepan­ cies between test results and field expe­ specimen rience are not uncommon. GTAW torch Despite these drawbacks, weldabil­ ity testing is generally perceived as an efficient and economical method for •>/////;/////// sin? predicting susceptibility to liquation-re­ n PI lated cracking during welding. Among the weldability testing techniques, the longitudinal-Varestraint (mini-Vare­ straint), spot-Varestraint and hot-ductil­ ity tests are three of a few methods that ^\r- can be utilized in quantifying the HAZ liquation cracking susceptibility or de­ Front View veloping a fundamental understanding of the cracking phenomena. Despite e = t/(2R+t) their widespread use, there has been lit­ force tle effort to correlate the data generated by these tests or to rigorously apply these data to real-world situations. Fig. 2 — Schematic illustration of the spot-Varestraint test.

136-s I APRIL 1993 Longitudinal-Varestraint test Table 1 — Criteria for Interpreting Hot-Ductility Test Results for HAZ Cracking Assessments The original Varestraint test was de­ Authors, veloped by Savage, etal. (Refs. 7, 8), at Criterion (Refs.) Year Rensselaer Polytechnic Institute in the mid-60s, and soon became one of the Classification of hot ductility Nippes, ef al. (25, 26) 1955 most widely utilized weldability testing curves techniques for quantifying the suscepti­ Incorporation of the extent of Williams (27), Kreischer 1963 bility of a material to weld solidification ductility and strength (28), Weiss, ef al. (29) cracking. Since that time, three modi­ recovery Nil-ductility temperature range Williams (27), Dahl, et al. 1963 fied versions have been developed between the NST and DRT (30, 31), Donati, ef al. (11) based on the original Varestraint con­ The extent of the nil-ductility Duvall, ef al. (32-33) 1966 cept, namely, the mini-Varestraint (or region and the amount and subscale Varestraint) test, the spot-Vare­ rate of ductility recovery straint test (originally called the Tigama- on-cooling jig test) (Ref. 9), and the Transvarestraint The zero ductility range and Yeniscavich (34, 35) 1966 test (Ref. 10). The mini-Varestraint test mid-temperature ductility dip uses a smaller test sample, nominally range 163 X 25 X 6.4 mm (6.5 X 1 X 0.25 in.) The temperature range Arata, ef al. (36) 1977 as compared with the original Vare­ between the NDT and DRT straint test, nominally 300 X 50 X 12.7 The ratio of ductility recovery Lundin, ef al. (14) 1991 mm (1 2 X 2 X 0.5 in.). In order to avoid (RDR), ductility recovery rate (DRR) and nil-ductility confusion with the spot-Varestraint and temperature range (NDR) Transvarestraint tests, the mini-Vare­ straint (or subscale Varestraint) test is called the longitudinal-Varestraint test in this report. Although this test is uti­ Spot-Varestraint Test ing or the degree of cracking (quantified lized primarily for characterizing weld by MCL or TCL) over a range of strain solidification cracking, it has also been The spot-Varestraint test (Ref. 9) has levels have been generally accepted as used to determine HAZ liquation crack­ been widely applied for evaluating HAZ cracking indexes since the introduction ing susceptibility (Refs. 11-15). liquation cracking susceptibility. A of this test (Refs. 9, 16-23). schematic of the test apparatus is shown A schematic of the longitudinal-Vare­ in Fig. 2. During spot-Varestraint test­ Hot-Ductility Test straint test is shown in Fig. 1. A speci­ ing, a GTA spot weld is produced in the men is supported as a cantilever beam center section of a small specimen, nom­ The concept in the design of the hot- and a gas tungsten arc weld (GTAW) is inally 140 X 25 X 6.4 mm (5.5 X 1 X0.25 ductility test is different from the major­ produced along the center section of the in.). After a predetermined weld time, ity of weldability testing techniques. In­ specimen. When the arc approaches the the arc is extinguished and the speci­ stead of quantifying the cracking sus­ center of a radiused die block (marked men is forced to conform to the surface ceptibility by the degree of cracking, it A in Fig. 1), a pneumatically operated of a radiused die block. In this manner, characterizes the ductility of the mate­ ram is triggered forcing the specimen to HAZ liquation cracks can be generated rial at elevated temperatures and relates conform to the surface of the die block. on the surface of the specimen adjacent this ductility data to cracking suscepti­ Meanwhile, the arc travels onward and to the GTA spot weld. The applied aug­ bility. Basically, small tensile samples is subsequently interrupted in the run­ mented strain (e) of the top surface of are fractured rapidly at some specific off area C. Two auxiliary bending bars the specimen is approximated in the temperatures during either the on-heat­ are added in order to ensure that the same manner as for the longitudinal- ing or on-cooling portion of a duplicated specimen conforms to the contour of the Varestraint test. weld thermal cycle in a thermomechan­ die block. The applied augmented strain Cracking susceptibility is determined ical simulator called a Gleeble™. The (e) of the top surface of the specimen can by measuring the length of each crack transverse reduction-in-area of the frac­ be varied by adjusting the radius of the on the as-tested specimen surface. The tured sample is subsequently deter­ die block (R) following the equation, e mined providing a measure of ductility. = t / (2R + t), where t is the specimen threshold level of strain to cause crack­ thickness. In this manner, both weld so­ lidification cracks and HAZ liquation cracks can be produced. The HAZ li­ quation cracks are normally located ad­ jacent and perpendicular to the fusion thermal cycle hot ductility curve boundary. re A^von-cooling CD ^ ^or i-heating Cracking susceptibility is assessed by < measuring the length of each crack on s c c the as-tested specimen surface. The ro o threshold strain (Appendix B) to cause q> Q. 13 "•^on-cooling cracking and the degree of cracking at E -o a specific strain level have been gener­ d> CD 1- Jpon-heating CC ally utilized as cracking indexes (Refs. 11 -1 5). The degree of cracking can be quantified by the maximum crack length DRTN DTNSTTL (MCL), the total crack length (TCL) or Time Temperature the cracked HAZ length (CHL) (see Ap­ pendix B for definition of terms). Fig. 3 — Schematic illustration of the hot-ductility test.

WELDING RESEARCH SUPPLEMENT I 137-s 4. These results stimulated a research 2.0 program to fundamentally evaluate the NST-DRT NST MCL TL -NDT hot-ductility, spot- and longitudinal- Varestraint tests. 300 300 E 1.6 E Objectives

O. The principal objectives of this study § 1.2 were as follows: 2 200 _i 200 3 2 1) Study the physical relationship -3 |5 "ro among weldability test results, cracking CD ro Q. S3 theories and material properties. E "0.8 o_ CD E E 2) Develop a generic methodology rj CD 100 E 100 and define a material-specific parame­ x ter, based on the hot-ductility, longitu­ •§ 0.4 dinal- and spot-Varestraint test results, that quantifies HAZ liquation cracking susceptibility. 0.0 3) Investigate the correlations among 903 909 903 909 903 909 these three techniques. (A) (B) (C) 4) Re-evaluate the current method­ ologies for interpretation of the three Fig. 4 — Comparison of test results for Incoloy Alloys 903 and 909 from the spot-Varestraint tests in the light of the new methodol­ and hot-ductility tests; A — the temperature range between the NST and Df?7"v57 obtained ogy which is proposed. from the hot-ductility test; B — the maximum crack length obtained from the spot-Varestraint test; C — the temperature range between the T and NDT obtained from the hot-ductility test L Experimental Design (Ref. 37). Materials Both the on-heating and on-cooling duc­ these criteria, the extent of the nil-duc­ tility curve in the vicinity of the solidus tility region and the amount and rate of A-286 and Type 31 0 stainless steels temperature can be obtained, as shown ductility recovery (Refs. 32, 33), and the were investigated in this study. A-286 is in Fig. 3, and the nil-ductility tempera­ nil-ductility temperature range between a precipitation-hardened iron-base alloy ture (NDT), nil-strength temperature the NST and DRT (Refs. 11, 27, 30, 31) designed for applications requiring (NST) and ductility recovery tempera­ are the most widely accepted. moderately high strength to 700°C ture (DRT) determined (see Appendix B The hot-ductility and spot-Varestraint (1292°F) and oxidation resistance to for definition of terms). In addition to tests are normally considered comple­ about 815°C (1499°F). The excellent the hot ductility, the hot strength of the mentary in quantifying HAZ liquation room and elevated temperature material can also be measured. cracking susceptibility and yield quali­ strengths exhibited by this alloy result from the presence of Ni, Ti and Al, which In the past, experimental methodolo­ tatively consistent results. A previous in­ promote formation of the Y Ni (AI,Ti) gies and data interpretation for hot-duc­ vestigation (Ref. 37), however, showed 3 strengthening phase. This alloy is often tility testing have been the subject of that these two tests were inconsistent in used for jet engine and gas turbine ap­ considerable investigation. A review by predicting the HAZ liquation cracking plications due to its highly desirable Lin (Ref. 24) indicated that the methods susceptibility of Incoloy® Alloys 903 and combination of properties. for interpreting hot-ductility test results 909, when "conventional" techniques for HAZ cracking assessments vary sig­ were used to interpret test results. Fig­ Extensive studies in the past have nificantly from investigator to investiga­ ure 4 shows that the MCLs of Incoloy Al­ shown this alloy to be susceptible to tor. Table 1 lists the several criteria that loys 903 and 909 as obtained from the both weld solidification and HAZ liqua­ have been proposed to date. Among spot-Varestraint test (Fig. 4B) were con­ tion cracking (Refs. 38-41). Vagi, et al. tradictory to the on-cooling nil-ductility (Ref. 40), associated HAZ liquation

temperature range (NST-DRTNST) (see cracking with the formation of a Fe2Ti Table 2 — Chemical Compositions of A-286 appendix for symbols and abbreviations) Laves phase. They proposed that grain and Type 310 Stainless Steel Investigated in as obtained from the hot-ductility test — boundary liquation by the Fe-Fe Ti eu­ This Study (wt-%) 2 Fig. 4A. However, the MCL correlated tectic (1310°C; 2390°F) allows the very well with the on-heating nil-ductil­ Element Type 310 A-286 ity temperature range (TL-NDT) — Fig. c 0.054 0.083 Table 4 — Conditions for Mn 0.19 1.42 Longitudinal-Varestraint Testing of A-286 P 0.016 0.024 Stainless Steels S 0.009 0.004 Table 3 — Conditions for Hot-Ductility B 0.006 Not Analyzed Testing of A-286 and Type 310 Stainless Current:190 A Si 0.25 0.46 Steels Voltage:12 V, DCEN \i 24.22 18.72 Travel speed: 2.54 mm/s Cr 14.58 24.75 Heating time:12.15 s Augmented strain: 0.5-7% Mo 1.32 0.27 Holding time at peak temperature:0.03 s Shielding gas: Argon, 30 CFH Cu 0.10 0.32 Cooling rate: 50:C/s (90 = F/s) Electrode: W-2%ThO.>, 2.4 mm (%2 in.) Al 0.34 0.062 Holding time at test temperature:0.03 s diameter., 60-deg included angle Ti 2.29 0.014 Stroke rate: 5 cm/sec (2 in/s) Air Pressure: 0.55 MPa (80 psi) V 0.22 0.054 Sample freespan:25 mm (1 in.) Strain rate: approximately 13%/s Fe bal. bal. Atmosphere: argon Weld bead width: 7.5 mm

