failure mode transition of triplethinsheet aluminum ... · ture is partially melted zone (pmz),...

WELDING RESEARCH

DECEMBER 2016 / WELDING JOURNAL 479-s

Introduction Resistance spot welding (RSW) hasbeen widely used in the manufacturingof aerospace, electronics, and especial-ly the automotive industry because ofits high productivity, flexibility, andsuitability. The automotive industrymakes extensive application of RSWwith typically between 2000 and 5000spot welds in a motor vehicle (Ref. 1).It can be argued that the mechanicalperformance of the spot welds deter-mines the vehicle crashworthiness,which is defined as the capability of avehicular structure to provide ade-quate protection to its passengers dur-ing a crash (Ref. 2). Nowadays, with the demand forlightweight vehicular structures,

three-sheet RSW is increasingly exert-ed in some complex structures, such asfront longitudinal rails, A-, B-, and C-pillars, and the bulkhead to the innerwing (Ref. 3). Compared to two-sheetRSW, joining three sheets is morecomplicated because of the extra inter-face introduced and different materialand sheet thickness combinations.Therefore, it is important to under-stand the nugget growth and failurebehavior of the three-sheet spot weldjoint. Some researchers have investigatedthe nugget growth and mechanical be-havior of three-sheet RSW. Harlin etal. found that the position of the ini-tial heat generation is independent ofmaterial thickness and stack configu-ration (Ref. 4). They also found that

increasing the electrode force leads toa shift in the initial position of theweld nugget formation from thesheet/sheet interface to the center ofthe middle sheet (Ref. 5). Nielsen et al.studied the weldability of thin, low-carbon steel to two thicker, high-strength steels through factorial ex-perimentation and statistical analysis.They found that it is feasible to obtaina good weld with acceptable weldstrengths (Ref. 6). Pouranvari andMarashi investigated weld nugget de-velopment during RSW of three steelsheets of equal thickness. They founda critical sheet thickness of 1.5 mm atwhich the size of the fusion zone atthe sheet/sheet interface is nearlyequal to that of the fusion zone at thegeometrical center of the joint (Ref. 7).They also investigated the failure behavior of three-sheet low-carbonsteel under a different joint type andpointed out that the joint design sig-nificantly affects the mechanical prop-erties and the tendency to fail in theinterfacial failure mode (Ref. 2). Many other studies using finite ele-ment (FE) simulation investigated thenugget formation process of three-sheet RSW. Shen et al. performed acoupled electrical-thermal-mechanicalmodel to predict the weld nugget for-mation process of RSW of three steelsheets of unequal thicknesses (Ref. 3).Lei et al. built a two-dimensional FEmodel considering the thermal-electri-cal coupling for the RSW process ofmild steel (Ref. 8). Ma and Murakawadeveloped an FE program considering

Failure Mode Transition of TripleThinSheetAluminum Alloy Resistance Spot Welds under

TensileShear LoadsThe failure mechanism of threesheet 6061T6 aluminum alloy

resistance spot welds was investigated

BY Y. LI, Y. ZHANG, Z. LUO, H. SHAN, Y. Q. FENG, AND Z. X. LING

ABSTRACT This paper investigates the failure mode transition of triplethinsheet aluminum alloyresistance spot welds under tensileshear loads. Two stackups, i.e., 1.0/1.0/1.0 mm and1.5/1.0/2.0 mm, were examined. The tensileshear tests were performed for four different joint designs for each stackup. The failure process and failure mode transition of thefour types of joints in the 1.0/1.0/1.0 mm stackup were investigated through stepbystep experimental methods. An analytical model, which is suitable for the threesheetaluminum alloy resistance spot weld, was proposed to ensure the pullout failure mode.The results showed that a type of columnar grain, which has large secondary dendritearm spacing, was the weak area in the threesheet aluminum alloy resistance spot welds.The critical weld button size required to ensure the pullout failure mode was obtained.The proposed analytical models can be used to predict the critical button size for threesheet aluminum alloy resistance spot welds.

KEYWORDS • Resistance Spot Weld • ThreeSheet Spot Welding • Aluminum Alloy • Failure Mode

Y. LI, Y. ZHANG, Z. LUO, H. SHAN, Y. Q. FENG, and Z. X. LING are with the School of Materials Science and Engineering, Tianjin University, Tianjin, China. Z. LUO is also with the Collaborative Innovation Center of Advanced Ship and Deep Sea Exploration, Shanghai, China.

