Magnetic Force Upset Welding Dissimilar Thickness Stainless Steel Tee Joints

When evaluating resistance welded tee joints in Type 304 and 17-7PH steels, best results are obtained with sheet beveled at the joint interface and with the more ductile Type 304 plate

BY KENNETH H. HOLKO

ABSTRACT. Sheet-to-plate tee joints were However, most of the conventional welding current. After a few millisec­ resistance welded using the magnetic welding processes are not readily ap­ onds of interfacial heating, the force force upset welding (MFUW) process. plicable to making a tee joint between wave may be applied. MFUW con­ Both AISI Type 304 and 17-7PH tee sections of greatly different thickness. trols can be set for short weld current joints were evaluated. The best results Usually, the thinner sheet section is times (< 8 millisec) and high peak were obtained with sheet beveled at the joint interface and the more ductile Type heated more rapidly than the thicker currents. Short times at high current 304 plate. An unwelded area under the plate section. Since heating is not con­ densities cause rapid interfacial weld fillet caused a stress concentration centrated at the tee joint junction (in­ heating to minimize the effect of the that lowered the strength of 17-7PH terface), a poor quality (low strength) unequal heat dissipation. Materials weldments. Removal of the weld fillet or weld results. with different thermal conductivities the use of more ductile Type 304 plate Resistance welding lends itself to may be joined by this method.3 greatly improved weld strength. Welds in the sheet-to-plate tee joint problem MFUW may be used to make solid 17-7PH were as strong as the plate after since maximum heat is developed at state welds in a great variety of mate­ removal of the weld fillet. MFUW was 4 successful because of short heating times the interface where resistance is high. rials. and magnetic force delay relative to weld However, in the conventional mode of The purpose of this study was to current capability. several cycles (0.05 sec) or more of evaluate the potential of MFUW for current, there is sufficient time for joining stainless steel sheet and plate unequal heat dissipation to occur in a tee joint configuration with the Introduction away from the interface. And in­ sheet as the leg of the tee as illustrated Many aerospace components re­ sufficient interfacial heating results in in Fig. 1. Sheet thicknesses were Vic quire tee joints between dissimilar a weak weld.1' 2 in. (1.59 mm) and plate thicknesses thickness materials (e.g., compressor Magnetic force upset welding were ^/32 in. (18.2 mm). AISI and turbine blades to rotors and discs). (MFUW) differs from conventional Type 304 and 17-7PH* stainless steels resistance welding in that a forging were welded. 17-7PH was selected for weld force is applied by a 120 Hz KENNETH H. HOLKO is with the Na­ tional Aeronautics and Space Administra­ electromagnet. The force wave can be "Tradename of Armco Steel Corp. and ten­ tion, Lewis Research Center, Cleveland, electronically timed in relation to the tative AISI Type. Ohio.

Paper to be presented at the AWS National 1 Fall Meeting in Baltimore, Md., during Table 1—Room-Temperature Base Metal Tensile Properties of October 5-8, 1970. 17-7PH and AISI Type 304 Stainless Steels Elonga Ultimate 0.2 % tion i tensile yie Id 'A in. 1 strength strength (1.27 MN/ MN/ cm), Material Condition ksi m2 ksi m2 % Fracture 1 17-7PH plate (tested TH1100 (precipita- 144 990 2 Brittle in short transverse tion hardened) to roll'ng direction) 1 17-7PH sheet (tested TH1100 177 1220 149 1025 17 Ductile parallel to rolling H --««— direction) ^ 17-7PH sheet 9,h Solution anneal d 130 895 40 275 30 Ductile .•** Type 304 shee* (tested Mill annealed 94 645 Ductile parallel to rolling direction)

a Averaged values of three or more tests. Fig. 1—Typical welded tee joint b Data is for a 2-in, (5,1-cm) gage length.

WELDING RESEARCH SUPPLEMENT | 427-s 17 its high strength and Type 304 for its /32 in. (1.59 X 23.8 X 64.3 mm) of sheet to 17-7PH plate, although other good weldability and high ductility. 17-7PH and Type 304 stainless steel 23 combinations of the two materials Welding parameters evaluated in­ were welded to plate specimens /32 were tried. JAII contacting surfaces on cluded timing and magnitude of force X 2 X 3V4 in. (18.2 X 50.8 X 82.6 both types of specimens were ma­ and current, stickout of the sheet mm) of the same materials. chined to a 16 rms (40.7 X 10-18 cm from the electrode tooling, and the The nominal chemical compositions rms) or better finish. amount of upset at the joint. Varia­ (%) of the two materials are: All the 17-7PH material was weld­ tions in tooling and joint configuration 17-7PH Type 304 ed in the solution-annealed condition were also studied. Weld integrity was C 0.07 0.06 and the Type 304 was in the mill- determined by metallographic sections Mn 0.70 1.50 annealed condition. All the 17-7PH to and tensile tests. Si 0.40 0.50 17-7PH welds and some 17-7PH to Cr 17.0 | 19.0 Type 304 welds were heat treated to Materials and Procedure Ni 7.01 10.0 the TH1100 condition (1400° F Al 1.15 — (760° C) /1V hr./A.C. -f 1100° F Materials and Heat Treatment Fe Bal Bal 2 15 (593° C) /1V2 hr./A. C). Tensile Sheet specimens V1G X /i6 X 2 Most of the velds were 17-7PH properties for the 17-7PH sheet and plate (short transverse) in the Current and force TH1100 condition are shown in Table 1. Also shown are tensile strengths for Upper platen annealed 17-7PH sheet and Type 304 (copper) sheet. Note the low strength and duc­ tility of the 17-7PH plate in the short transverse direction as compared to those of 17-7PH sheet in the rolling Z3 ^Electrode tooling direction. (copper) 17-7 PH space Welding Procedure