138-s I APRIL 1993 grains to be separated by thermally in­ duced strain. Blum, ef al. (Ref. 38), agreed with Vagi, ef al., but they also proposed that grain boundary precipita­ tion of a continuous film of TiC and in­ tragranular precipitation of y" inhibits deformation during cooling, which causes base metal cracks to propagate and relieve the weld stresses. Later works by Brooks (Ref. 41) indicated that the loss of hot ductility above 1150°C (2102°F) was attributed to the presence of boron. He suggested that the consti­ tutional liquation of borides caused grain boundary liquation and resulted in cracking. Type 310 is a fully austenitic stain­ less steel developed for applications re­ quiring high corrosion resistance. Due to the fully austenitic nature of the reso- lidified weld metal, this alloy suffers from weld solidification cracking as well as HAZ liquation cracking (Refs. 20, 42-46). Comparing the HAZ liquation cracking susceptibility of this alloy with other austenitic stainless steels, Kujan­ paa, etal. (Ref. 20), and Morishige, ef Fig. 5 — Experimental setup for the spot-Varestraint test. al. (Ref. 42), indicated that Type 310 is more susceptible than Types 316, 347, 321, 304 and 309. A fundamental inves­ tigation of the HAZ liquation cracking mechanism of this alloy by Tamura, ef load sufficient to overcome the frictional Longitudinal-Varestraint Testing al. (Ref. 46), suggested that Cr and Ni in force of the fixture, approximately 10 kg the solute-rich zones in the matrix were (22 Ib). On-heating tests were conducted Longitudinal-Varestraint tests were swept up and assimilated into the mi­ by heating samples to the peak temper­ performed only on A-286 stainless steel grating grain boundaries during weld ature in 12.15 s and pulling them to fail­ at strains ranging from 0.5 to 7%. Three thermal cycling. A subsequent eutectic ure at a rate of 5 cm/s (2 in./s). The on- samples were tested at each strain level reaction involving Cr and Ni at the grain heating time to peak temperature was on a subscale Varestraint test unit using boundaries caused a depression in the based on the NST test. On-cooling tests specimens with dimensions of 1 63 X 25 effective solidus. were performed after heating to the NST X 6.35 mm (6.5 X 1 X 0.25 in). Two aux­ The chemical compositions of A-286 in 12.15 s and cooling to the desired iliary bending bars were located on each and Type 310 stainless steels investi­ temperature at 50°C/s (90°F/s). This side of the sample to ensure that the gated in this study are listed in Table 2. cooling rate was the maximum achiev­ specimen fully conformed to the surface Both materials were in the form of 6.5- able with the materials and specimen of the die block during testing. The con­ mm (0.25-in.) thick plate. A-286 exhib­ freespan utilized without externally as­ ditions for longitudinal-Varestraint test­ ited a recrystallized austenitic structure sisted cooling. On-cooling test samples ing are listed in Table 4. A rapid strain with a grain size of ASTM No. 9 and con- were also pulled to failure at 5 cm/s. This rate ( approximately 13%/s) was selected tained a bimodal distribution of titanium rapid stroke rate was selected to mini­ to minimize the change in specimen carbides and/or carbonitrides. Type 310 mize the change in specimen tempera­ temperature in the HAZ during bending. was fully austenitic with a grain size of ture during fracture. Sample ductility, in A 7.5-mm (0.3-in.) weld bead width was ASTM No. 6. Note that the nickel con­ terms of reduction in area, was subse­ obtained with these welding parameters. quently measured using a vernier tent of Type 31 0 is slightly lower than A binocular microscope with 40X caliper. The conditions for hot ductility the specification limits of 1 9 to 22%. magnification was utilized to identify testing are summarized in Table 3. Both alloys exhibited relatively low lev­ the locations of each crack tip relative els of impurities, particularly with re­ spect to sulfur and phosphorus contents.

Hot-Ductility Testing

Hot-ductility tests were performed under an argon atmosphere using a Gleeble™ 1000 system. Standard speci­ mens 6.35 mm in diameter and 1 00 mm long (0.25 X 4 in.) were machined from specimen the plate. A specimen freespan (jaw spacing) of 25 mm (1 in.) was used thermocouple wires throughout the investigation. The NST 1/16" Hole was determined by heating samples at a rate of 111 °C/s (200°F/s) under a static Fig. 6 — Schematic illustration of the setup for temperature measurement.

WELDING RESEARCH SUPPLEMENT I 139-s Table 5 — Conditions for Spot-Varestraint Testing of A-286 and Type 310 Stainless Steels Table 6 — Hot-Ductility Test Results of A-286 Type 310 A-286 and Type 310 Stainless Steels Weld time 35 s 35 s Current 96 A 110 A A-286 Type 310 Voltage 16 V 12 V Tl(a) 1422°C 1386°C Weld diameter 11.5 mm 12 mm NDT 1200°C 1325°C Shielding gas Argon, 20 ft3/h Argon, 20 ft3/h NST 1350°C 1350°C Air pressure 0.55 MPa (80 lb/in.2) 0.55 MPa (80 lb/in.2) DRTNsT 1050°C 1325°C Strain rate approximately 13%/s approximately 13%/s TL-NDT 222°C 61°C Augmented strain 0.5-5% 1-7% 300 °C 25°C Cooling time 0-4.5 s 0-0.45 s NST-DRTNST Electrode diameter, 60-deg included angle W-2%Th02, 2.4 mm ( & in. (a) TL was determined as the peak temperature experienced at the fusion boundary during spot-Varestraint testing.

to the interception of the instantaneous 31 0. A rapid strain rate (approximately riod between the instant at which the solid/liquid interface and the fusion 13%/s) was also selected to minimize welding current drops to essentially zero boundary. These crack tip locations and the change in specimen temperature and the time at which the die block con­ the fusion boundary circumscribed a during bending. tacts the specimen. A special setup was cracked HAZ region. The maximum The spot-Varestraint tests were con­ devised to precisely monitor the actual width of the cracked HAZ was designated ducted in two stages. The first set of tests, cooling time. as the maximum crack length (MCL), called the on-heating spot-Varestraint It is also important to point out that which is the distance between the crack tests, was aimed at determining the for Type 310, the weld pool became ir­ tip of the longest crack and the fusion cracking behavior of the material on regular for weld times longer than about boundary projecting in a direction per­ weld heating. This was achieved by per­ 20 s. In order to produce a uniform cir­ pendicular to the fusion boundary. forming the tests with no cooling time cular weld pool, an electromagnetic coil (see Appendix B for definition). Thus, was custom-built and attached to the Spot-Varestraint Testing the HAZ was on-heating when it was weld torch as shown in Fig. 5. The size straining. The on-heating spot-Vare­ of the coil is 25 mm in height with 35- Spot-Varestraint tests were performed straint tests were performed from 0.5 to mm ID and 50-mm OD (1 in. H X 1.375 on a Tigamajig-type test unit (Ref. 9) 7% augmented strain with three sam­ in. ID X 2 in. OD). There were 280 turns using specimens with dimensions of 140 ples for each strain level. The second set of 1 9-gauge wire in this coil. By ener­ X 25 X 6.35 mm (5.5 X 1 X 0.25 in.). of spot-Varestraint experiments, called gizing this coil with 4.0 A of alternating Slots (rather than holes) at both ends of the on-cooling spot-Varestraint tests, current, the weld pool could be forced the sample allowed samples to be accu­ was designed to investigate the crack­ to rotate, resulting in a symmetrical, cir­ rately located and fixtured, and also ing behavior of the material on weld cular pool shape. minimized any axial tensile force dur­ cooling. This was accomplished by per­ Quantitative cracking data were ob­ ing specimen bending. The conditions forming the tests above the saturated tained from as-tested samples by mea­ for spot-Varestraint testing are listed in strain level (see Appendix B for defini­ suring the length of the longest crack be­ Table 5. The criterion for determining tion) with variable cooling times. The tween the fusion boundary and crack tip the welding parameters was to obtain a cooling time was varied by adjusting the projecting in a direction perpendicular cooling rate that was comparable with time period between arc extinction and to the fusion boundary using a binocu­ that used during on-cooling hot-ductil­ sample bending. Based on the cracking lar microscope under 40X magnifica­ ity testing. For the parameters selected, behaviors of these two alloys, cooling tion. Thus, the maximum crack length the average cooling rate from 1 350° to times ranging from 0 to 4.5 s were uti­ (MCL) reported here represents the dis­ 1200°C was about 84°C/s (1 51 °F/s) for lized for A-286 and from 0 to 0.45 s for tance between the isotherms at the crack A-286 and 11 2°C/s (202°F/s) for Type Type 310. tip and the fusion boundary, rather than 310. The average cooling rate over a Because cooling time is a critical pa­ the actual length of the observed crack. wider temperature range, from 1 350° to rameter in this test, it was precisely cal­ 1 000°C, was about 60°C/s (1 08°F/s) for ibrated and monitored throughout the Thermal Cycle Measurement A-286 and 70°C/s (126°F/s) for Type test. The actual cooling time is the pe- The thermal cycles experienced dur­ ing both the spot- and longitudinal-Vare­ straint testing were measured using fine On-Heating On-Cooling wire S-type (Pt - Pt 1 0%Rh, 0.2 mm in CSR CSR diameter) and K-type (Chromel and E' Alumel, 0.25 mm in diameter) thermo­ Jffi^WjSl;^ couples. K-type thermocouples were uti­ lized only to measure the thermal cycles •NDT NST DRTNS T with a peak temperature below 1 350°C. \ \ I These thermocouples were percussion / welded at the bottom of 1.6 mm (% in.) holes, which were drilled from the bot­ tom of specimens to within about 0.3 mm from the surface on which the weld was placed, as shown in Fig. 6. The tem­ perature was recorded using a Labtech Notebook software package (Ref. 47) at Fig. 7 — Theoretical hypothesis of the thermal crack-susceptible region.