Yang Li Dec 2015120-REV.qxp_Layout 1 11/9/16 5:23 PM Page 479

WELDING RESEARCH

WELDING JOURNAL / DECEMBER 2016, VOL. 95480-s

the coupling of electrical, thermal, andmechanical fields to study the nuggetformation in RSW of high-strengthsteels (Ref. 9). Although many studies have beenperformed on the weld growth processof three-sheet spot welds, these re-searchers have all focused on mild orhigh-strength steels. The usage of alu-minum alloys in the automotive indus-try is gradually increasing due to itslight weight, good formability, andhigh corrosion resistance (Ref. 10).However, very little work in open liter-ature has studied the RSW of multiplealuminum alloy sheets. Although Li etal. investigated the weld growth mech-anism of three-sheet RSW for alu-minum alloys (Ref. 11), the failure behavior of the spot welds was not thefocus of the paper, especially the fail-ure transition mode, which is consid-ered an important characteristic of thespot weld joint. The present article investigates thefailure mechanism of three-sheet6061-T6 aluminum alloy resistance

spot welds (RSWs), especially the fail-ure mode transition behavior of thespot welds. Four types of joints are de-signed in this paper. The mechanicalproperties of three-sheet RSWs arealso investigated.

Experimental Procedures In this study, 6061-T6 aluminumalloy sheets with thicknesses of 1, 1.5,and 2.0 mm were used. Table 1 liststhe chemical composition of the mate-rials, while Table 2 lists the mechanicalproperties. Two thickness combinations wereused in the experiments. From the up-per electrode tip to the lower one, thetwo thickness combinations were1.0/1.0/1.0 mm and 1.5/1.0/2.0 mm,respectively. Four types of three-sheetjoints for each thickness combinationwere designed, as shown in Fig. 1. Inthe Type I and II joints, only one inter-face bore the tensile force during thetest. In the Type III and IV joints, bothof the two interfaces bore the tensile

force during the test. It is obvious thatthe stiffness of these joint designs andconsequently the tendency of the sam-

A B

Fig. 1 — Joint designs of the threesheet AA6061T6 RSW: A — 1.0/1.0/1.0 mm stack; B — 1.5/1.0/2.0 mm stack.

Table 1 — Chemical Composition of 6061T6 Aluminum Alloy (wt%)

Si Mg Zn Cu Mn Fe Cr Ti Al

0.56 1.10 0.25 0.25 0.15 0.70 0.18 0.15 Balanced

Table 2 — Mechanical Properties of 6061T6 Aluminum Alloy

Yield Strength Tensile Strength Elongation (MPa) (MPa) (%)

285 310 6

Fig. 2 — The exemplary microstructureof the 6061T6 resistance spot weldnugget in the 1.0/1.0/1.0 mm stack (18kA, 200 ms).

A

B

C

D

E


ples to rotate during a tensile-sheartest are different. The Type III jointexperienced the largest rotation whilethe Type IV joint bore pure shear dur-ing the tensile-shear test. These affect-ed the failure behavior and the suscep-tibility to fail in interfacial mode. Spot welding was performed usinga 220-kW direct current (DC) inverterRSW machine. The welding parame-ters are shown in Table 3. For the1.0/1.0/1.0 mm stack, ten samplewelds were performed per weldingcondition including three samples forthe complete tensile-shear test andseven samples for the step-by-steptensile-shear test. For the 1.5/1.0/2.0mm stack, three sample welds wereperformed per welding condition. Thesample dimensions used in this studywere 100 × 25 mm with a 25-mm-wideoverlap area. The tensile-shear testswere performed at a crosshead of 1mm min−1 with a CSS-44100 materialtest system. The maximum load of the CSS-44100 material test system is 200 kN and the initial distance be-tween the crosshead was 125 mm (the gripped zone on each sheet was25 mm). The peak load was evaluated usingthe average value of the three com-plete tensile-shear tests. Button sizewas measured from the failure surfaceof the welded joint. Here the “buttonsize” is used to evaluate the weld quali-ty rather than the “nugget size” be-cause the button size is easier to meas-ure in industrial production (Ref. 12).The seven step-by-step tensile-sheartests were used to investigate the fail-ure processes of the weld joints. Sevenspecimens were obtained from differ-ent stages (load raising stage, peakload stage, load drop stage, and final

fracture stage) during the tensile-shear test. Afterward, the seven speci-mens were ground, polished, andetched using standard metallographyprocedures. The cross sections of thewelds were etched by Keller’s reagent(1 mL hydrofluoric acid, 1.5 mL hy-drochloric acid, 2.5 mL nitric acid, and95 mL water). The Vickers microhard-ness test was performed using an in-denter load of 100 g for a period of 10 s.