Sheet specimen, 1/16(1.6) Joint Configuration. The electrode tooling and joint configurations are illustrated in Fig. 2. Two types of Stickout sheet specimens were used. The un- ___——Weld joint beveled sheet is shown in place in the (I2R heating) tooling in Fig. 2(a). Beveled sheets Plate specimen were also used as shown in Fig. 2(b), 23/32(18.3)-^ with various initial contact thicknesses and angles of bevel. This type was used to concentrate current and elimi­ nate undesirable effects to be de­ Lower platen scribed later. copper) Force and current are applied as / (a) Electrode tooling with un- shown in Fig. 2(a). Welding occurs beveled sheet in place. as I2R (I—current, R—resistance) resistance heating takes place at the interface between the sheet and plate. _0.062__ Fillets are formed at the sheet-plate / (1.6) interface as the sheet is upset. / a Electrode Tooling Configuration. / L The tooling used to hold the sheet specimen and introduce current and I force is shown in Fig. 2. The tooling / (b) Sheet specimen bevels. Bevel angle, material is RWMA Class II copper. A f a= 15°, 30°, 45°; initial contact thickness, typical sheet specimen is shown in place. The stickout of the sheet from 4___ I = 0.010 to 0.030 inch (0.25 to 0.76 mm). UI28.6) the tooling is shown in Figs. 2(a) and 2(c), and is discussed in a later sec­ tion. Mo- The most frequently used tooling i: Ll/4 (6.4) had a 45 deg chamfer with a 0.030 in. StickouUUtl (.76 mm) leg (see X dimension in l/8orl/4_ Fig. 2(c) ). This chamfer is used to (3.2 or 6.4) 0.030 form the fillet at the junction of the (0.76) thin and thick members. Variations of (c-l) Straight chamfer: (c-2) Radiused (c-3) Molybdenum inserts this chamfer are shown in Fig. 2(c) standard, X = 0.030 chamfer, (shown at right angle and include a smaller chamfer, a radi­ in. (0.76 mm); small, R= 0.030 in. to other views). used chamfer, and the use of a Mo X= 0.015 in. (0.38 mm). (0.76 mm). insert for part of the chamfer. Magnetic Force Upset Welding (c) Various tooling configurations. (MFUW). MFUW differs from con­ Fig. 2—Joint configurations and electrode tooling variations for magnetic ventional resistance welding in that force upset welding (MFUW) sheet-to-plate tee joints the forge force is applied by a 120 Hz

428-s | SEPTEMBER 1970 electromagnet—Fig. 3. The advantage is that the force half-waves can be Air cylinder timed in duration and phase shifted in relation to the current half-waves. The Y////////////A heating may be more effectively de­ veloped and concentrated at the inter­ Drive shaft- face in this manner. By delaying the Magnet armature initiation of the force half-wave until (connected to after the initiation of the current half- drive shafts wave, the current can flow through an interface under low pressure, and Magnet stator thus, high resistance. So, the R in the ^.^Magnet air gap (not connected PR term may be maximized during ~p'" (adjustable) to drive shaftH current flow, thus maximizing the resistance heating at the interface.6- 6 v\ This is essential if unequal sections are to be welded. MFUW is used with high welding 6 Magnet current densities (typically 1.5 X 10 transformer A/in.2 (0.233 X 106 A/cm2) and short current times (< 16 millisec). This convention follows from the use Upper platen of MFUW for solid state welding where melting is avoided. The short —\— Weld voltage times and high current densities may Electrode tooling and specimens control also be used to counteract the heat (current) balance problem.2 The rate of heat input into the interface may be ad­ justed to be much greater than the Weld rate of heat dissipation away from transformer the interface. This makes the dissimi­ lar heat dissipation rates of the two Fig. 3—Simplified diagram of magnetic force upset welding (MFUW) sections less influential in determining equipment. The methods of applying pneumatic force, magnetic force, the location of maximum heating. The and weld current are shown current and force capabilities for the MFUW machine used in this study are less than 4 and 6 cycles of force. tron setting and magnet gap then, the shown in Table 2. Wave initiation, duration, and mag­ time relation between force initiation MFUW Force—Current Relation­ nitude are dependent upon conven­ and current initiation may be varied ships. For most of the work reported tional ignitron "heat" settings. In the while force magnitude is held relative­ here, less than one full cycle of 60 Hz case of the magnetic force wave, how­ ly constant. current and less than 2 full cycles of ever, magnet gap may also be adjusted There is an inherent time delay of 120 Hz magnet force were used. This to vary the force magnitude at one the force wave of about 0.5 millisec consisted of a small preheat half-cycle ignitron setting. By adjusting the igni- due to the mechanical inertia of the and a larger welding half-cycle as shown in Fig. 4. The preheat half- cvcle provides both macro and micro Weld magnet force- alignment between members, incipient welding, and an increase in interfacial temperature. The welding half-cycle Preheat time provides the bulk of heating and upset delay (from necessary in the formation of the mechanical inertia) weld. A few welds were made with less than 2 and 3 cycles of current and Pneumatic

force-v Table 2—Maanetic Force Upset Welding Machine Capabilities" Welding current: 200 ka at short circuit Single-phase, 60 Hz, 440 v, 400 kva transformer Force: Magnet force, 6.5 kips (27.9 N) peak Preheat current-7 Pneumatic force, 3.7 kips (16.5 N) Welding time delay maximum Single-phase, 60 Hz, 440 v, 100 kva 'ransformer producing 120 Hz force Time, t wave Fig. 4—Typical current and force waves used in magnetic force upset * Current and force waves independently welding (MFUW) for less than one cycle of 60 Hertz current and two controlled by phase shift heat controls. cycles of 120 Hertz force