140-s I APRIL 1993 a sampling rate of 20 Hz on an IBM 386 compatible personal computer. Under 100 these conditions, the thermal cycles at • a A-9flR different locations from the fusion boundary into the HAZ were obtained. 80 The peak temperatures and cooling rates \ On-Heating were determined and temperature gra­ D \ dients approximated. •£ 3 60 \ < \ Metallurgical Characterization c t In Representative longitudinal- and § 40 T3 1 spot-Varestraint test specimens were Z3 I metallographically prepared by grind­ 1 On-Cooling nl ing about 0.2 mm (0.008 in.) of mate­ rr 20 •^^from the NST 1 rial from the top surface and then pol­ *\ I NST ishing through 0.05-micron alumina. Hot-ductility specimens were ground to 0 \ L ! the center section of the sample and then 900 1000 1100 1200 1300 1400 polished through 0.05-micron alumina. Temperature (C) Both A-286 and Type 310 stainless steels were etched with a mixed acid solution (A) A-286 comprised of equal parts of concentrated nitric, hydrochloric and acetic acids. Mi­ crostructures were characterized using an optical microscope at magnifications 100 upto 1000X. * - Tvp"?1ft • o.^--"-— " —s. TO D On-Cooling X n \ Development of the 80 . from the NST \ \ On-Heating Crack-Susceptible Region 5 \ \ 3TO^ \ 1 Theoretical Hypothesis of the Thermal £ 60 . \ 1 Crack-Susceptible Region < \ 1 c \ A Metallurgically, HAZ liquation c Xr cracking is associated with the occur­ 1 40 - o \ 1 D \ I rence of grain boundary liquation. In the 73 \ i CD \p past, most efforts in studying HAZ liqua­ cr o\r tion cracking mechanisms have been di­ 20 - u rected toward characterizing the evolu­ 1 NST tion and distribution of liquid in the HAZ. However, the mere presence of 0 L 1 liquid films at grain boundaries is not 1200 1250 1300 1350 1400 sufficient to induce a liquation crack. In Temperature (C) order to cause cracking, it is essential that the crack-susceptible microstruc­ (B) Type 310 ture be subjected to a sufficient tensile strain (or stress). During welding, the Fig. 8 — Hot-ductility behavior of A-286 and Type 310 stainless steels. tensile strain (or stress) does not gener­ ally develop until the weld begins to cool (Ref. 48). As a result, liquation nological strength theory (Refs. 5, 6) are cracking occurs during the solidification the best known. Although these theories Table 7 — Pertinent Longitudinal-Varestraint Test Results of A-286 of the liquid films. The cooling cycle of differ in their approach to cracking the liquid films in the HAZ is similar to mechanistics, there is general agreement the final stages of weld solidification, al­ that cracking occurs in a discrete tem­ A-286 though the origin of the liquid films and perature envelope called the brittle tem­ MCL (at saturated 0.80 mm the microstructural boundaries may be perature range (BTR) (see Appendix B strain = 3%) different from those present in the weld for definition). Metallurgically, the BTR CHLOC.NST 1 1.5 mm metal. Consequently, to a first approxi­ describes a range between the tempera­ CHLoc.TL 14.5 mm mation, the criteria that govern weld so­ ture where liquid is confined within the CT 286°C/mm lidification cracking can be adopted to solidification structure to that where the CRT at NST from 66°C/s explain the solidification of liquid and boundary liquid is partially or com­ NST over resultant cracking in the HAZ. pletely solidified and the material recov­ CHLOC.NST. (E-E' in Fig. 13) ers its ductility. Mechanically, the BTR Many mechanisms have been pro­ CRT at TL from TL 69°C/s represents the regime over which the posed to describe weld solidification over CHLoc TL, cracking. Among these, the shrinkage- ductility of the material is essentially (A-A' in Fig. 13) brittleness theory (Refs. 49-51), the zero and, thus, susceptible to cracking. strain theory (Refs. 52-54), the general­ The ductility of the material during a ized theory (Refs. 55-58), and the tech- weld thermal cycle can be determined

WELDING RESEARCH SUPPLEMENT I 141-s , it follows Based on this discussion t at t a poin d represen s (T) woul liquidu h experi­ t in the HAZ whic resent a poin r away l to the NST. Farthe perature equa the cor­ s rapidly and perature decrease DRT increases. responding ) in the HAZ could pendix B for definition temperature field be constructed on a t the weld is pro­ in Fig. 7. Assuming tha k , while a pea boundary d fusion the wel rep­ T would l to the NS e equa temperatur k tem­ l cycle with a pea ences a therma lines represent the liquidus. The dashed t temperatures. The isotherms of differen Point A represents the instantaneous liq­ periences a peak temperature equal to k tem­ n boundary, the pea from the fusio n (NDR) (see Ap­ that a nil-ductility regio pool, as illustrated surrounding the weld o left at an instant gressing from right t located at point O, with the weld center e equal to the alloy pool at a temperatur L e of the weld taneous solid/liquid interfac fusion boundary is represented by AA'. uidus temperature. Point B represents the NDT. CB represents the isotherm of and on-cooling, would then represent various peak temperatures used to de­ a DRT can be determined that is specific NST. This thermal cycle achieves its isotherm at point E'. Based on the on- s the instan­ the ellipse HDAG represent the point at which the thermal cycle ex­ the NDT. Temperatures along the line AB, the transition between on-heating set of the on-cooling hot-ductility tests, parallel to the fusion boundary toward same thermal cycle. For example, Point E and E' are in the same thermal cycle peak temperature at point E and falls off resents a regime in which the material NDR is defined by the NDT isotherm, line CB, and the various DRTs, line BA', havior, respectively. This NDR repre­ hibits negligible ductility. According to nisms, which states that cracking results NDR should be equivalent to an area in termine on-cooling ductility. For each to a given peak temperature along AB. tersection of the DRT isotherm and a line from the point of the peak temperature and the peak temperature are in the along line EE' reaching the DRT cooling hot ductility data, line EE' rep­ process can be used to determine the 7. Thus, the thermal envelope of the sents an area in which the material ex­ ment of the liquation cracking mecha­ This DRT can then be located at the in­ the cooling direction, because the DRT with a peak temperature equal to the exhibits negligible ductility. A similar entire DRT curve BA', as shown in Fig. the criterion assumed in the develop­ from the localized loss of ductility, this cracking. By definition, it is the thermal CSR (Appendix B). for the on-heating and on-cooling be­ the HAZ that is susceptible to liquation NST Y 5% augmented strain A-286 stainless steel steel stainles s A-286 On-Cooling CSR range is constant, the on-cooling nil- previous study (Ref. 37) showed that dif­ range. From a physical sense, different peak temperatures for the on-cool ing test the on-heating nil-ductility temperature general, as the peak temperature is in­ creased above the NDT, the DRT de­ creases, resulting in a net increase in the represent different locations in the HAZ. ductility temperature range is a function of the peak temperature. Results from a ferent peak temperatures resulted in a variation in the corresponding DRT. In on-cooling nil-ductility temperature Thus, a peak temperature equal to the 0 15 20 0 5 1 2% — 3%, 5% and 7% /^"\ / //N. NcV- s weld enter (mm) boundary from the instantaneou j

On-Heating CSR

0 -5

n c D D

O 10-5 0 5 10 15 20

p p

q p

0.2 1.0 0.6 0.8

y y boundr n fusio e th m ro 0 0.2 g. (0 9? o n Distance along the fusio

(LULU) 1 00 b •" "o o CD O c E ro 13 o !3 o c o & 03 c to 0.0 Distance along the fusion boundary from the instantaneous weld center (mm) •8 JQ Tr. 0.4 'cn •o HAZ region for A-286 at different strain levels F/g. 9 — The locus of crack tips in the cracked using the hot-ductility test. According are two temperature ranges within ductility. These are the on-heating nil- to this test (Fig. 3), a material loses all ductility when the temperature reaches obtained from the longitudinal-Varestraint test. ature range between a peak temperature (T) and the corresponding DRT. While ductility temperature range, which is the temperature regime between the NDT and T, and the on-cooling nil-ductility temperature range, which is the temper­ ductility test results indicate that there which the material exhibits negligible the NDT on heating and recovers duc­ tility at the DRT on cooling. The hot- Fig. 10— The crack-susceptible region (CSR) of A-286 obtained from the longitudinal-Vare­ 142-s I APRIL 1993 and solid from alternate sides of the same sample. P L straint test. Symbols represent crack tips in both sides of three individual samples, with open Hot ductility test data are not suffi­ cient, however, to absolutely define the entire thermal CSR in Fig. 7. Since on- 1.0 cooling hot ductility tests are very diffi­ A-286 stainless steel cult to perform using peak temperatures above the NST, the value for the DRT's E 0.8 between point E' and A' in Fig. 7 can­ E 4 -i not be obtained using the conventional hot-ductility test. 0.6 Hot-Ductility Test Results 4g o CO In this study, A-286 and Type 310 6 0.4 stainless steels were investigated to ver­ E ify the proposed hypothesis. The on- E heating and on-cooling hot ductility 'ro 0.2 curves for the two alloys are presented 2 in Fig. 8. The NST for both alloys was approximately 1350°C (2462°F). The 0.0 on-heating ductility of the A-286 de­ 2 4 creased rapidly above 11 50°C (21 02°F) Augmented Strain (%) and approached zero at 1200°C (2192°F). On-cooling from the NST, the ductility of this alloy did not recover sig­ Fig. 11 — The maximum crack length at different augmented strain levels for A-286 obtained nificantly until it had been cooled to from the longitudinal-Varestraint test. below 1050°C (1922°F). Thus, for A- 286, the NDT is 1200°C and the DRTNST the size of the cracked HAZ region ex­ creased from 0.5% up to 3%, with a is 1050°C. Type 310 stainless steel ex­ panded for the strain level ranging from threshold strain of about 0.5%. Above hibited similar on-heating ductility be­ 0.5 to 3% as shown in Fig. 9. Above a 3% strain the MCL was essentially con­ havior with a NDT of 1 325°C (241 7°F). saturated strain level of 3%, the size and stant at 0.80 mm (0.03 in.) from the fu­ However, ductility recovery was much shape of this cracked HAZ region is es­ sion boundary. more rapid as evidenced by a DRTNST sentially constant. This saturated of 1325°C (241 7°F). The hot-ductility cracked HAZ region was previously de­ Development of the Thermal CSR Based on results (Table 6) indicate that the on- fined as a crack-susceptible region the Longitudinal-Varestraint Test Results heating nil-ductility temperature ranges (CSR). As shown in Fig. 10, the weld pool (TL- NDT) for A-286 and Type 31 0 are progressed from right to left. The CSR The size of the CSR shown in Fig. 10 222° and 61 °C (400° and 11 0°F), and expanded on weld heating and shrank depends on the material tested and the the BTR at a point that experiences a upon cooling. The maximum width of welding parameters employed during peak temperature equal to the NST (NST the CSR occurs at the transition between testing. In order to directly compare the - DRTNST) are 300° and 25°C (540° and the on-heating and on-cooling regions. CSR among different materials, the weld­ 45°F), respectively. The maximum widths of the cracked ing process variables must be isolated HAZ region at different strain levels, as from the material factor. One way to Experimental Verification of the defined by the maximum HAZ crack achieve this is to normalize the test re­ Thermal Crack-Susceptible Region length (MCL), are shown in Fig. 11. The sults with respect to the thermal condi­ MCL increased as augmented strain in­ tions the HAZ experienced during test- The thermal CSR is material-specific and is inherent to the HAZ surrounding weld pools made with any of the con­ i^__^ Longitudinal-Varestraint Test ventional fusion welding processes. This 140d '^"i>j-a_^3 rn A-286 stainless steel region can be revealed by applying a N5 —Sin • sufficient tensile strain on a material, 1200 while a fusion welding process is occur­ ring. This can be effectively achieved §1000 using the spot- and longitudinal-Vare­ CD straint tests.