Results and Discussion

Joint Microstructure

Figure 2 shows the exemplary microstructure of the 6061-T6 resist-ance spot weld nugget in the1.0/1.0/1.0 mm stack. The microstruc-ture of the resistance spot weld nuggetin the 1.5/1.0/2.0 mm stack was simi-lar to that in the 1.0/1.0/1.0 mmstack, and it will not be given here. As shown in Fig. 2B, from nuggetedge to nugget center, the microstruc-ture is partially melted zone (PMZ),columnar grain zone (CGZ), andequiaxed grain zone (EGZ). The PMZrefers to the area where temperaturewas between the solidus temperatureand the liquidus temperature duringwelding (Ref. 13). The material be-

comes a solid-liquid mixture, that is, itis partially melted. The formation ofthe CGZ is due to the relatively hightemperature gradient and low consti-tutional supercooling at the edge ofthe weld nugget. In the center of theweld nugget, a low-temperature gradi-ent and high constitutional supercool-ing contributes to the formation of theEGZ. Note that the columnar grain hastwo morphologies (Fig. 2B, 2D): thecolumnar grain with large secondarydendrite arm spacing (LCGZ) and thecolumnar grain with small secondarydendrite arm spacing (SCGZ). The for-mation of LCGZ was due to recessionof the nugget perimeter during weld-ing. The authors found that the LCGZwas easier to form at the lower inter-face (close to the negative electrode).This is because the Peltier effect (Refs.14, 15) leads to the coldest areas of theliquid nugget being adjacent to thenegative electrode when the weldingcurrent is switched off. This area con-tains less alloy content and is prone tocracking under stress. The creation ofa significant LCGZ is an indication of apoorly designed weld schedule (rela-tively low heat input). When the heatinput is high enough (a more suitableweld schedule), less or no LCGZ willform. Figure 3 shows the microhardnessdistribution of the weld nugget. The

WELDING RESEARCH


A

Fig. 3 — Microhardness profile of the threesheet aluminum alloy resistance spot weldjoint in the 1.0/1.0/1.0 mm stack.

Fig. 4 — Photos of the failure surface in the 1.0/1.0/1.0 mm stack: A — Interfacial failure;B — partial thicknesspartial pullout failure; C — pullout failure.

Table 3 — Experimental Parameters

StackUps Welding Current Welding Time Electrode Force Electrodes (mm) (kA) (ms) (kN) 1.0/1.0/1.0 16, 18, 20, 22 200 3.6 Truncated cone 50–300 with electrodes with a 1.0/1.0/1.0 20 increments 3.6 6mmdiameter tip of 50 1.5/1.0/2.0 18–34 with 120 4.0 Domed electrodes increments of 2 with a sphere radius of 50 mm and face diameter of 20 mm

B C


WELDING RESEARCH


lowest microhardness appears in theLCGZ, which has a coarser structureand less alloy content. This low hard-ness zone has a detrimental effect onthe mechanical properties of the joint,which will be discussed in the follow-ing sections.

Failure Mode Transition inThree Equal Thickness Stacks

Failure of Joint Types I and II

Three types of failure modes, inter-facial (IF) failure, partial thickness-partial pullout (PT-PP) failure (Ref.16), and pullout (PO) failure, are ob-served in joint Types I and II, asshown in Fig. 4. Only the load-displacement curvesof the Type II joint are given here dueto the similarity of the mechanical be-havior of the Type I and II joints, asshown in Fig. 5. The load smoothlydropped to zero after it reached itsmaximum value in the IF failure, whilea residual force remained after the forcebegan to drop in the PT-PP and PO fail-ures (Ref. 17). More details about thefailure process are shown in Fig. 6. Figures 6A and 6B show themacro/microstructures of the fracturesurface cross section of welds thatfailed in the IF mode. Figure 6A lo-cates the section where the forceachieved its maximum value, and acrack formed, explaining the subse-quent load reduction. The crack initial-ly formed between the PMZ and LCGZand then propagated through the inte-

rior of theLCGZ, and fi-nally failed asan interfacialcharacteriza-tion — Fig. 6B.The subopti-mized weldingparameters (16kA, inadequateheat input)contributed tothe formationof the LCGZ,which has a lowhardness andstrength to resist the crack propaga-tion. Note that the weld size in Figs.6A,B is inconsistent. This variationmay be caused by local differences incontact resistances of workpiece/work-piece and electrode/workpiece, whichwill alter the heat input and affect thenugget formation (Ref. 18). Figures 6C and 6D show fractureinitiation location of the welds thatfailed in the PT-PP and PO mode, re-spectively. In the PT-PP mode, the fail-ure location was the PMZ while thefailure of the PO mode was due tonecking of the base metal. This sug-gested that the PT-PP mode is a sub-optimized failure mode. Figure 7 shows the effect of buttonsize on the peak load and energy ab-sorption of joint designs I and II. Sim-ple linear regression was applied toboth the data obtained from joints Iand II, and a best fit line with a coeffi-cient of determination (R2) of 0.878,was obtained. The relatively high value

of R2 suggested that a linear relation-ship exists between the peak load andbutton size. This phenomenon is alsoobserved by Han et al. (Ref. 18) andSun et al. (Ref. 19). However, a two-order-polynomial relation exists be-tween the energy absorption and but-ton size, indicating that a larger weldnugget could not only improve thepeak load, but also the microstructure(less LCGZ due to suitable heat input)and relieve the stress concentrationaround the weld nugget and, in turn,improve the ductility of the weld joint.