WELDING RESEARCH SUPPLEMENT I 429 s welding head (or movable platen). cording instrument (Visicorder). The other two types of tensile spec­ imens were designed to test the entire This is the time required to transmit Weld Evaluation the force from the magnet to the tee joint (termed "full size tensile workpieces. Thus, there will actually Tensile Tests. The welded tee joints specimens"). One had a stress concen­ be a slight force delay when the force were sectioned in various ways to tration present at the sheet-plate in­ and current ignitron settings are the provide several types of tensile and terface under the weld fillet—Fig. same. metallographic specimens. Three types 5(b). The other had the weld fillet of tensile specimens were used. To delay the force wave even more hand filed to remove the stress con­ The sheet-type specimen (termed centration—Fig. 5(c). Both types of behind the current wave, the ignitron "small tensile specimen"), shown in setting is decreased (e.g., 50 to 40% specimens were tested with conven­ Fig. 5(a), was machined from the tional serrated tensile grips and a "heat" setting) while keeping the cur­ sheet and planes parallel to the sheet rent setting constant. A constant threaded pull rod that fit in the tapped surfaces through the plate. A tab from hole in the plates as shown in Fig. force magnitude is retained by de­ the unaffected sheet material was creasing the magnet gap which effec­ 5(b). An accurate weld area could welded on to the plate side of the not be measured for the full size tively increases peak force back to the small specimen to provide increased original value. This is necessary since tensile specimens. So the tee joints length for testing purposes. The gage were compared by taking the failure a decrease in ignitron setting also section was V in. (12.7 mm) long decreases peak force. 2 loads, dividing by the sheet cross- with a 0.188 X 0.060 in. (4.77 X 1- sectional areas, and expressing the re­ Peak welding current was measured 53 mm) cross section. Tensile testing sultant sheet stresses at weld failure as with a Duffers' current analyzer. was done in air at room temperature percentages of the sheet ultimate ten­ Waveshapes of primary and secondary with a strain rate of 0.1 in. in./min. sile strengths. It should be pointed out current, magnetic force, and primary Only specimens of 17-7PH sheet to that the sheet ultimate tensile strength and secondary voltage were recorded 17-7PH plate heat treated to TH1100 for the particular sheet material used on an ultraviolet light sensitive re­ were tested in this configuration. in each tee joint was used to calculate this percentage (i. e., Type 304, Tensile testing 17-7PH, and 17-7PH in the TH 1100). direction Metallographic Evaluation. Metal­ lographic specimens were machined from some of the specimens at two sections transverse to the weld joint at ^Full-size tensile the center and edge as indicated by A / specimen and B in Fig. 5(a). They were mounted, polished, and etched with oxalic acid. An average metallograph­ ic area (A,„) was measured for each Small tensile specimen—Fig. 6). From many ob­ specimen servations, the Ara was found to ap­ proximate the fusion weld area on a plane through the original interface. The large arrows shown in Fig. 6 point to the original interface of the „,—Metallographic tee joint. Area Am was obtained by sections averaging the straight line weld width (along the original interface) and multiplying it times the length of the weld.* The average depth of maximum penetration (Dm) of the fusion weld below the original interface was ob­ tained by averaging the maximum penetrations for the two sections. Also, an interface-to-center-of-upset (I to C) distance was measured for the unbeveled joints—Fig. 6(a). These three dependent variables are indica­ tors of the extent of interfacial heating. They are used to determine the effects of variations in weld set­ tings. Welding Parameters. The decrease in length (A L) of the sheet was measured for each welded specimen and used as a measure of the upset. Essentially, all of the upset takes place in the material sticking out of thi Fig. 5—Location and types of tensile and metallographic specimens ma­ tooling since it is heated and not fully chined from tee joint. Top (a)—location of small tensile specimen and metal­ lographic sections; left (b)—full size tensile testing with serrated grips and •Am is actually a relative, rather than tapped hole; right (c)—full-size tensile specimen with weld fillet filed to absolute, weld area; therefore, area units remove stress concentration are not used.

430-s ! SEPTEMBER 1970 (a) Center section. Etched; X36. lb) Enlargement of block in (a). Etched; X250.

(cl Edge section. Etched; X36. Fig. 6—Typical magnetic force upset sheet-to-plate weld in 17-7PH (weld made with < 1 cycle of 60 Hz current) and un- beveled joint configuration (specimen 18-25). Section locations shown in Fig. 5. Etchant: oxalic acid. Parameters used to indicate amount of interfacial heating are shown: Average metallographic area, A,„ = tWi -f W, j- W3)/2] L. Average Dx+ D, depth of maximum penetration, D„, = .

constrained. study to evaluate the effects of weld­ best combinations of welding parame­ Stickout (S) is measured from the ing variables on joint properties. The ters found, based on tensile strengths last point of contact with the tooling to the end of the sheet, as shown in Fig. 2, for a 0.030 inch (.76 mm)/45 Table 3—Optimum Weld Parameters for Sheet-to-Plate Tee Joints in deg chamfer. However, in order to 17-7PH and AISI Type 304 Stainless Steels (Less Than One Cycle of compare variations of this chamfer 60 Hz Current) (Fig. 2(c)), an "effective" stickout was Parameter Unbeveled Beveled used. For the 0.015 in. (.38 mm) 45 Material (sheet/plate) 17-7PH/17-7PH 17-7PH/17-7PH 17-7PH/Type 304 deg chamfer, 0.0078 in. (.20 mm) or Bevel, in. (mm)/deg 0.015 (0.38)/15 V128 in. was added to the measured Tooling chamfer, 0.015 (0.38)/45 or 0.030 (0.76)/45 stickout. And for the 0.030 in. (.76 in. (mm)/deg 0.030 (0.76)/45 mm) radiused tooling, 0.0039 in. (.10 Stickout, S, in. (mm) 13/128 to 14/128 13/128 to 14/128 (2.6 to 2.8) mm) was subtracted. These correc­ (2.6 to 2.8) tions follow from heat balance consid­ Current (peak)," kA 45.5/60 34/50 35/51 erations. Magnetic force (peak) « 3.45/3.7 1.5/4.0 (6.65/17.8) .5/3.8(6.65/16.9) Delay (D) is defined as the time kips (kN) (15.3/16.4 Delay," D, msec 0.5/1.5 1.5/1.0 delay of the initiation of the force Current duration," msec 4.7/6.0 4.6/5.7 4.6/5.9 wave after the beginning of the weld­ Force duration," msec 5.7/5.9 4.9/6.2 4.9/6.1 ing current wave. It was estimated Decrease in length, AL, 0.060 to 0.067 0.062 to 0.063 (1.57 to 1.6) from the Visicorder traces. in. (mm) (1.5 to 1.7) Ratio of decrease in length 0.92 to 0.96 Results and Discussion to stickout, AL/S12

Over 150 welds were made in this Preheat-weld—Fig. 4.