Longitudinal-Varestraint Test Results CF > O O O emperat u The longitudinal-Varestraint tests were performed only on A-286, with | 400 Q_ augmented strain levels ranging from 0.5 to 7%. The locations of each crack tip 200 in the HAZ of the as-tested specimen Temperature Gradient = 286°C/mm surface were identified relative to the in­ n terception of the instantaneous solid/liq­ 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 uid interface and the fusion boundary. Distance from the fusion boundary (mm) The profile of these crack tips and the fusion boundary circumscribed a Fig. 12 — Peak temperatures of weld thermal cycles at different locations in the HAZ during cracked HAZ region. It was found that longitudinal-Varestraint testing.

WELDING RESEARCH SUPPLEMENT I 143-s cooling direction. With a very narrow peak temperature range from the liq­ N Temperature at crack tip uidus, as discussed in this study, this line < On-Heatin can be assumed to be perpendicular to thermal CSR the fusion boundary. Line A"BA' repre­ sents the profile of crack tip tempera­ tures. The crack tip temperature (Ttip) 2 :5 can be experimentally derived by the to following equation using a point along a> a_ line AB with the same distance from the E fusion boundary as a reference. CD

T tip • (CRT x CHL OH/OC /VT) (2)

where, TP = the peak temperature of a thermal cycle that the point of interest = tne experienced; CHL0H/OC length measured parallel to the fusion bound­ ary between the crack tip and the point which experiences the peak tempera­ Fig. 13 — Schematic illustration of thermal CSR obtained from the longitudinal-Varestraint ture of the same thermal cycle. This test. length is designated as CHLQH for the crack tip located in the on-heating por­ r ing. This, in fact, provides a material- represents the instantaneous liquidus tion, and CHLQC f° the on-cooling por­ tion (Fig. 1 0); CRT = the average cool­ specific parameter, termed the thermal temperature (TL). The temperature at the CSR, which uniquely describes the CSR fusion boundary after the weld pool has ing (or heating) rate over CHLoc (or CHLQH); V = welding speed during lon­ in terms of temperature. passed is represented by AA'. Point B T gitudinal-Varestraint testing. The thermal conditions of the HAZ represents the crack tip temperature at during testing of A-286 were obtained the transition between on-heating (left For example, Point E and E' are in the using implanted thermocouple wires. of AB) and on-cooling regions (right of same thermal cycle with a peak temper­ The peak temperatures at different loca­ AB). The temperature at Point B (TB) can ature equal to TE, as they are equidistant tions in the HAZ are shown in Fig. 1 2. be derived by the equation: from the fusion boundary. This thermal The approximate temperature gradient cycle achieves its peak temperature at point E and falls off along line EE'. Crack­ in the region immediately adjacent to TL - (MCL X GT) (1) the fusion boundary was 286°C/mm ing is observed within the area of A"BE'A'A". Thus, CHLoc at point E is (515°F/0.04 in.). where, GT is the gradient of peak tem­ In conjunction with the thermal cycle peratures during longitudinal-Vare­ represented by the length EE'. The time data, the spatial CSR (Fig. 10) can be straint testing. Temperatures along the required (tEE<) for the temperature to fall translated into the thermal CSR as illus­ line AB would then represent various from point E to point E' is the distance between these two points (EE') divided trated in Fig. 1 3. The weld is progress­ peak temperatures (TP) the HAZ experi­ ing from right to left at an instant when enced during testing. Note that in actual by the welding speed during testing (VT), the solid/liquid interface intercepts the welding fabrication, line AB would in­ tEE' = EE' / VT. The temperature in the fusion boundary at point A. Point A thus cline to the fusion boundary toward thermal cycle reaches point E' (TE) at a time period tEE> after it has achieved the peak. Or, mathematically, TE. = TE-(CRT X EE' / VT). CRT is the average cooling 2.5 Spot-Varestraint Test rate from the peak temperature TE over cooling time = 0 sec a time period of tEE*. Pertinent thermal CSR data for A-286 obtained from the E 20 A-286 longitudinal-Varestraint test are summa­ I- rized in Table 7. c~ fj) CD 1 S Spot-Varestraint Test Results _l Y m<> The on-heating spot-Varestraint test »_ results for both A-286 and Type 310 al­ O 1 0 loys are shown in Fig. 14. Because there b"3 E was no time for the HAZ to cool before cracking occurred, the maximum HAZ 0.5 Type 310 crack length (MCL) represented the ja o- width of the cracked HAZ under a spe­ cific strain level when the HAZ was at 0.0 peak temperature. For Type 31 0 stain­ 2 3 4 5 less steel, above a threshold strain level Augmented Strain (%) of 1 %, the MCL increased as augmented strain increased up to 3%. Above 3% strain, designated the saturated strain, Fig. 14 — On-heating spot-Varestraint test results of A-286 and Type 310 showing the width of the MCL was essentially constant at 0.39 the cracked HAZ at different augmented strain levels.

144-s I APRIL 1993 mm (0.01 5 in.) from the fusion bound­ ary. 3.0 A threshold strain could not be de­ Spot-Varestraint Test augmented strain = 5% termined for A-286, since significant cracking was observed at the lowest level of augmented strain (0.5%). MCL increased slightly between 0.5% and 3.0% and was essentially constant above On-Cooling CSR 3.0%. The maximum width of the of A-286 cracked HAZ for A-286, as defined by the MCL in the saturated strain region, was 1.93 mm (0.076 in.). Note that above a saturated strain level, the width of the cracked HAZ during weld heat­ ing is essentially constant. As defined previously, the width of the cracked HAZ at the saturated strain level repre­ sents the extent of the CSR during heat­ 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 ing. Cooling Time (sec) The on-cooling spot-Varestraint tests were performed with a saturated strain Fig. 15 — On-cooling spot-Varestraint test results of A-286 and Type 310 showing the on- of 5% for both alloys to characterize the cooling portion of the CSR. extent of the CSR during weld cooling. It is important to point out that at a longer cooling time, cracks may propagate 1400°-1450°C for Type 310) (Ref. 59). fusion boundary (T ) after arc extinc­ across the fusion boundary into the fu­ FB In conjunction with the thermal cycle tion can be derived from the cooling sion zone. The portion of crack length data in the HAZ, Fig. 1 5 can be trans­ time (t ) following the equation: inside the fusion zone did not represent c lated to represent the on-cooling por­ the material behavior in the HAZ, thus, tion of the thermal CSR, as schemati­ T = T - (CR X t ) (4) was not taken into account in this study. FB L T c cally shown in Fig. 1 7. Point A repre­ Results from the on-cooling tests in Fig. sents the instantaneous liquidus temper­ where, CR is the average cooling rate 1 5 show the on-cooling portion of the T ature (T ). Point B represents the crack at the fusion boundary over a period of CSR. Again, the MCLs represent the dis­ L tip temperature determined with no tc. Line BA' represents crack tip temper­ tance of crack tips from the fusion cooling time (t = 0 s). The temperature atures at various cooling times. These boundary. Both alloys behaved in a sim­ c at Point B (T ) can be derived following crack tip temperatures (Ttip) can be de­ ilar manner, in that the width of the CSR B the equation: rived using a point along line AB with shrank as cooling time increased. For A- the same distance from the fusion 286, it disappeared after a cooling time (MCL X GT (3) boundary as a reference and following of 4.1 3 s, while for Type 310 this region the equation: persisted for only 0.41 s. where, GT is the gradient of peak tem­ peratures during spot-Varestraint test­ T (5) Development of the Thermal CSR Based on tip - Tp • (t- X CRT) the Spot-Varestraint Test Results ing, and MCL is the maximum HAZ crack length performed with no cooling where, TP = the peak temperature of a time (t = 0). Temperatures along line The methodologies utilized to de­ c HAZ thermal cycle which the point of AB represent different peak temperatures velop the thermal CSR based on the spot- interest experienced during spot-Vare­ achieved in the HAZ. Since the cooling Varestraint test results are basically the straint testing; tc = the cooling time over time is equivalent to the period after same as those for the longitudinal-Vare­ which cracking is observed at the point weld pool has passed in a continuous straint test. The on-cooling thermal CSR of interest; CRT = the average cooling weld. Line AA' would then represent var­ can be developed by combining the spa­ rate of a thermal cycle with a peak tem­ ious temperatures at the fusion bound­ tial CSR (Fig. 1 5) and the thermal con­ perature of TP from the peak tempera­ ary after arc extinction. Point A' locates ditions experienced in the HAZ during ture over a time period of tc. the greatest cooling time in the spot- testing. The thermal conditions in the For example, Point E and E' are in the Varestraint test above which cracks do HAZ during testing were determined same thermal cycle with a peak temper­ not propagate across the fusion bound­ using the implanted thermocouple tech­ ature equal to TE, as they are equidistant ary into the HAZ. The temperature at the nique, as described previously. The peak from the fusion boundary. Cracking at temperatures of thermal cycles at differ­ ent locations in the HAZ are shown in Fig. 16 with an approximate gradient of Table 8 — Pertinent Thermal CSR Data Obtained from the Spot-Varestraint Test 101°C/mm (182°F/0.04 in.) for A-286 and 156°C/mm (281°F/0.04 in.) for Type A-286 Type 310 310. The peak temperature at the fusion MCL (t = 0) 1.93 mm 0.39 mm boundary was determined to be 1422°C L tc.NST 4.10 s 0.20 s (2592°F) for A-286 and 1 386°C (2527°F) tc.TL 4.13 s 0.41 s for Type 310. These temperatures are CT 101°C/mm 156°C/mm believed to approximate the liquidus CRT at NST from NST over tc NST, 68°C/s 140°C/s temperatures (TL) since these data are (E-E' in Fig. 17) very close to published melting ranges CRT at TL from TL over tc TL, 94°C/s 220°C/s (1370°-1430°C for A-286 and (A-A' in Fig. 17)