Failure of Joint Type III

Similar to joint Types I and II, IF,PT-PP, and PO failures were observedin the Type III joint. Only the PO fail-ure mode will be discussed here be-cause the analysis of the PT-PP failuremode is similar to that for joint TypesI and II. Figure 8 shows the macro/mi-

A B

C B

Fig. 5 — Typical loaddisplacement curves of the Type II joint inthe 1.0/1.0/1.0 mm stack.

Fig. 6 — Macro/microstructures of Type II weld joints in the1.0/1.0/1.0 mm stack that failed in A, B — IF mode (16 kA, 200 ms);C — PTPP mode (20 kA, 200 ms); D — PO mode (20 kA, 200 ms).

Yang Li Dec 2015120-REV.qxp_Layout 1 11/9/16 5:25 PM 82

WELDING RESEARCH


crostructures of the fracture surfacecross section of welds that failed in IFmode and PO mode. From Fig. 8B, itcan be seen that the crack began toform at the interior of LCGZ and thenpropagated through the interface be-tween SCGZ and LCGZ. A crack wasalso found on the other workpiece/workpiece interface —Fig. 8C. This in-ferred that there is a competition be-tween the two interfaces in a three-sheet spot weld joint, and that failurewill occur on the weaker one. As can beseen with Fig. 8B, C, the weld failed atthe interface where LCGZ formed (Fig.8B) while the crack propagation wasrestrained at the interface where noLCGZ formed — Fig. 8C. This againverifies that the LCGZ is the weak areain a spot weld. In the PO mode, theweld joint failed in the PMZ — Fig. 8E.There is also a crack on the otherworkpiece/workpiece interface (Fig.8F), indicating competition betweenthe two interfaces during the tensile-shear test. Figure 9 shows the effect of buttonsize on the peak load and energy ab-sorption of joint design III. The mini-mum button size that guarantees a POmode was 5.1 mm. The results aresimilar to the case of joint Types I andII, i.e., a linear relation and two-order-polynomial relation exists between the peak load and button size, and energy absorption and button size, respectively.

Failure of Joint Type IV

Since the Type IV joint experienced

a pure shear dur-ing the tensile-shear test, thefailure modes ofthis type of jointare a little differ-ent from thoseof joint Types I,II, and III. Whenthe nugget sizewas small, bothof the two inter-faces failedthrough the in-terfacial mecha-nism — Fig.10A. This iscalled a doubleinterfacial fail-ure (DIF) in thispaper. Note thatthe nugget devi-ated from itsoriginal posi-tion. The de-tailed processwill be discussedin the followingtext. When thenugget size grew larger, one interfacefailed through the interfacial mecha-nism while the other one failedthrough the pullout mechanism (Fig.10B), and this is called a one interfa-cial/one pullout (IF/PO) failure. Whenthe nugget size was large enough, thebase metal fractured during the tensileprocess (Fig. 10C), and this is called abase metal fracture (BMF) failure. Figure 11A shows the typical load-displacement curves of the Type IV

weld joint that interfacially failed. Theload-displacement curve has twopeaks. The first drop of the load corre-sponds to the initial formation of acrack, as shown in Fig. 11, B1. Figure11C corresponds to the second peak. Itcan be seen that the middle sheet waspulled out along the tensile direction— Fig. 11, C1. The nugget wassqueezed (Fig. 11, C2) by the middlesheet because of the movement ofmiddle sheet. At the same time, cracks

DC = 3tIDP

FL

WN

cos IFcos PO

8( )

Fig. 7 — Effect of button size on the peak load and energy ofjoint designs I and II in the 1.0/1.0/1.0 mm stack.

Fig. 8 — Macro/microstructures of Type III weld joints in the1.0/1.0/1.0 mm stack that failed in A, B, C — IF mode; D, E, F — POmode.

Fig. 9 — Effect of button size on the peak load and energy absorption of the Type III joint in the 1.0/1.0/1.0 mm stack.