WELDING RESEARCH SUPPLEMENT I 431-s and metallographic observations, are unsatisfactory. for the scatter in the data shown. shown in Table 3. Only parameters Selected specimens were tensile Mechanical Properties for short-time welds made with less tested and metallographically exam­ than one cycle of 60 Hz current and ined. The tensile test results are shown Unbeveled Joints. The tensile re­ less than two cycles of 120 Hz magnet in Table 4 for unbeveled joints and in sults are shown in Table 4 for unbev­ force are shown. Longer time welds Table 5 for beveled joints. Variations eled joints. The best weld settings of were tried, but were judged to be in welding parameters are responsible those used are shown in Table 3. For the 17-7PH/17-7PH (TH1100) joints tested full size (stress concentration present), the best weld strength ob­ Table 4—Tensile Test Results for Unbeveled Tee Joints in tained was 63% of the sheet (i.e., 17-7PH and AISI Type 304 Stainless Steels when the weld failed, the sheet was stressed to 63% ultimate tensile (a) Full-size tensile specimens strength). Sheet stress has been used Sheet since the weld area could not be ulti­ mate accurately measured. All of the welds tensile were irregularly shaped combinations — Tooling Heat strength of fusion and solid state welds. Speci­ Chamfer •— Material — treat­ at fail­ For the full size specimens, the 1 men Material in. mm Sheet Plate ment" ure ', % Fracture mode stress concentration present at the B9-1 Copper 0.030 0.76 17-7PH 17-7PH TH1100 33.6 1] Solid-state junction of the sheet and plate under B9-2 46.0 Iweld and plate the weld fillet (see Fig. 6) caused C9-1 63.0 [heat-affected failure at low tensile strengths. Frac­ jzone ture, initiated at the stress concentra­ C9-3 53.8 Mostly solid- tion, easily propagated through the state weld 17-7PH plate which has low ductility E8-9 .015 .38 Type 304 As welded 83.4 Through sheet at fillet in the short transverse direction— E8-10 .015 .'8 37.8 1 Table 1. Type 304 sheet has only about half the strength of 17-7PH E8-12 .015 .38 37.' I Mostly solid- E8-16 .030 .76 40.5 | state weld sheet (in the TH1100 condition). Al­ E8-17 50.0 though tensile data are not available E8-18 73.3 )Tensile hole for Type 304 plate in the short trans­ E8-19 83.8 }in plate verse direction, it is generally accepted E8-20 75.0 j initiated that the Type 304 plate is both weak­ 41.5 Through sheet E8-21 er and more ductile than the 17-7PH J9-9 Type 304 93. 'Solid-state J9-10 92.0 Iweld and plate plate in the short transverse direction. J9-11 85.7 [heat-affected However, joints with Type 304 plates [zone were as strong as those with 17-7PH J9-12 100.0 1 plates (see specimens C9-1 and 18-12, J9-13 Through sheet Table 4). The reason for this is felt to J9-14 be the higher ductility of Type 304 J9-15 I plate in the short transverse direction 18-12 Mo inserts .. 17-7PH TH1100 63.0 Mostly0 solid- state weld and the resistance to crack propa­ gation from the stress concentration it offers. Further evidence of the (b) Small tensile specimens in 17-7PH, heat treated to TH1100 beneficial effect of the Type 304 duc­ Elonga- tility is given by the 100% ultimate Ultima'e 0.2% tion in tensile strengths values for the Type tensile yield Vs 'n. 304—Type 304 joints in Table 4. Tnnl inet strength strength (1.27 1 UUIII Ig Speci Chamfer MN/ MN/ cm), When the stress concentration was men Material in. mm ksi m2 ksi m2 % Fracture mode removed by machining the small ten­ sile specimen in 17-7PH/17-7PH 18-24 Coppe r 0.015 0.38 144 992 134 926 4 Two-thirds solid-state weld, one-third heat joints, base metal strength was at­ affected zone tained. 18-25 143 985 135 930 6 For the small size tensile specimens 18-26 111 765 — — 2 ! Mostly solid-state of 17-7PH, ultimate tensile strengths 18- 7 67 464 — — [weld over 140 ksi (964 MN/m2) with yield 18-28 i 85 586 — — strengths of 134 ksi (920 MN/m2) 18-30 i 129 890 124 857 Through plate d d and 4-6% elongation were achieved L8-4 .030 .76 96 663 — — \ Mostly solid-state (see specimens 18-21, 24, and 25 in L8-5 86 595 — — • (weld L8-6 23 157 — — 0 tol \All solid-state weld Table 4). These weld strengths are L8-11- 83 572 — — 1 i good when compared to the aver­ 18-21 Mo in ;ertsr .030 .76 142 975 135 926 4 Mostly plate heat- age plate strength of 144 ksi (990 2 affected zone MN/m ) shown in Table 1. Weld ductility is also higher than the plate a All materials were welded in annealed condition. b 2 average. The higher sheet strength of Based on an average sheet ultimate tensile strength of 177 ksi (1220 MN/m ) for 17-7PH sheet in the TH1100, 130 ksi (895 MN/m') for 17-7PH sheet annealed, and 94 ksi (645 MN/m2) 17-7PH could not be attained proba­ for Type 304 sheet. bly because weld strength was limited c '/s in. (3.18-mm) Mo taper (see Fig. 2(c)). d Radius. by the relatively weak plate. e Weld made with less than three, rather than less than one cycle of60 Hz current; and less Beveled Joints. Tensile tests on full- than six, rather than less than two cycles of 120 Hz magnet force. 1 '/< in. (6.36-mm) Mo taper (see Fig. 2(c)). size specimens produced from beveled