WELDING RESEARCH SUPPLEMENT I 145-s spot-Varestraint tests and comparing the

1 extent of the HAZ thermal CSR obtained 1400 from the three methods. E A 286 1200 ^ kfi-TXL__^,n " Characterization of the Ductility of Material in the Thermal CSR Obtained from 5-1000 ~"~° . Longitudinal- and Spot-Varestraint Tests Type 310 One of the important criteria as­ sumed in the development of liquation theories is that cracking results from the c n o o o localized loss of ductility. In order to

temperatur e ( prevent liquation cracks from propagat­ f 400 - ing, the material must exhibit sufficient CO ductility to accommodate the thermally Temperature Gradient 200 " A-286 = 101°C/mm induced and/or externally applied strain. • Type 310= 156°C/mm Based on this criterion, to a first approx­ o imation, it would be appropriate to as­ 0.0 0.5 1.0 1.5 2.0 2.5 3.0 sume that the material at the crack tip Distance from the fusion boundary (mm) exhibits localized ductility with the mag­ nitude equal to the applied augmented Fig. 16 — Peak temperatures of weld thermal cycles at different locations in the HAZ for A- strain during both the longitudinal- and 286 and Type 310 during spot-Varestraint testing. spot-Varestraint testing. Thus, in con­ junction with the thermal cycle data, Figs. 11 and 14 can be translated to point E is observed over a cooling time tion, and the various DRTs for the on- physically show the localized ductility of the material at different temperatures period of tE-. Thus, the temperature in cooling portion — Fig. 7. In the longi­ during weld heating as shown in Fig. 1 8 the thermal cycle reaches point E' (TE0 tudinal-Varestraint test, the profile of the at a time period t < after it has achieved crack tip temperatures in the HAZ for for the longitudinal-Varestraint test and E Fig. 19 for the spot-Varestraint test. In the peak. Mathematically, T -(CR the test performed with the saturated E T these figures, the localized ductility is X t 0, where, CR is the average cool­ strain determines the envelope of the E T analogous to the augmented strain. The ing rate from T over a period of t >. Per­ HAZ thermal CSR — Fig. 1 3. In the spot- E E temperature of the material represents tinent thermal CSR data for alloys A-286 Varestraint test, the crack tip tempera­ the temperature at the crack tip (T ) dur­ and Type 31 0 obtained from the spot- ture of the on-heating tests performed tip ing testing, which is translated from the Varestraint test are summarized in Table with the saturated strain determines the transition temperature of the on-heating MCL following the equation: thermal CSR, and the various crack tip temperatures of the on-cooling tests per­ Discussion Ttip = TL - (MCL X GT) (6) formed with the saturated strain enclose Results from this study have indicated the on-cooling thermal CSR — Fig. 1 7. Thus, results from the on-heating por­ that the HAZ thermal CSR of a material The correlation and the methodology of tion of the longitudinal-Varestraint test can be determined using any of the three data interpretation of the three tests can (Fig. 1 8) showed that A-286 exhibited tests. In the hot-ductility test, the HAZ be illustrated by characterizing the duc­ substantial localized ductility (greater thermal CSR is circumscribed by the tility of the material in the thermal CSR than 7% in this case) for temperatures NDT isotherm for the on-heating por­ obtained from the longitudinal- and below 1194°C (21 81 °F). As the temper­ ature increased beyond 1194°C, the lo­ calized ductility dropped dramatically to 3%, and further decreased to 0.5% at higher temperatures below TL. In con­ trast, results from the on-heating spot- Varestraint tests (Fig. 19) indicated that A-286 exhibited considerable localized ductility (greater than 5% in this case) Temperatures at crack tip for temperatures below 1227°C (2241 °F). The localized ductility On-Cooling thermal CSR dropped substantially to 3% as the tem­ perature increased beyond 1227°C. With a further increase in temperature, the localized ductility decreased gradu­ ally to 0.5% at a temperature of 1 240°C, and A-286 exhibited essentially no duc­ tility (less than 0.5%) for the tempera­ ture range between 1 240°C (2264°F) and TL. On the other hand, Type 310 A A' (Fig. 1 9) exhibited greater than 7% lo­ Temperature at the fusion boundary after arc extinction calized ductility for temperatures below (Translated from cooling time) 1 325°C (241 7°F). The localized ductil­ ity dropped rapidly to 3% as the tem­ Fig. 17 —Schematic illustration of the on-cooling portion of the thermal CSR obtained from perature exceeded 1 325°C, and further the spot-Varestraint test.

146-s I APRIL 1993 decreased to approximately 1 % at tem­ peratures slightly below T . L Longitudinal-Varestraint Test Note that there is a transition temper­ ature, 1 1 94°C or 1 227°C for A-286 as obtained from longitudinal- and spot- _ 6 Varestraint, respectively, and 1325°C / for Type 310, above which the localized ductility of the material dramatically drops to essentially zero. The tempera­ 8 On-Heating thermal ture range between TL and the transition Q "CSR of A-286 temperature defines the extent of the on- TD CD heating thermal CSR. In order to deter­ N to mine this transition temperature, it is es­ 0 sential the saturated augmented strain o be employed during both longitudinal- and spot-Varestraint testing. It is appar­ I ent that saturated augmented strain is one of the important parameters for both 1450 1350 1250 1150 of these tests, yet the importance of this parameter has generally been neglected Temperature at crack tip (C) (Ttp =TL-MCLXGT) by most investigators who have used these two tests to quantify liquation Fig. 18 —The localized ductility in the on-heating thermal CSR of A-286 obtained from the cracking. longitudinal-Varestraint test. It is interesting that the transition tem­ peratures for A-286 defined with the lon- in the experimentally determined local­ did not influence the results. gitudinal-and spot-Varestraint tests are ized ductility in the on-heating thermal The localized ductility in the on-cool­ essentially the same, if reasonable ex­ CSR from these two tests, with a greater ing portion of the thermal CSR deter­ perimental error is taken into account. localized ductility in the former. mined with the longitudinal- and spot- However, the localized ductility in the It was observed that both the longi­ Varestraint was not evaluated in this on-heating thermal CSR is inconsistent. tudinal- and spot-Varestraint test speci­ study. However, it is definitely less than This discrepancy can be rationalized by mens did not conform perfectly to the 5% for the two alloys investigated, since considering the different local variations surface of the die block. Specimens in 5% augmented strain was employed for in strain accommodation on the speci­ both tests tended to kink slightly at the men surface during testing. the on-cooling spot-Varestraint tests, highest augmented strain level (7%). The and the CSR was obtained for A-286 at As shown in Fig. 20, the HAZ liqua­ kinking of the specimen may thus result 5% augmented strain from the longitu­ tion cracks are parallel to each other in in a higher applied strain than that cal­ dinal-Varestraint test. Thus, the thermal the longitudinal-Varestraint test. Be­ culated. However, it did not affect the CSR shown in Figs. 13 and 1 7 represents cause the applied augmented strain is ductility transition temperature (Figs. 18 a regime within which the material (both theoretically calculated based on the and 19), since it was determined at a sat­ A-286 and Type 310) exhibits less than elongation of the top surface of the spec­ urated strain level of 3% for both alloys. 5% ductility (or perhaps 3%, since 3% imen, the opening of the cracks would As shown in Figs. 18 and 19, even with is the saturated strain for both alloys in accommodate some amount of the ap­ a higher level of augmented strain both tests). plied strain. The as-tested specimen sur­ (>3%), the transition temperature does Considering the complexity of the face clearly showed that the opening of not change and, thus, specimen kinking the crack is wider at the trailing edge of strain field during the longitudinal- and the weld pool and becomes narrower toward the welding direction. This im­ plies that the strain distribution is not uniform with a greater strain at the trail­ 8 Spot-Var sstraint Test TL T up On-Heati ig thermal CSR ing edge of the weld pool. Thus, the ac­ 7 i tual augmented strain for the crack lo­ k*-Type -»J cated at the transition between on-heat­ 6 310 • ing and on-cooling regions would be £ TL I T lower than the calculated strain. In con­ Up = 5 i c trast, the HAZ liquation cracks are radi­ I ally oriented relative to the weld center n Q ' I in the spot-Varestraint test, as shown in 4 > Fig. 21. The crack with a maximum = 3 i length is always the widest and located ro ° perpendicular to the longitudinal di­ 8 1 mension of the specimen, indicating that - 2 / > the localized strain was highest at that 1 point. It is postulated, therefore, the lo­ i * f o calized augmented strain is greater than 0 the calculated strain at this location. 1' t50 1400 1350 1300 1250 1200 11 50 Based on this argument, it follows that Temperature at crack tip (C) (T = T - MCLXG ) the difference in the localized strain dis­ ap L T tribution in the longitudinal- and spot- Varestraint tests results in the difference Fig. 19 — The localized ductility in the on-heating thermal CSR of A-286 and Type 310 ob­ tained from the spot-Varestraint test.