A

B

C

D

E

F


formed and propagated at both of thetwo interfaces. Figure 11, C3 showsthe deformed EGZ. The nugget can beconsidered as experiencing work hard-ening. It can be seen that the micro-hardness of EGZ increased with an in-creasing deformation degree. Thiscauses the load to rise again. Figure11D is the stage when only one inter-face failed. Figure 11, D1 shows thefracture occurred in the interior of theLCGZ. It is interesting to note that onthe other interface, the deformedgrain induced the crack propagatedinto the interior of weld nugget — Fig.11, D2. As a consequence, a failed weldjoint in the final stage of the load-displacement curve as shown in Fig.11A is formed. The load-displacement curve of theType IV weld joint that failed in theIF/PO mode is similar to that in theDIF mode. Figure 12 shows the load-displacement curve and microstruc-tures of the Type IV weld joint thatfailed in the BMF mode. The curve hasa “platform,” which indicates that thecrack is propagating in the base metaland, therefore, the load is relativelystable. The weld nugget had very smalldeformation during the tensile processcompared with those that failed in theDIF and IF/PO modes. This indicatesthat the weld nugget was large enoughto resist being squeezed by the middlesheet and therefore the crack formedaround the edge of the weld nuggetand then propagated to the base metal. Figure 13 shows the effect of weld-ing time on the peak load and energyabsorption of joint design IV. The ener-gy absorption in the Type IV joint is de-fined by the area under the load-displacement curve up to the secondpeak. Although a welding time of 250ms should be used to guarantee theBMF mode, the peak load and energyabsorption were almost unchanged af-ter a welding time of 200 ms. This is be-cause the hardening of the weld nuggetcompensated the relatively small size ofthe weld nugget when the weld jointfailed in the DIF and IF/PO modes. From the macroscopic photos of thefailed joint and the microstructure ofthe failed weld joint, it can be seen thatthe weld nugget either deformed (DIFand IF/PO modes) or became invisible(BMF mode) after the joint failed. As aresult, the accuracy of the button sizecannot be measured from the failed

weld joint, and the effect of button sizeon the peak load and energy of jointdesign IV cannot be obtained.

Failure Mode Transition in Three Unequal Thickness Stacks

The overall failure rules of the

1.5/1.0/2.0 mm stack were similar tothat of 1.0/1.0/1.0 mm stack. Figure14 shows the macrostructures of weldjoints in 1.5/1.0/2.0 mm stack. For allfour types of joints, the IF failure loca-tion moved from LCGZ to EGZ. Infact, no obvious LCGZ formed in the1.5/1.0/2.0 mm stack. This is because

WELDING RESEARCH


Fig. 10 — Photos of failure modes of the Type IV joint in the 1.0/1.0/1.0 mm stack: A —Double interfacial failure; B — one interfacial/one pullout failure; C — base metal fracture failure.

Fig. 11 — Typical loaddisplacement curve and microstructures of Type IV weld joints inthe 1.0/1.0/1.0 mm stack which failed by the interfacial mechanism (18 kA, 200 ms).

A

AA

B

B1

C1 C2 C3

D1 D2 D3

B2 B3

C


the domed electrodes and hard normwelding parameters lead to a moreconcentrated heat generation and lessheat dissipation, which is beneficial toinhibiting the formation of LCGZ, asdiscussed in the section of “joint mi-crostructure.” For the Type I joint, the PO failurelocation was located at the SCGZ. Allthe Type II joints failed in IF mode, al-though the button size reached toabout 10 mm. This will be discussed in

the following text. For the Type IIIjoint, the PO failure location was thePMZ. For the Type IV joint, the BMFfailure location was also located at thePMZ. Figure 15 shows the effect of but-ton size on the peak load of joint de-signs I, II, and III. Similar to the1.0/1.0/1.0 mm stack, good linear re-lationships exist between the peakload and button size. The critical but-ton sizes for the Type I joint and Type

III joint were 9.1 and 8.2 mm, respec-tively. However, all the Type II jointsfailed in IF mode during the tensile-shear test. The failure mode of jointdesign IV in the 1.5/1.0/2.0 mm stackwas similar to that in the thicknesscombination of 1.0 /1.0 /1.0 mm. Thecritical button size was about 6.2 mm,which is nearly the same as that in the1.0/1.0/1.0 mm stack (6.25 mm). Thisindicates that for the joint design ofpure shear, the critical weld nuggetsize or button size may be controlledby the thickness of the middle sheet.

Analytical Model to Predict Failure of ThreeSheetAluminum Spot Welds

Pouranvari et al. proposed a simpleanalytical failure model for the RSW ofsteel (Refs. 21, 22). However, the weldrotation was not considered in theirmode. In this paper, an analyticalmodel considering the weld rotationwas developed based on weld areastress analysis. VandenBossche analyzed the stressdistribution when a spot weld failed inthe IF and PO modes (Ref. 23). Asshown in Fig. 16B, once the weld ro-tates, the load on the weld interfacecan be decomposed to two compo-nents: the force N normal to the fayingsurface and the force S parallel to it.They are related to F by

S=F cos (1) N=F sin (2)

In the tensile-shear test, the driv-ing force for the IF mode is the shearstress at the sheet/sheet interface(Ref. 24). The shear load S generates ashear stress S distributed across theinterface. If the average value of theshear stress is V/A, then the maximumvalue is (Ref. 23)

where IF is the weld rotation anglewhen the joint experiences IF failure. The driving force for the PO modeis the tensile stress around the nugget(Ref. 24). As shown in Fig. 16C, thetensile stress due to S is

τSMAXIF = 3S

2A= 6FcosθIF

πd2 3( )

WELDING RESEARCH


Fig. 12 — Typical loaddisplacement curve of Type IV weld joints in the 1.0/1.0/1.0mm stack that failed in BMF mode (22 kA, 200 ms).