432-s I SEPTEMBER 1970 joints indicated that removal of the concentrations are present. Hot sheet this point. Hardly any current enters stress concentration under the fillet material at the outside has come into the 17-7PH sheet along the rest of its resulted in higher joint strengths. For contact with the cold plate surface length because of its high resistance as example, as high as 86% sheet ulti­ and the resultant stress concentration compared to the copper tooling. mate tensile strength was reached for is similar to a "cold shut" sometimes The reason for the "dumbbell- 17-7PH/17-7PH beveled joints tensile found in castings and forgings. shaped" appearance of heated materi­ tested full size with the stress concen­ In both sections shown in Fig. 6, al in the center section (Fig. 6(a) ) is tration removed (see specimens 19-5, localized heating has occurred just the occurrence of the "skin effect." 6 in Table 5). This is an increase of above the fillet, actually at the last This effect is normally associated with about 25 percentage points over bev­ point of contact between the sheet higher frequency current as common­ eled joints tested with the stress con­ and tooling, before upset. The reason ly used in induction heating. However, centration present (specimens J9-3, for this is that most of the current the occurrence of this effect at 60 Hz 4). Welding parameters used are enters the sheet from the tooling at has been described.7 The rapid rise of shown in Table 3. By using the ductile Type 304 plate, slightly higher strength welds (88% sheet ultimate Table 5—Tensile Test Results for Beveled Tee Joints in tensile strength) were obtained even 17-7PH and AISI Type 304 Stainless Steels with the stress concentration present (a) Full-size tensile specimens; tooling material, copper; chamfer, 0.030 in. (0.76 mm) (specimens J9-7, 8). Again, the ad­ Sheet vantage of the more ductile Type 304 ulti­ plate is seen. If the stress concentra­ —Sheet bevel—. mate tion is not removed, the beveled 17- Initial tensile 7PH/17-7PH joints are no stronger contact Heat strength than the unbeveled joints, as can be Speci­ — Material — thickness Angle, treat- atfail- a b seen by comparing Tables 4 (Speci­ men Sheet Plate in. mm deg ment ure , % Fracture mode mens C9-1 and C9-3) and 5 (Speci­ J9-3 17-7PH 17-7PH 0.015 0.38 15 TH1100 57.6 mens J9-3 and J9-4). J9-4 63.9 Tests on the small tensile specimens J9-5 85.6" Mostly plate heat- indicated ultimate tensile strengths as J9-6 i 84.2" affected zone J9-7 high as 160 ksi (1,100 MN/m2) with Type 304 45 88.4 J9-8 Type 304 i 87.0 1 i yield strengths over 143 ksi (985 C9-6' 17-7PH .020 21.2 All solid-state weld 2 .51 MN/m ) were achieved with the bev­ C9-7'i .020 19.0 .51 eled 17-7PH/17-7PH joints (see Ta­ C9-10 .025 30 51.7 .64 ble 5). However, actual weld strengths C9-11 .030 30 44.6 Half solid-state .76 C9-12 .020 15 43.3 weld, half heat- were higher than this in some cases .51 C9-14 .020 45 62.2 affected zone since failure occurred in the plate. The .51 D9-1 .010 45 42.1 relatively poor plate properties then .25 D9-2e Knife edge 45 24.6 Through cracked limit the theoretical joint strength I I I fusion weld (i.e., that of the 17-7PH sheet). J9-1 Type 304 Type 304 0.015 0.38 15 As welded 58.0 Solid-state and plate Although high strengths were heat-affected zone achieved with multiple cycle beveled J9-2 Type 304 Type 304 .015 .28 15 As welded 99.4 Mostly plate heat- joints (Table 5(b)), they are not rec­ affected zone ommended. The marginal solid state welds formed had poor resistance to (b) Small tensile specimens in 17-7PH, heat treated to TH1100 crack propagation, as shown by the Elonga - low strength of the full size tensile —Sheet bevel —. Ultimate 0.2% tion in specimens (Table 5(a)). Initial tensile yield Vs in. . Tooling contact strength strength (1.27 Metallography Speci- Mate- Chamfer thickness Ang le, MN/ MN/ cm), Fracture 2 2 Unbeveled Joints. The unbeveled men rial in. mm in. mm de g ksi m ksi m % mode sheet-to-plate tee joint consisted of L8-1 Cop 3er «0.030 *0.76 0.015 0.38 45 124 855 — — 1 1 Through both solid state and fusion welds L8-2 2.030 £.76 .015 .38 45 122 840 — — 0 i plate (principally grain boundary melting) C9-8 .030 .76 .015 .38 30 56 383 — — 2 ] Mostly for the settings shown in Table 3. C9-9 .025 .64 30 75 517 — — 1 ^solid-state Transverse metallographic sections j weld from the center and edge of a 17- C9-17 .015 .38 15 145 995 0 Through 7PH joint are shown in Fig. 6. The plate C9-5 .015 .38 45 — — 0 Mostly solid- large arrows indicate the location of state weld the original interface. As can be seen C9-15d .020 .51 161 1105 143 985 0 } in the center section (Fig. 6(a) and C9-16Through (b)), fusion welding occurred at the 18-22 Mo in­ .032 .81 152 1048 118 812 6 ! plate outside and incipient solid state weld­ serts' ing in the center. In the edge section 18-23 Mo in­ . .023 .58 137 940 126 870 4 Half plate (Fig. 6(c)), only fusion welding oc­ serts' heat-affected curred. zone, half plate The fillets formed at the base of the sheet section (Fig. 6) are the result a All materials welded in annealed condition. of the chamfer in the electrode >' See footnote (b), Table 4. r Stress concentration removed (see Fig. 8(c)). tooling. However, notice that the d See footnote (e), Table 4. sheet and plate are not completely •' Weld made with less than two cycles of current and four cycles of magnet force. ' '/«in. (6.36-mm) Mo taper (see Fig. 8(c)). welded under the fillets and stress e Radius.