WELDING RESEARCH SUPPLEMENT I 147-s Table 10 summarizes the procedures for determining the thermal CSR using the hot-ductility, longitudinal- and spot- Varestraint tests. It is of importance to note that the shape and size of the ther­ mal CSR reflects the compositional and metallurgical nature of the material and is independent of any process or restraint factors. The thermal CSR allows com­ parison of cracking susceptibility among different materials. Using this method­ ology, the thermal CSR surrounding a moving weld pool has been determined for A-286 and Type 310, as shown in Fig. 22. It is clear that A-286 exhibits a larger BTR and correspondingly wider -- i" thermal CSR than Type 310, indicating that A-286 would exhibit greater extent 1 of HAZ liquation cracking than Type 310 v; 1 under the same thermal and restraint s *WMP^PP I conditions. In applying this material-specific en­ velope to actual welding conditions, the I A spatial extent of the CSR traveling with Fig. 20 — A micrograph showing the HAZ liquation cracks of A-286 tested withI 2 5%mm aug­ 1 the weld pool can be determined as a mented strain from the longitudinal-Varestraint test. The welding direction is from right to function of the welding parameters, left. which determine the thermal conditions in the HAZ. Table 11 provides a trans­ formation matrix for determining the magnitude of the thermal and spatial spot-Varestraint testing and different straint test performed with saturated thermal and mechanical conditions for strain. It also correlates with the temper­ CSR. As a result, the thermal envelope hot-ductility testing, it would be appro­ ature range over which the MCL occurs determined via hot-ductility, spot- and priate to equate this very low-ductility in the spot-Varestraint test performed longitudinal-Varestraint weldability tests thermal CSR to the nil-ductility region with saturated strain at no cooling time can be directly applied to actual weld­ ing situations if the thermal gradient (NDR) as determined from the hot-duc­ (tc = 0). The temperature range between tility test. The equivalence between the a peak temperature and the correspond­ (Gw) and cooling rate (CRW) can be CSR and the NDR provides experimen­ ing DRT is the temperature range in a measured or derived empirically. The tal evidence for the criterion assumed cooling cycle within which the material contribution of the welding process/pa­ in the development of liquation crack­ exhibits negligible ductility, previously rameter factor in cracking susceptibility ing theories, which states that liquation defined as the BTR. This temperature can also be visualized from the con­ cracking results from the localized loss range in turn correlates with the cool­ struction of the spatial CSR. of ductility. ing time (tc) over which cracking is ob­ served in the spot-Varestraint test, and Evaluation of Traditional Correlation and Interpretation of Test the on-cooling portion of the cracked Methodologies Results Obtained from the Three Methods HAZ length (CHLoc) in the longitudi­ nal-Varestraint test. Despite the differ­ The development of this methodol­ Two basic elements with metallurgi­ ence in testing techniques, the correla­ ogy has provided a means to link to­ cal significance can be visualized in the tion obtained for both A-286 and Type gether weldability test results, cracking development of the thermal CSR. These 31 0 is excellent, reinforcing the com­ theories and material properties, and to are the extent of the thermal CSR during plementary nature of these techniques elucidate the interpretation of the hot- weld heating and the BTR on cooling. for predicting HAZ liquation cracking. ductility, spot- and longitudinal-Vare­ The correlation between the theoretical This good correlation also provides evi­ straint test results. The utility of tradi­ hypothesis and the experimental results dence that the thermal CSR is a mate­ tional methodologies can also be eval­ can be demonstrated by comparing the rial-specific parameter, and a true quan­ uated based on this approach as dis­ magnitude of these two elements. As tification of HAZ liquation cracking sus­ cussed below. listed in Table 9, very good correlation ceptibility. was obtained for A-286 and Type 31 0 Hot-Ductility Test stainless steels among the three tests. Application to Actual Welding Conditions The development of this correlation The construction of the NDR using also clarifies the interpretation of results Although the thermal CSR is devel­ the hot-ductility test was first demon­ from the three tests. Physically, the tem­ oped using weldability testing tech­ strated by Duvall, ef al. (Refs. 32, 33). perature range between the TL and NDT niques, it is a material-specific parame­ Unfortunately, they could not develop represents the extent of the thermal CSR ter and can be applied to any of the con­ a relationship between the NDR and during weld heating. This temperature ventional fusion welding processes. The HAZ liquation cracking susceptibility. range correlates with the temperature application of the thermal CSR in a con­ Based on a rough calculation of the weld range over which the MCL occurs at the tinuous weld can be clearly demon­ size, he suggested that HAZ liquation transition between on-heating and on- strated with the longitudinal-Varestraint cracks propagated outside of the NDR cooling regions in the longitudinal-Vare­ test as illustrated in Figs. 10 and 20. [i.e., via solid-state cracking) and the

148-s I APRIL 1993 amount and rate of ductility recovery were the most sign ificant factors govern­ ing liquation cracking susceptibility. De­ spite the high levels of strain used in this investigation during spot- and longitu­ dinal-Varestraint testing, cracks were never observed to propagate outside of the NDR. Thus, the importance of duc­ tility recovery below the DRT, as sug­ gested by Duvall, et al., appears to be negligible based on the results from both A-286 and Type 310 stainless steels. The NST-DRT temperature range has been utilized extensively as a quantita­ tive cracking index (Refs. 11,27, 30, 31). This index is restrictive in that it only represents the BTR at a specific location in the HAZ. Since the highest peak tem­ perature that can be easily achieved in the hot-ductility test is in the vicinity of the NST, this approach only provides in­ formation for a point in the HAZ that is removed from the fusion boundary. When considering that HAZ liquation cracking is commonly observed in the region immediately adjacent to the fu­ Fig. 21 — A micrograph showing the HAZ liquation cracks of A-286 tested with 5% aug­ sion boundary, this cracking index may mented strain from the spot-Varestraint test. not be representative of actual behavior or predictive of liquation cracking sus­ ceptibility. This is particularly true if the difference between TL and the NST is large. This problem is further exacer­ bated when peak temperatures below the NST are used to develop on-cooling hot ductility behavior, as is the case in many of the published reports which use hot-ductility data to predict HAZ liqua­ tion cracking susceptibility. The hot strength criterion proposed 1050°C in 1 960s (Refs. 27-29) is commonly uti­ lized to rationalize the inconsistency be­ 1035°C tween the hot-ductility results and crack­ ing observed in the field experience. However, considering the fact that all materials exhibit extremely low strength in the temperature range in which crack­ ing occurs, the elastic strain induced by this hot strength would be negligible. In addition, most alloys exhibit a critical y/^///////////////^J^^ level of threshold strain for cracking in the spot- and longitudinal-Varestraint (A) A-286 testing (e.g., 1% for Type 310 in Fig. 14). The observation of significant threshold strain to cause cracking indicates that the materials have the ability to resist crack­ V1350°C ing even under a substantial applied \1325°C 1325°C stress. This suggests, therefore, that it is the hot ductility rather than hot strength 1386°C 1296°C that dominates liquation cracking.

Spot-Varestraint Test

The conventional cracking indexes for the spot-Varestraint test, the thresh­ ZZZZZZZZZ- old strain to cause cracking or the de­ gree of cracking (MCL or TCL) over ar­ (B) Type 310 bitrary augmented strain levels and an arbitrary cooling time, are even more complicated from a metallurgical stand- Fig. 22 — Thermal CSR of A-286 and Type 310 stainless steels.

WELDING RESEARCH SUPPLEMENT I 149-s However, the thermal histories in this Table 9 — Correlations among the Hot-Ductility, Spot- and Longitudinal-Varestraint Test cracked region along the crack are dif­ Results ferent among these tests. In the Transvarestraint test, the weld center- Spot-Varestraint Longitudinal- line crack falls in the same thermal cycle Hot-Ductility Test Test Varestraint Test (Ref. 1 0). In the case of HAZ cracks in A-286, Extent of the - NDT = MCL X CT = MCL X GT the longitudinal- and spot-Varestraint on-heating CSR 1422- 1200 = 222°C 1.93 X 101 = 195°C 0.80 X 286 = 2293C tests, the crack which is perpendicular to the longitudinal direction of the sam­ Type 310, Extent of TL - NDT = MCL X GT = the on-heating CSR 1386 - 1325 = 61 °C 0.39 X 156 = 60°C Not Determined ple follows different thermal cycles in the HAZ, but cracking occurs at the A-286, NST - DRTNST = tc X CRT = CRT X CHLOC/VT = same instant relative to the peak tem­ BTR at NST 1350- 1050 = 300°C 4.10 X 68 = 279°C 66 X 11.5/2.54 = 299°C perature. In the case of solidification

Type 310, NST - DRTNST = tc X CRT = Not Determined cracks in the longitudinal-Varestraint BTR at NST 1350- 1325 = 25°C 0.20 X 140 J 28°C test, cracking occurs following different thermal cycles and at different instants relative to the peak temperature. As a point. First, the crack length (either MCL of the test results in relation to the met­ result of the difference in thermal his­ or TCL) does not provide a material-spe­ allurgical aspects of liquation cracking tory of the cracks generated in the three cific quantification; rather, it represents susceptibility. This specific width of the tests, the cracking temperature range in a specific material/process/restraint CSR would neither correlate with the on- the Transvarestraint test determined by combination. Any variation in the test cool ing behavior obtained from the hot- the MCL defines the BTR. The HAZ conditions such as the welding parame­ ductility test nor reflect the cracking be­ cracking temperature range determined ters and the augmented strain level havior of the material during actual by the MCL in both the longitudinal- and would affect the test results. Second, the welding. spot-Varestraint tests, on the other hand, saturated strain has been found to be the The metallurgical significance of the defines a thermal width of the cracked parameter which must be used to MCL can also be revealed from this region. uniquely determine the size of the CSR. study. By representing MCL as the dis­ As shown in Fig. 14, the MCL is depen­ tance from the fusion boundary to the Longitudinal-Varestraint Test dent on the augmented strain for the crack tip, it can be directly correlated strain level below the saturated strain. with the extent of the CSR. Notice that The discussion of the conventional Thus, the MCL determined below the this manner of representation of the MCL cracking indexes for the longitudinal- saturated strain, such as in most of the also dramatically reduces the human Varestraint test, TCL or MCL, foi lows that published literature, would only repre­ error in crack length measurement. His­ for the spot-Varestraint test, except that sent part of the CSR. It does not provide torically, this cracking index has not TCL is the cumulative length of all cracks a comprehensive measure of cracking been used as extensively as the TCL in with the same strain level at different susceptibility. It is also of importance to the published literature. In the spot- thermal conditions in the longitudinal- note that the MCL determined in this Varestraint test, the TCL represents the Varestraint test. Both the MCL and TCL manner is very sensitive to the localized cumulative length of all cracks with the cracking indexes represent results for a strain distribution of samples during test­ same thermal conditions at different specific material/process/ restraint com­ ing. Thus, test results from a laboratory strain levels. It is not material-specific bination. The magnitude of these in­ may not be easily reproduced in another and may be influenced by grain size and dexes would vary if different welding laboratory due to the difference in equip­ strain localization at the grain bound­ parameters or augmented strain are em­ ment setup. Third, the cooling time has ary. As a result, TCL is not metallurgi­ ployed. The CHL criterion, proposed by been proved to be the parameter which cally significant with respect to HAZ li­ Lundin, etal. (Ref. 14), can be normal­ governs the HAZ temperature during quation cracking. ized with the thermal conditions expe­ testing and, thus, the on-cooling behav­ rienced during testing to obtain a BTR, ior of material. Tests performed with an Since spot-Varestraint was modified from the original Varestraint test, the ter­ if the tests are performed with a satu­ arbitrary, fixed cooling time will only rated strain and only the on-cooling por­ reveal a specific width of the on-cool­ minologies utilized in quantifying the degree of cracking (MCL or TCL) were tion of the CHL (CHLoc) is considered, ing CSR as shown in Fig. 1 5. The use of as illustrated in Fig. 10 and Table 11. unsaturated strain level during testing also adopted. Physically, the MCL in all three types of Varestraint tests (spot, lon­ However, it is very unlikely that the would, then, only reveal part of this spe­ CHL c at a specific location in the HAZ cific width and increase the complexity gitudinal and transverse) characterizes 0 a region over which cracking occurs. can be precisely determined without the development of the entire CSR.