Fig. 13 — Effect of welding time on the peak load and energy of joint design IV in the1.0/1.0/1.0 mm stack (20 kA).


WELDING RESEARCH


where PO is the weld rotation anglewhen joint experiences PO failure.The rotation models of the four typesof joints are schematically shown inFig. 17. It is obvious that the aboveequations can be applied to the jointTypes I and II directly. For joint TypeIII, although both the two interfacesbear the tensile-shear load during thetensile-shear test, each interface bearsthe total tensile-shear load not thehalf of it. Accordingly, the above equa-tions are also suitable for the Type IIIjoint. Joint IV experiences pure shearduring the tensile-shear test, i.e., thejoint will not rotate during the test.The above model is not suitable forjoint Type IV. When the maximum shear stressexceeds the shear strength of the weldnugget, a crack will form at the weldinterface. This moment correspondsto where the tensile-shear forcereached its peak, as shown in Figs. 5,6. Letting the maximum shear stressequal to the shear strength of the weldnugget, and then the failure load atthe IF mode FIF can be expressed as

where d is the weld nugget, and WN isthe shear strength of the weld nugget.For a three-sheet RSW, d should be re-placed by dIN, which is the weld nuggetdiameter at the failure interface. Con-sidering that the aluminum spot weldsare more sensitive to porosity or voids,a porosity factor P can be introducedinto Equation 5 (Ref. 19)

where P = (Atotal - Aporosity)/Atotal. Atotal isthe area of the fusion zone on the frac-ture surface and Aporosity is the area ofthe porosity on the fracture surface ofthe weld. Letting the shear stress equal thetensile strength of the pullout failurelocation, then the peak load for a weldto fail in the pullout mode under thetensile-shear test can be approximated

as

where tID is the local sheet thicknessaround the nugget accounting for in-dentation, and σFL is the shearstrength of the failure location. In order to ensure pullout failurefor a spot weld, the failure load for aPO failure should be less than that forIF failure, i.e., FPO < FIF. Thus, the criti-cal nugget diameter DC can be ob-tained from Equations 6 and 7.

Applying the linear relationship be-tween the strength and hardness, andthe linear approximate between shearstrength and tensile strength, Equa-tion 8 can be rewritten as

where HFL is the hardness of the failurelocation, HWN is the hardness of theweld nugget, and f is a constant coeffi-cient. For aluminum alloys, f is about0.6 (Ref. 25). In this study, HWN shouldbe replaced by HLCGZ because the fail-ure location of the IF mode occurred inthe LCGZ. Therefore, Equation 9 canbe rewritten as a more widely applica-ble form

FIF = πd2

6cosθIFτWN 5( )

FIF = PπdIN

2( )6cosθIF

τWN 6( )

FPO = dINtID2cos PO

FL 7( )

DC = 3tIDP

FL

WN

cos IFcos PO

8( )

DC = 3tIDPf

HFLHWN

cosθIFcosθPO

9( )

σSPO = S

A= S

πdt / 2= 2FcosθPO

πdt4( )

Fig. 14 —Macrostructures of weld joints in 1.5/1.0/2.0 mm stack: A — IF failure in Type Ijoint (18 kA); B — PO failure in Type I joint (32 kA); C — IF failure in Type II joint (18 kA); D —IF failure in Type II joint (34 kA); E — IF failure in Type III joint (18 kA); F — PO failure in TypeIII joint (26 kA); G — DIF failure in Type IV joint (18 kA); H — BMF in Type IV joint (22 kA).

BA

DC

FE

HG


WELDING RESEARCH


where HPO is the hardness of pulloutfailure location, and HIF is the hard-ness of interfacial failure location. Apply Equation 10 to the Types Iand II joints of the 1.0/1.0/1.0 mmstack, The failure location in the POmode was the BM. The average hard-ness of the BM was 95 Hv. In the caseof Types I and II joints, the failure lo-cation of the IF mode was the failurelocation in the interior of the LCGZ,where the porosity is hard to form.Thus, the Aporosity equals 0 and P is 1.The average indentation was about70% of the original sheet thickness,i.e., tID was 0.7 mm. The rotation anglewas measured after the tensile-sheartest. It was nearly zero when the jointfailed in the IF mode (16 kA, 200 ms),while it was 2 deg when the joint failedin the PO mode (20 kA, 200 ms). The critical nugget diameter forTypes I and II joints can be obtained asfollows:

It can be seen that the predictedvalue is very close to the experimentalresult of 5.9 mm. In the case of a Type III joint, thefailure location in the IF mode was stillin the interior of LCGZ or along the in-

terface between the LCGZ and SCGZ.Note that the failure location in thePO mode changed to PMZ, as shownin Fig. 9. The average indentation wasabout 90% of the original sheet thick-ness, i.e., tID was 0.9 mm. The averagerotation angle was about 2 deg whenthe joint failed in the IF mode (16 kA,200 ms), while it was 7 deg when thejoint failed in the PO mode (18 kA,200 ms). Thus, the critical nugget di-ameter for the Type III joint can be ob-tained as follows:

The result is a little larger than the ex-perimental value (5.1 mm). Equations 1–11 are not suitable forthe Type IV joint because the failuremode of the Type IV joint is differentfrom the other types of joints. This pa-

DC( )Types I&II = 3tIDPf

HBMHLCGZ

cos IFcos PO

= 3 0.70.6

9555

1

cos2 deg6.0 mm 11A( )

310)(= θ

θD

tPf

HH

coscosC

ID PO

IF

IF

PO

DC( )Type III = 3tIDPf

HPMZHLCGZ

cos IFcos PO

= 3 0.90.6

6555

cos2degcos7deg

5.3 mm 11B( )

Fig. 15 — Effect of button size on the peak load of the Types I, II, andIII joints for the 1.5/1.0/2.0 mm stack.

Fig. 16 — Stress analysis in the weld area: A — Weld rotation; B— IF failure; C — PO failure (Ref. 23).

A B C

Fig. 17 — Schematic of joint rotation in the 1.0/1.0/1.0 mm stack: A — Type I joint; B —Type II joint; C— Type III joint; D — Type IV joint.

B

A

C

D


per constructs a model for predictingthe failure mode for the Type IV joint. From the above discussion, it canbe known that in the DIF failure of theType IV joint, one interface failsthrough the LCGZ while the other in-terface fails through the interior of theweld nugget, i.e., through the EGZ.For simplification, the deformationand the work hardening of the weldnugget were ignored. The maximumshear stress at each interface is

Accordingly, the failure load at theDIF mode can be expressed

where EGZ is the shear strength of theequiaxed grain zone, and LCGZ is theshear strength of the columnar grainwith a large secondary dendrite armspacing. In order to determine the mathe-matical equation of the failure load forthe BMF mode, Fig. 18 depicted thefailure analysis of the BMF in the TypeIV joint. LO is the length of overlap-ping, which is equal to the width ofthe workpiece W. The failure locationaround the weld nugget of the BMFmode was the PMZ as shown in Fig.12. Thus, the failure load for the BMFmode can be expressed as

Note that the thickness t in Equa-tion 14 is the thickness of the middlesheet, which was not influenced by theindentation. Combining Equations 13 and 14,the following equation can be obtained

The solution for Equation 15 is

Applying the linear relationship be-tween the strength and hardness, andthe linear approximation between theshear strength and tensile strength,Equation 16 can be rewritten as

Using W = 25 mm, t = 1 mm, f = 0.6,HPMZ = 65 Hv, HBM = 95 Hv, HEGZ = 60Hv, and HLCGZ = 55 Hv, the critical nugget diameter of the Type IV joint is

Although the predicted value is small-er than the experimental result (about6.25 mm), based on the experimentalresults, when the button size wasabout 6 mm (the corresponding weld-ing parameters were 20 kA and 200ms), both the peak load and energy ab-sorption of weld joints were similar tothe weld joints that failed in BMFmode. Accordingly, the predicted re-sult is also acceptable for the Type IVjoint. For the Type I joint of the1.5/1.0/2.0 mm stack, the failure loca-tion in the PO mode was the SCGZ.The average hardness of the SCGZ was85 Hv. The increased hardness in theSCGZ maybe due to the hard norm ofwelding parameters. The failure loca-tion in the IF mode was the EGZ. Theaverage hardness of the EGZ was 60Hv. The average indentation was about85% of the uppersheet i.e., the tIN wasabout 1.3 mm. The rotation angle wasnearly zero when the joint failed in theIF mode, while it was about 2 degwhen the joint failed in the PO mode.