WELDING RESEARCH SUPPLEMENT I 433-S current (I) with respect to time (t) weld settings (see Table 3), a fusion plate during upsetting if the angle of dl/dt used in this program also favors weld was obtained for the entire sheet bevel is small. Localized heating at the the skin effect. Briefly, the skin effect width in the beveled joint. The skin last point of contact is only present in is the tendency for current passing effect was avoided since the bevel the center section. through a conductor to concentrate in constricted the current. This caused a Influence of Welding Variables on a tubular element near the surface. high current density across the initial Joint Properties The current density is then highest just contact thickness. However, if the ini­ below the surface and decreases tial contact thickness was too large, Extensive analysis was made to de­ towards the center of the conductor. the skin effect still occurred. Typical termine the effects of the welding The higher current density results in transverse center and edge sections variables on the strengths of the melting near the outside surface. The with fusion welding are shown in Fig. joints. The results of this analysis lower density results in various de­ 8. Figure 8(b) shows the typical grain (summarized in Figs. 10-15) indicate grees of solid state welding at the boundary melting that occurred. The the relative importance of each weld­ center—Fig. 6(b). The skin effect is weld is similar for both sections, but ing variable studied in this program. not normally seen in conventional the penetration is greater at the edge For unbeveled joints stickout (5), de­ resistance welding because multiple (large arrows in Fig. 8 indicate origi­ creases in sheet length (l\L), the cycles of current are used. In this nal interface location). The reason for melting near the outside surface, the study, welds made with three cycles of this is the "edge effect" as described in ratio (/\L/S), and delay (D) were current did not have the skin effect. the Appendix. all important in obtaining highest strengths. For beveled joints only A L The shape of the heated material in Very little welding is present under and delay were important. the weld joint is shown schematically the fillets in either section shown in by the three sections in Fig. 7. The Fig. 8. This situation can be improved Slight variations occur in the weld section drawings have been construct­ by decreasing the bevel angle which parameters (of the type shown in ed from many actual metallographic increases the weld width as shown in Table 3) that are considered constant sections, and the reasons for their Fig. 9. The sheet in Fig. 8 had a 45 for each point plotted in Figs. 10-15. shapes will be discussed in the next deg bevel and the one in Fig. 9 a 15 But, from observation of many speci­ section. deg angle. In effect, more sheet mate­ mens, it is felt that these variations do not significantly influence the relation Beveled Joints. With the proper rial can come into contact with the between the parameters shown in each figure. vSolid-state weld Stickout (S). Figure 10 indicates r Incipient solid-state weld the effect of stickout on the strength Fusion weld t \ of unbeveled joints tested in both the full and small size tensile configura­ tions. A stickout range of about 12 14 in ?>>»/?}/s/s/>y;7;/7>/y?7s /i28- /i28 - (2-4-2.7 mm) yielded Section A-A: view of plane the highest sheet stresses at failure through original interface. for unbeveled joints. This range pro­ vided the best heat balance at the sheet plate interface. Larger stickouts resulted in: 1. Too much sheet heating away from the interface. 2. The center of upset moving well away from the interface—Fig. 6(a). Small stickouts did not allow suffi­ cient heat to be developed at the inter­ face because the closeness of the cop­ per tooling dissipated the heat too rapidly. In either case, little fusion welding and a weak incipient solid state weld resulted. Removal of the stress concentration, by machining the small tensile specimen, resulted in about a 35 ksi (241 MN/m2) strength increase. This is seen by comparing the two curves in Fig. 10. For the unbeveled joints, decreases in sheet length (A^) varying directly with stickout as shown in Fig. 11 if welding current, force, and delay are held constant. Since more sheet sticks out from the tooling, a greater materi­ al volume is heated, softened, and upset. This follows since the bulk of the sheet sticking out is resistance Section B-B: Just inside sheet Section C-C: Middle of sheet heated as well as the portion near the specimen surface (heated specimen thickness, interface. material crosshatched). Beveled joint strength was not as Fig. 7—Various sections through unbeveled welded joint showing geome­ dependent upon stickout as that of try of heated material unbeveled joints. The reason is that

434-s SEPTEMBER 1970 • -J

**(.- '

(W Enlargement of block in (al. Etched; X250.

(c) Edge section. Etcned: X36. Fig. 8—Typical magnetic force upset sheet-to-plate weld in 17-7PH (made with less than 1 cycle of 60 Hz current) and beveled (0.020 in. (0.51 mm)/45 deg) joint configuration (specimen H8-12). Etchant: oxalic acid

81

3sii'W ' •-: l '• fig

w. " / ' ' /: ' ' '' -,""-'<"'' "' ' '

• Stress B concentration B

MJL ^^^^^^^^m *^|S^5W 1••* _J^H ' "• -THL-..-' C-70-1389 (a) Center section. Etched; X36. (b) Edge section. Etched; X36. Fig. 9—Typical magnetic force upset sheet-to-plate weld (made with less than 1 cycle of 60 Hz current) and beveled (0.015 in (0.38 mm/15 deg) joint configuration (specimen C9-17). Etchant: oxalic acid heating is concentrated at the inter­ The 13/i28-14/i28 in. (2.6-2.7 mm) Decrease in Sheet Length. (AL). face by the sheet bevel. The bevel stickout range is recommended for As upsetting occurs, the center of constricts the current causing a high both unbeveled and beveled joints in upset in the sheet moves towards the density, and the interface is locally Table 3 because successful welds were interface. This causes increased inter­ heated. made at these settings. facial heating and more fusion and

WELDING RESEARCH SUPPLEMENT I 435-s O Sheet stress at weld failure (full-size tensiles) A Weld stress (small tensiles) \ Failure away from weld and weld-heat-affected zone 160 1000 17-7 PH plate UTS

800 120

60 CD .— OO 600 l/"> QJ c — CD ist *^ C £ 3 TOCD CD — 40 E CD 400 Wl CD

40 ^E 20 200 "5

10 12 14 16 18 Stickout, S, 1/128 in.

1. 2.2 2.6 3.0 3.4 Stickout, S, mm Fig. 10—Effect of stickout on joint strength of unbeveled 17-7PH/Type 304 and 17-7PH/17-7PH tee joints. solid state welding. For unbeveled For unbeveled joints and the stick­ For the beveled joints, interfacial joints, correlation of interfacial outs recommended in Table 3, at least heating varied directly with AL as heating and strength with AL was 0.060 in. (1.5 mm) AL was required shown in Fig. 12. Both weld strength only found when stickout remained for the highest 17-7PH/17-7PH joint (small tensile) and interfacial heating constant. Variations in stickout strengths. The 17-7PH sheet/Type (the product—(Am X Dm) ) in­ changed the heat balance enough to 304 plate joints required at least 0.063 creased indicating increasing interfa­ override the effect of AL. So, by in. (1.6 mm) AL and the Type cial heating. The product (Am X Dm) itself, £\L is not a reliable indication 304-Type 304 joints required 0.057 and the tensile strength are directly of interfacial heating for unbeveled in. (1.4 mm) (not shown in Table related but are not shown here. Varia­ joints. 3). tions in stickout did not interfere with the relation because of the beveled joint's insensitivity to stickout. A good 2. Or— relation between AL and sheet stress for the full-size specimen was not found because of the variable influ­ ence of stress concentration—Figs. 8 1.6 .060 and 9. Extremely small stickouts (12/i28 in. (2.4 mm) or less) required at least 2 — 0.052 in. (1.3 mm) AL to avoid a "ring weld" due to the skin effect. For .040 — larger stickouts ((14/i2s in- (2-8 mm) and up), AL should be between 0.046 and 0.063 in. (1.2 and 1.6 mm) to reach maximum strength and avoid cracking. Less severe bevels (.015 in. .020 (.38 mm)/15°) can tolerate more 10 12 14 16 upset than those shown in Fig. 12 Stickout, S, 1/128 in. (.015 in. (.38 mm)/45°) without cracking. This is because a greater volume of sheet material is available 1.6 2.0 2.4 3.2 to compensate for expulsion during Stickout, S, mm upsetting. Fig. 11—Effect of sheet stickout from electrode tooling on decrease in The joints made with the Mo insert sheet length. Unbeveled sheet-to-plate tee joints in 17-7PH tooling cracked at, lower values, of