Table 10 — Development of the Thermal CSR (Fig. 7) Summary

Point/ Theoretical Hot Ductility Spot-Varestraint Longitudinal-Varestraint This paper has described an experi­ I ine Quantity Test Test Test mental approach that allows HAZ liqua­ tion cracking susceptibility to be quan­ A TL — TP at fusion boundary Tp at fusion boundary tified in a manner that is metallurgically CB NDT NDT TL - (MCL X CT) TL - (MCL X GT) E SST NST — significant and material-specific. E' DRTNST DRTNST NST - (tc,NST X CRT) TL - CRT X CHL0C,NST/VT Methodologies have been developed BE' DRT's DRT's Tip as function of tc Tip as function of CHLoc and tested for defining the thermal crack- E'A' DRT's — Ttjp as function of tc Ttip as function of CHLoc susceptible region (CSR) for a material. A' DRTR — TL-UCTLXCRT) TL - CRT X CHLQCJI/VT Any one of three weldability tests can be used to quantify the thermal CSR, namely, the Gleeble hot-ductility test,

150-s I APRIL 1993 and the spot- and longitudinal-Vare­ Table 11 — Transformation Matrix for the Derivation of the Thermal and Spatial CSR straint tests. The results presented here have also provided new insight into the A. The Hot-Ductility Test interpretation and correlation of test data for each of these tests and form the basis Test Result Thermal CSR Spatial CSR of a unified approach for defining a ma­ Width of CSR TL - NDT TL - NDT (TL - NDT)/Gv\ terial-specific temperature regime BTR TP- DRT TP - DRT (TP - DRT) X Vw/CRw within which HAZ liquation cracking B. The Longitudinal-Varestraint Test occurs. Recent results have shown that Test Result Thermal CSR Spatial CSR this methodology can also be success­ Width of CSR MCL MCL X GT MCL X Gr/Gw fully applied to other materials, includ­ BTR CHLoc CHLoc X CRT/VT CHLoc X CRT/VT X VV\ /CRVV ing Alloy 625 and aluminum Alloy 6061 C. The Spot-Varestraint Test (Refs. 60, 61). Test Result Thermal CSR Spatial CSR During the development of this Width of CSR MCL MCL X CT MCL X CT/CW BTR tc t X CR methodology some inherent limitations t T tc X CRT X Vw/CRw in utilizing any of the three tests for pre­ dicting the occurrence of HAZ liquation cracking were also identified. Although The temperature range between the liq­ period between arc extinction and spec­ the thermal CSR represents the temper­ uidus (TL) and nil-ductility temperature imen bending and is the parameter ature regime over which HAZ liquation (NDT) in the hot-ductility test represents which governs the temperature in the cracking may occur, it only serves as an the extent of the on-heating thermal HAZ during on-cooling spot-Varestraint indirect measure of the actual cracking crack-susceptible region. This tempera­ testing. By varying the cooling time, the susceptibility. In order to fully quantify ture range correlates with the tempera­ on-cooling crack-susceptible region can this susceptibility during welding, other ture range over which the maximum be determined. factors such as ductility within the BTR, crack length occurs at the transition be­ 7) For the A-286 stainless steel tested the effect of liquid healing and local tween on-heating and on-cooling re­ in this study, the extent of the on-heat­ HAZ strain accommodation must be gions in the longitudinal-Varestraint test ing thermal crack-susceptible region is considered. Unfortunately, it is very dif­ performed with saturated strain. It also 222°C. The brittle temperature range at ficult to incorporate these factors when correlates with the temperature range the location of the nil-strength tempera­ interpreting weldability test data. As a over which the maximum crack length ture is 300°C. The brittle temperature consequence, the methodology devel­ occurs in the spot-Varestraint test per­ range at the fusion boundary is 387°C. oped here, while providing a material- formed with saturated strain for no cool­ 8) For the Type 310 stainless steel specific quantification of HAZ liquation ing time after arc extinction. tested in this study, the extent of the on- cracking, cannot easily accommodate 3) The temperature range between heating thermal crack-susceptible re­ the variations in weld restraint that often the peak temperature (TP) and the cor­ gion is 61 °C. The brittle temperature dictate the occurrence and severity of responding ductility recovery tempera­ range at the location of the nil-strength cracking. ture (DRT), as obtained from the hot- temperature is 25°C. The brittle temper­ Despite these limitations, the devel­ ductility test, represents the brittle tem­ ature range at the fusion boundary is opment of this methodology has shown perature range (BTR). This temperature 90°C. that the Gleeble hot-ductility and Vare­ range in turn correlates with the cool­ straint tests are effective tools for both ing time over which cracking is observed Acknowledgments quantifying the HAZ liquation cracking in the spot-Varestraint test, and the on- susceptibility in terms of the size of the cooling portion of the cracked HAZ The authors would like to acknowl­ thermal CSR and developing a funda­ length in the longitudinal-Varestraint edge the members of Edison Welding In­ mental understanding of the liquation test. stitute for providing financial support for phenomena. Part 2 of this investigation 4) The localized ductility within the this work. The generosity of Rolled Al­ will describe the metallurgical behavior thermal crack-susceptible region is es­ loys, for providing the Type 31 0 stain­ in the CSR in support of this test method­ sentially zero. These results provide ex­ less steel, is also appreciated. ology. perimental evidence supporting the cri­ terion assumed in the development of References liquation cracking theories, which states Conclusions that liquation cracking results from the 1. Hemsworth, B., Bonizewski, T., and localized loss of material ductility. Eaton, N. F. 1969. Classification and defini­ 1) A new methodology has been de­ tion of high-temperature welding cracks in veloped for quantifying HAZ liquation 5) The importance of saturated aug­ alloys. Metal Construction and British Weld­ cracking susceptibility. This methodol­ mented strain in both the longitudinal- ing lournal 1 (2):5-1 6. ogy quantifies a thermal crack-suscepti­ and spot-Varestraint tests was evaluated. 2. Lippold, J. C, Harwig, D. E., Ernst, S. ble region in the HAZ in which HAZ li­ The saturated strain refers to the applied C, and Nelson, T. 1 989. Development of a quation cracking may occur. It is a ma­ strain above which the maximum crack weldability test data base WELDTEST. un­ terial-specific parameter and represents length remains constant. It was found published research. Edison Welding Institute. a true quantification of HAZ liquation that saturated strain conditions must be Columbus, Ohio. cracking susceptibility. This region can used to uniquely define the temperature 3. Baeslack, W. A. Ill, and Lippold, ). C. be determined using the Gleeble hot- above which the ductility of the mate­ 1987. Evaluation of weldability testing tech­ niques. Proceedings of EWI Annual North ductility, spot- or longitudinal-Vare­ rial drops to essentially zero. This tem­ perature correlates to the on-heating nil- American Welding Research Seminar on The straint tests. Influence of New Materials Developments 2) The thermal crack-susceptible re­ ductility temperature in the hot-ductil­ on Weldability. Columbus, Ohio. ity test. gion can be described by two basic ele­ 4. Lippold, |. C. 1988. Physical simula­ ments: the extent of the crack-suscepti­ 6) The importance of the cooling time tion of welding: a perspective on weldability ble region during weld heating and the in the spot-Varestraint test was evalu­ testing. Proceedings of International Sympo­ brittle temperature range on cooling. ated. Cooling time is defined as the time sium on Physical Simulation of Welding, Hot