436-5 ! SEPTEMBER 1970 (A„, X Dm) than the joints made the weld interface was not adequately delay), little interfacial heating oc­ with solid Cu tooling—Fig. 12. More heated. curred. For example, one weld with a heat is built up in the sheet near the When the force half-wave preceded negative delay of 1.0 millisec had an D fnr its interface because of the lower heat the current half-wave (negative exceptionally low (Am X m) conductivity of the Mo. AL/S1-2 ratio. Upset to Stickout Ratio (AL/5). A Weld ultimate tensile strength Both AL and S determine the amount • Interfacial heating of heating at the interface for unbev­ Solid symbols denote weld made with eled joints. For larger S's, more AL is Mo insert tooling required to bring the center of upset Failure away from joint (Fig. 6(a) ) to the interface. So when S varies, it is the ratio of AL to S t that determines how much interfacial "3 heating occurs. (AL is measured af­ 1000 r— ter the weld is made but can be indirectly controlled by selection of weld settings.) 800 — 12.0x10" Actually, S is slightly more impor­ tant in the ratio, AL/5. So a relation of AL/51-2 was empirically derived from these data. In Fig. 13, as the 600 E ratio, AL/51-2, increases, tensile < strength and the product (Am X Dm) do also. These dependent variables are 400 both indications of interfacial heating. And too much heating (AL/51-2/— 0.975) results in cracking. Too little heating (AL/51-2 < 0.9) results in a 200 weak weld. The I-to-C distance (see Fig. 6(a) ) is an indication of the amount of interfacial heating. As the center of . 040 . 050 .060 Decrease in length, AL, in. upset moves closer to the interface, more heating and welding occur. For small I-to-C's, high weld strengths re­ 1 5 sulted. The empirical ratio AL/5 - 1.0 1.2 1.4 1.6 determines the location of the center Decrease in length, AL, mm of upset. As previously described, Fig. 12—Effect of decrease in sheet length on weld ultimate tensile strength when this ratio increased, so did the and interfacial heating in beveled 17-7PH joints interfacial heating. Thus, the center of upset moved closer to the interface, as shown in Fig. 14. Again, too much A Weld ultimate tensile strength heating resulted in cracking. • Interfacial heating Solid symbol denotes weld made with Since the sheet bevel overrides the Mo insert tooling influence of 5, (within reasonable lim­ its) the ratio AL/5 does not correlate t Failure away from joint with interfacial heating for beveled 160 r— 16.0x10" joints. Delay (D). Interfacial heating in­ 1000 creases as the delay of the force half- wave behind the current half-wave increases—Fig. 4. This effect is shown 12.0 in Fig. 15 in terms of fusion weld -5 800 fracture area and sheet stress at fail­ ure for various unbeveled joints. The fracture area is the approximate area 600— = of melted material pulled out of the -0 ••= plate in full-size specimen fracture— Fig. 7. For example, increased delay shifted the fracture from the plate to 400 the sheet in a series of Type 304- Type 304 joints as the fusion weld area increased. Relative strengths for these joints are shown in Table 4 200 (19-9 through J9-15). Percent sheet ultimate tensile strength at weld fail­ ure for 17-7PH/Type 304 joints in­ creased from 38 to 50% for a delay .7 .8 .9 1.0 1.1 1 L increase of 1.0 to 1.5 millisec. These Ratio of decrease in sheet length to stickout, AL/S - strengths are low because a low pre­ Fig. 13—Effect of ratio of decrease in sheet length to stickout on weld ultimate heat current half-cycle was used and tensile strength and interfacial heating. Specimens machined from unbeveled sheet-to-plate tee joints of 17-7PH/Type 304 and 17-7PH/17-7PH

WELDING RESEARCH SUPPLEMENT | 437-s Only 1.0 millisec delay was used to Limitations of Program some of the other variables should obtain a full section fusion weld for This program was limited in scope also be studied. Among these, the the beveled joints. This was due to the since many of the possible variables effect of smoother and rougher faying current and force concentrating effect were held constant so that the few surface finishes on joint properties of the bevel. described could be studied. I feel that should be determined.

.40 .016 r—

£

.30

3

20

E o 10

1.6 1.8 2. 0 2. 2 2.4 Ratio of decrease in sheet ength to stickout, AL/S1-5 Fig. 14—Effect of ratio of decrease in sheet length to stickout on distance from weld interface to center of sheet upset. Unbeveled sheet-to-plate tee joints in 17-7PH/Type 304 and 17-7PH/17-7PH

O Sheet stress, 17-7PH/Type 304 460 r— • Fusion-weld fracture area, Type 304/Type 304 O Fusion-weld fracture area, 17-7PH/17-7PH

420 40

380 — .30 s

340

20 .50 1.00 1.25 1.50 1.75 Delay of magnet force, D, msec Fig. 15—Effect of delay of magnet force initiation (relative to weld current) on sheet stress at weld failure and fusion-weld fracture area. Unbeveled sheet-to-plate full-size tee joints

438-s | SEPTEMBER 1970 The value of using a weld current the ratio (AL/5), and delay of force Current is forced inward, from both reading as an indication of weld quali­ relative to current (D) were all im­ long and short edges, toward the ty should be examined. It is the au­ portant in obtaining highest strengths. cooler, lower-resistance material. Por­ thor's opinion that weld current read­ For beveled joints only AL and D tions of the interface near the short ings have been overemphasized in were important. edges receive increased current from resistance welding technology and can 6. MFUW was a successful welding both long and short edges. The cur­ be very misleading. In this study, for process in this study because heating rent density is necessarily higher near example, it was possible to have rela­ was concentrated at the interface be­ the short edges. This causes the fusion tively high weld current readings and tween the two dissimilar section thick­ weld front to move inward faster no interfacial heating or welding. It is nesses. This was accomplished by from the short edges. felt that weld voltage and resistance delaying the initiation of the magnetic Third, as the material near the readings are much more valuable. force wave and by using short weld edges is softened by resistance Because of the highly localized current times set by unique MFUW heating, it no longer is able to support heating capability offered by MFUW, controls. the load. The pressure on the remain­ the process may be applicable to ma­ ing material increases with a corre­ terials considered unweldable. Materi­ References sponding drop in contact resistance. als with low ductility and those ad­ 1. Del Vicchio, E. J., editor, Resistance This allows more current through the versely affected by heating from con­ Welding Manual, Third Edition, Resistance Welder Manufacturers' Association, Vol. 1, remaining center portion of the inter­ ventional welding processes may not Chapters 1 and 5 (1956). face. Thus, resistance heating in­ be more weldable. Among these may 2. Phillips. A. L., editor, Welding Hand­ creases, and the fusion weld front book, Fifth Edition, American Welding So­ be cast super-alloys and fiber-rein­ ciety. Sect. 2, Chapter 30 (1963). moves inward. forced composites. 3. Park, F. R., "Magnetic Force Weld­ Thus, increased current density and ing." Product Engineering, 29 (39), p. 4. Summary of Results (Sept. 15. 1958). more heating occur near the edges, 4. Boolen, R. F., "Magnetic Force Weld­ just ahead of the moving fusion weld Sheet-to-plate joints in AISI Type ing Applications at Battelle Northwest." Battelle Northwest Report, BNWL^22 front. So, all three of the factors 304 and 17-7PH stainless steels were (May 1967). described, together, cause the inward resistance welded using the magnetic 5. Schueler, A. W., "Magnetic Force Butt movement of fusion welding shown in Welding for Nuclear Cladding." WELDING force upset welding (MFUW) proc­ Fig. 7. ess. The results are summarized as JOURNAL, 47 (8), 638 to 643 (1968). 6. Funk. E. J., "Recent Developments in Delay of the magnetic force initia­ follows: Magnetic-Force Welding," WELDING JOUR­ tion allows more inward movement of 1. Full strength welds were NAL, 36 (6), 576 to 582 (1957). 7. Stevenson, W. D., Jr., Elements of the fusion weld front. In fact, too achieved in dissimilar thickness tee Power System Analysis, McGraw-Hill Book much delay results in excessive joints with both Type 304 and 17-7PH Co., He. (1962). 8. Timoshenko, S. P.. and Goodier, J. N., heating, expulsion and cracking. Too sheet and plate. Theory of Elasticity, Second Edition, Mc­ little delay does not allow enough heat 2. A stress concentration existed at Graw-Hill Book Co.. Inc. (1951). to be developed resulting in very little the sheet/plate interface under the 9. Anon.: Properties of Some Metals and Alloys. Chemical Composition-Mechanical fusion and only a marginal solid state weld fillet due to the presence of an Properties-Hardness-Physical Constants. unwelded area. Removal of the weld International Nickel Co., Inc., 1943. wHd. For the unbeveled joints, a delay fillet (and elimination of the stress of 1.5 millisec provided optimum concentration) resulted in about a Appendix: Edge Effect amounts of fusion and solid state welding. Complete fusion welding in 40% strength increase in full size Figures 6 and 7 show that there is unbeveled joints was accompanied by 17-7PH/17-7PH tee joints. Almost all more fusion welding in the edge sec­ shrinkage cracking. failures in these tee joints occurred tion than at the center. This is termed through the weld heat-affected zone the "edge effect." As resistance As magnetic force is apDlied, pres­ on the plate side of the weldment. The heating occurs at the sheet-plate in­ sure increases sharply on the nonfu- greater ductility of the Type 304 plate terface, the "fusion weld front" moves sion welded area of the interface. compared to that of the 17-7PH plate toward the center principally from the Contact resistance droos and even lessened the effect of the stress con­ short edges as shown by the arrows in though high current is available, little centration under the fillet. Fig. 7. Increases in delay (force rela­ resistance heating takes place due to 3. Because of the possible existence tive to current) increases the amount low contact resistance. Solid state of a stress concentration in this type of fusion weld front movement. There welding occurs because upsetting of tee joint, ductility is a more impor­ are three interrelated reasons for the brings hot sheet material into close tant criterion than strength in select­ edge effect. contact with the plate along the inter­ ing the plate material. Type 304 is The first factor is mechanical. Pres­ face. preferred over 17-7PH as the plate sure is not distributed uniformly Less delay (1.0 millisec) was used material because of the low ductility across the sheet thickness or width at with the molybdenum-insert tooling of 17-7PH plate in the short trans­ the original interface.8 Rather, it is when the unbeveled sheet was welded. verse direction. highest at all four edges of the cold The purpose of the molybdenum was 4. From metallurgical observations, interface and decreases parabolically to counteract the edge effect and the best welds were achieved with as the center is approached. This force more current into the center of beveled joint configurations. A small causes contact resistance near the the joint. The molybdenum inserts sheet bevel angle and initial contact edges to be lower. And this favors were shaped to form a more resistive thickness (.015 in. (.38 mm) 15°) more current flow and a higher cur­ path along the edge than the center. are preferred. Beveled joints were rent density near the edges. More heating took place toward the principally fusion welded; whereas, Second, heating occurs in the form center, and less delay was required. unbeveled joints were a combination of a thin outer ring at the sheet/plate However, the^e was a tendency to of fusion and solid state welding. interface because of the skin effect. overheat the sheet at the last point of 5. For unbeveled joints stickout However, as the material temperature contact with the molybdenum due to (5), decrease in sheet length (AL), rises, so does its electrical resistance. its low heat conductivity.

WELDING RESEARCH SUPPLEMENT I 439-s