WELDING RESEARCH SUPPLEMENT I 151-s Forming and Continuous Casting Processes. Lippold, ). C. 1989. Weldability of a high- less steel. Welding Journal 53(1 1 ):51 7-s to Ottawa, Canada. strength, low-expansion superalloy. Welding 523-s. 5. Prokhorov, N. N. 1962. The techno­ Journal 68(10):418-s to 430-s. 42. Morishige, N., Kuribayashi, M., and logical strength of metals while crystallizing 23. Brooks,). A. 1974. Effect of alloy mod­ Okabayashi, H. 1979. Effects of chemical during welding. Welding Production ifications on HAZ cracking of A-286 stain­ compositions of base metal on susceptibility 9(4):1-8. less steel. Welding Journal 53(11):51 7-s to to hot cracking in austenitic stainless steel 6. Prokhorov, N. N., and Prokhorov, N. 523-s. welds. IIW Doc. IX-1114-79. Nikol. 1 971. Fundamentals of the theory for 24. Lin, W. 1991. A methodology for 43. Gooch, T. O, and Honeycombe, ). technological strength of metals while crys­ quantifying heat-affected zone liquation 1 980. Welding variables and microfissuring tallizing during welding. Trans. JWS cracking susceptibility. Ph.D. dissertation. in austenitic stainless steel weld metal. Weld­ 2(2):109-117. The Ohio State University. Columbus, Ohio. ing Journal 59(8): 233-s to 241 -s. 7. Savage, W. F., and Lundin, C. D. 1965. 25. Nippes, E. F., Savage, W. F., Bastin, 44. Lundin, C. D., Chou, C. P. D., and The Varestraint test. Welding Journal B. T., Mason, H. F„ and Curran, R. M. 1955. Sullivan, C. J. 1 980. Hot cracking resistance 34(10):433-s to 442-s. An investigation of the hot ductility of high- of austenitic stainless weld metal. Welding 8. Lundin, C. D., Lingenfelter, A. C, temperature alloys. Welding Journal Journal 59(8):226-s to 232-s. Grotke, G. E., Lessmann, G. G., and Math­ 34(4) :183-s to 196-s. 45. King, B. L. 1966. Weld metal and heat- ews, S. J. 1 982. The Varestraint test. Weld­ 26. Nippes, E. F., Savage, W. F., and affected zone cracking in wrought austenitic ing Research Council Bulletin No. 280. Grotke, G. 1 957. Further studies of the hot stainless steels. BWRA Report, C139/A/2/65. 9. Savage, W. F., Nippes, E. F., and Good­ ductility of high-temperature alloys. Welding The Welding institute. Cambridge, England. win, G. M.. 1 977. Effect of minor elements Research Council Bulletin No. 33. 46. Tamura, H„ and Watanabe, T. 1973. on hot cracking tendencies of Inconel 600. 27. Williams, C. S. 1963. Steel strength Mechanism of liquation cracking in the weld Welding Journal 56(8):245-s to 253-s. and ductility response to arc welding ther­ heat-affected zone of austenitic stainless 10. Senda, T., Matsuda, F., Takano, G., mal cycles. Welding Journal 42(1):l-s to 8-s. steels. Trans JWS 4(1)30-42. Watanabe, K., Kobayashi, T, and Matsuzaka, 28. Kreischer, C. H. 1963. A critical anal­ 47. Labtech Notebook software package. T. 1971. Fundamental investigations on so­ ysis of the weld heat-affected zone hot duc­ 1990. Laboratory Technologies Corp., Wilm­ lidification crack susceptibility for weld met­ tility test. Welding Journal 42(2):49-s to 59-s. ington, Mass. als with Transvarestraint test. Trans. JWS 29. Weiss, B., Grotke, G. E., and Stickler, 48. Demyantsevich, V. P. 1967. Mecha­ 2(2):1-22. R. 1970. Physical metallurgy of hot ductility nism of hot cracking during welding. Weld­ 11. Donati, J. R., and Zacharie, G. 1974. testing. Welding Journal 49(10):471 -s to 487-s. ing Production 14(3): 1-6. Evaluation of tendency toward hot crack in 30. Dahl, W., Duren, C, and Muesch, H. 49. Medovar, B. I. 1954. On the nature the welding heat-affected zone of austenitic 1973. Susceptibility of austenitic welded of weld hot cracking. Avtom. Svarka 18-10 stainless steels. Revue de Metallurgie joints to hot cracking. IIW Doc. 11-660-73. 7(4):12-28. Brutcher Translations No. 3400. 71(12):905-915. 31. Muesch, H. 1984. Welding of mate­ 50. Torpov, V. A. 1957. On the mecha­ 12. Ogawa, T., and Tsunetomi, E. 1982. rial TP 347 mod. Mannesmann Research In­ nism of hot cracking. Metallovedenie Hot cracking susceptibility of austenitic stain­ stitute. Duisburg, Germany. Obrabotka Metallov (6):54-58. Brutcher less steels. Welding journal 61 (3):82-s to 93-s. 32. Owczarski, W. A., Duvall, D. S., and Translations No. 3982. 13. Lundin, C. D., Lee, C. H., and Menon, Sullivan, C. P. 1966. A model for heat-af­ 51. Pumphrey, W.)., and Jennings, P. H. R. 1988. Hot ductility and weldability of free fected zone cracking in nickel base superal­ 1948. A consideration of the nature of brit­ machining austenitic stainless steel. Welding loys. We/d/'ngyourna/45(4):145-s to 155-s. tleness above the solidus in castings and yourna/67(10):119-sto 130-s. 33. Duvall, D. S., and Owczarski, W. A. welds on aluminum alloys. Journal of the In­ 14. Lundin, C. D., and Qiao, C. Y. P. 1967. Further heat-affected zone studies in stitute of Metals 75:235-256. 1991. Weldability of nuclear grade stainless heat resistant nickel alloys. Welding Journal 52. Pellini, W. S. 1 952. Strain theory of steels. Proceedings of the International Con­ 46(9):423-sto432-s. hot tearing. The Foundry 80(11 ):1 25-133. ference on New Advances in Welding and 34. Yeniscavich, W. 1966. A correlation 53. Apblett, W. R„ and Pellini, W. S. 1954. Allies Processes. Beijing, China. of Ni-Cr-Fe alloy weld metal fissuring with Factors which influence weld hot cracking. 15. Katoh, M., and Kerr, H. W. 1987. In­ hot ductility behavior. Welding Journal Welding Journal 33(2):83-s to 90-s. vestigation of heat-affected zone cracking of 45(8):334-s to 356-s. 54. Bishop, H. F., Ackerlind, C. E., and GTA welds of Al-Mg-Si alloys using the Vare­ 35. Yeniscavich, W. 1969. Correlation of Pellini, W. S. 1952. Metallurgy and mechan­ straint test. Welding Journal 66(12):360-s to hot ductility curves with cracking during ics of hot tearing. Trans. American Foundry- 368-s. welding. Proceedings of the Symposium on mans Society 60:818-833. 16. West, S. L., Baeslack, W. A. Ill, and Methods of High-Alloy Weldability Evalua­ 55. Borland, J. C, and Younger, R. N. Kelly, T. J. 1989. Morphology of weld heated- tion. 2-12. Welding Research Council. Troy, 1960. Some aspects of cracking in welded affected zone liquation cracking in Ta-modi- N.Y. Cr-Ni austenitic steels. British Welding Jour­ fied cast alloy 718. Metallography 23(3):219- 36. Arata, Y., Terai, K., Nagai, H., nal 6(1):9^16. 229. Shimizu, S., and Aota, T. 1977. Fundamen­ 56. Borland, j. C. 1961. Generalized the­ 17. Thompson, R. O, Cassimus, J. J., tal studies on electron beam welding of heat- ory of super-solidus cracking in weld and Mayo, D. E., and Dobbs, J. R. 1 985. The re­ resistant superalloys for nuclear plants - Re­ casting. British Welding Journal lationship between grain size and microfis­ port II. Trans JWRI 6(1 ):69-79. 7(81:508-512. suring in alloy 718. Welding Journal 64(4) :91- 37. Lin, W., Baeslack, W. A., Ul and Lip­ 57. Borland, J. C. 1979. Fundamentals of s to 96-s. pold, ). C. 1989. Hot ductility testing of high solidification cracking in welds. Welding and 1 8. Thompson, R. G. 1 988. Microfissur­ strength low expansion superalloys. Recent Metal Fabrication Parti, (1/2):19-29. Part II, ing of alloy 718 in the weld heat-affected trends in welding science and technology (3):99-107. zone. Journal of Metals 40(7):44-48. TWR '89. 609-614, ASM International, Ma­ 58. Matsuda, F., Nakagawa, H., and So- 19. Lippold, J.C., Baeslack, W. A. Ill, and terials Park, Ohio. rada, K. 1982. Dynamic observation of so­ Varol, I. 1992. Heat-affected zone liquation 38. Blum, B. S., and Witt, R. H. 1963. lidification and solidification cracking dur­ cracking in austenitic and duplex stainless Heat-affected zone cracking in A-286 weld­ ing welding with optical microscope. Trans. steels. Welding Journal 71(1):1-s to 14-s. ments. Welding Journal 42(8):365-s to 370-s. 7W/?l11(2):67-77. 20. Kujanpaa, V. P., David, S. A., and 39. Brooks, J. A., and Krenzer, R. W. 1974. 59. ASM Metal Handbook. Properties and White, C. L. 1987. Characterization of heat- Progress toward a more weldable A-286. selection: stainless steels, tool materials and affected zone cracking in austenitic stainless Welding Journal 53(6):242-s to 245-s. special-purpose metals. 9th Edition, Vol. 3, steel welds. Welding Journal 66(81:221-s to 40. Vagi, ). J., and Martin, D. C. 1956. ASM International, Materials Park, Ohio. 228-s. Welding of high-strength stainless steels for 60. Lin, W., Nelson, T. W., Lippold, ]. C, 21. Lippold, ). C. 1 983. An investigation elevated-temperature use. Welding Journal and Baeslack, W. A. III. 1992. A study of the of heat-affected zone hot cracking in alloy 35(3):137-s to 144-s. HAZ crack-susceptible region in alloy 625. 800. Welding Journal 62(1 ):1 -s to 11 -s. 41. Brooks, J. A. 1974. Effect of alloy mod­ to be published in Proceedings of the 3rd In­ 22. Ernst, S. C, Baeslack, W. A. Ill, and ifications on HAZ cracking of A-286 stain­ ternational Conference on Trends in Weld-

152-s I APRIL 1993 ing Research, Gatlinburg, Tenn. NST Nil-strength temperature HAZ region, surrounding the weld pool, 61. Nelson, T. W., Lin, W., and Lippold, PMZ Partially-melted zone within which the material is susceptible J. C. 1992. A study of the HAZ crack-suscep­ tc Cooling time over which to liquation cracking due to the local­ tible region in 6061 aluminum alloy, unpub­ cracking occurs in the ized loss of ductility. lished research. Edison Welding Institute. spot-Varestraint testing Ductility Recovery Temperature Columbus, Ohio. tc.NST lc at a point in the HAZ (DRT): Temperature on-cooling from a which experiences a peak peak temperature above the NDT at Appendix A temperature of NST which perceptible ductility (>5%) of the

L t at the fusion boundary material is apparent. Symbols and Abbreviations c,TL c TCL Total crack length Maximum Crack Length (MQJ: The T Temperature at the fusion maximum length of cracks on the as- BTR Brittle temperature range FR boundary after arc tested specimen surface in longitudinal- CHL Cracked heat-affected zone extinction or spot-Varestraint tests. Traditionally, length Liquidus temperature this crack length is measured from tip to CHL c On-cooling portion of the TL 0 Peak temperature of a tip of a crack. In this study, the MCL rep­ cracked HAZ Length thermal cycle resents the distance between the CHLQCNST CHL at a point in the OC Temperature at the crack isotherm at the crack tip and the HAZ which experiences a •up tip during weldability isotherm at the fusion boundary. peak temperature of NST testing Nil-Ductility Region (NDR): A region CHL CHL at the fusion OCTL oc V Travel speed during in the HAZ, surrounding the weld pool, boundary T longitudinal-Varestraint within which the ductility of the mate­ CHL On-heating portion of the OH testing rial, as determined with the hot-ductil­ cracked HAZ Length V Travel speed for actual ity test, is essentially zero. CR Average Cooling Rate w T welding conditions Nil-Ductility Temperature (NDT): during weldability testing Temperature on-heating at which the CR Average cooling rate for W ductility of the material drops to zero. actual welding conditions Appendix B Nil-Strength Temperature (NST): CSR Crack-susceptible region Definition of Terms Temperature on-heating at which the DRT Ductility recovery strength of the material drops to essen­ temperature Brittle Temperature Range (BTR): The tially zero. DRT DRT corresponding to a NST temperature range during weld cooling Saturated Strain: The applied aug­ peak temperature equal to within which the material is susceptible mented strain level above which the the NST to liquation cracking due to the local­ maximum crack length remains con­ DRT DRT corresponding to a a ized loss of grain boundary ductility. stant, or saturates, for the spot- and lon­ peak temperature equal to Cooling Time (t ) or Delay Time: The gitudinal-Varestraint tests. theT c L time period between arc extinction and Threshold Strain: The applied aug­ FB Fusion Boundary specimen bending for the spot-Vare­ mented strain above which cracking oc­ G Temperature gradient T straint test. curs in the longitudinal- or spot-Vare­ during weldability testing Cracked HAZ Length (CHL): The straint tests. G Temperature gradient for w length of the region in the HAZ, mea­ Total Crack Length (TCL): Cumula­ actual welding conditions sured parallel to the fusion boundary, tive length of all cracks on an as-tested HAZ Heat-affected zone over which cracking is observed in the specimen surface in spot- or longitudi­ MCL Maximum crack length longitudinal-Varestraint test. nal-Varestraint tests. NDR Nil-ductility region Crack-susceptible Region (CSR): The NDT Nil-ductility temperature

WELDING RESEARCH SUPPLEMENT I 153-s