3 EGZ +3 LGGZ d2

+ 12t BM 2

t PMZ d

W2t BM + 3W

4t BM = 0 15( )

FBMF = 32W2t BM + d

2t PMZ

+W d2

t BM 14( )

DC( )Type IV =

= 12t BM 2

t PMZ +

12t BM 2

t PMZ

2

+ 4+3 EGZ + 4+

3 LCGZ

W2t BM + 3W

4t BM

2+3 EGZ + 2+

3 LCGZ

DC( )Type IV =

34

tHPMZ tHBM( )

+ 32

12tHBM 2

tHPMZ

2

+ 4+3fHEGZ + 4+

3fHLCGZ

W2tHBM + 3W

4tfHBM

fHEGZ + fHLCGZ

DC( )Type IV 6.0 mmτSIMAX = 3F / 2

2A= 3F

πd2 12( )

FDIF =πdID

2( )3

τLCGZ +πdID

2( )3

τEGZ 13( )

WELDING RESEARCH


Fig. 18 — Failure analysis of the BMF in the Type IV joint in the1.0/1.0/1.0 mm stack. Fig. 19 — Effect of joint design on the failure mode transition.

(16)

(17)


The critical nugget diameter for theType I joint can be obtained as follows

This predicted value is very close tothe experimental result of 9.1 mm. For the Type II joint of the1.5/1.0/2.0 mm stack, all the jointsfailed in IF mode, assuming that thePO failure location of Type II joint isthe PMZ and the rotation angle is thesame as Type I joint. Note that the IFfailure location was the EGZ ratherthan the LCGZ because the nugget willshift to the thicker sheet. The criticalnugget diameter for Type II should be

However, the maximum button sizeobtained from experiments was about10 mm. Therefore, the prediction forthe Type II joint is also reasonable. For the Type III joint of the1.5/1.0/2.0 mm stack, the IF failurelocation was the LCGZ, while the POfailure location was the PMZ. The av-erage indentation was about 90% ofthe original sheet thickness, i.e., tIDwas 1.35 mm. The average rotation an-gle was about 3 deg when the jointfailed in the IF mode, while it was 10deg when the joint failed in the POmode. Thus, the critical nugget diame-ter for the Type III joint can be ob-tained as follows:

The predicted value is very close to theexperimental result of 8.2 mm. For the Type IV joint of the1.5/1.0/2.0 mm stack, it can be seenthat in the DIF failure (Fig. 14), bothof the two interfaces failed throughthe EGZ. Accordingly, all the HLCGZ inEquation 17 should be replaced byHEGZ. Using W = 25 mm, t = 1 mm, f =0.6, HPMZ = 75 Hv, HBM = 95 Hv, andHEGZ = 60 Hv, the critical nugget diam-

eter of the Type IV joint is

The predicted value is close to the pre-dicted result of the 1.0/1.0/1.0 mmstack. This is reasonable because theBMF failure is dependent on the prop-erty of the middle sheet. Since themiddle sheets in the two thicknesscombinations were the same, the ex-perimental and predicted resultsshould be similar.

Effect of Joint Design on theFailure Mode Transition

The effect of joint design on thefailure mode transition is shown inFig. 19. The data point for the Type IIjoint in the 1.5/1.0/2.0 mm stackcomes from the predicted results. For the 1.0/1.0/1.0 mm stack, thetendency to fail in the IF mode is in-creased in the order Type III, Types Iand II, and Type IV. This is consistentwith Pouranvari and Marashi’s work(Ref. 2). The failure of the weld joint isthe competition between shear stressat the sheet/sheet interface (i.e., IFfailure) and the tensile stress at thenugget circumference (i.e., PO failure)(Ref. 20). The higher the shear stressat the sheet/sheet interface, the high-er the tendency to fail in the IF mode.The Type III joint has the maximumrotation angle and the minimum shearstress at the sheet/sheet interface.Therefore, it has the minimum criticaldiameter DC to fail in the PO mode. Incontrast, the sheet/sheet interfaces inthe Type IV joint experienced pureshear. The weld joint has virtually norotation and therefore, it has thelargest critical diameter DC to fail inthe PO mode (BMF mode). For the 1.5/1.0/2.0 mm stack, thetendency to fail in the IF mode is in-creased in the order of Type III, TypeIV, Type I, and Type II. Without con-sidering the Type IV joint, i.e., thepure shear condition, the failure rulesfor the two thickness combinationsare similar. The Type III joint experi-enced the maximum rotation while theType II joint has the minimum rota-tion angle. However, although theType IV joint experienced pure shear,the strength of the middle sheet waslower than the shear strength of twosheet/sheet interfaces. Therefore, for

the three unequal thickness stacks, thethickness of the middle sheet shouldcontrol the critical weld nugget size ofpure shear joint.

Conclusions and FutureWork

In this paper, the failure mode tran-sition of three-sheet aluminum alloyresistance spot welds (RSWs) duringtensile-shear tests were investigatedthrough experiments and an analyticalmodel. Four types of joints were inves-tigated. The following conclusions canbe drawn: 1) The microstructure in the three-sheet 6061 aluminum alloy RSWs con-sists of a partially melted zone (PMZ),columnar grain zone (CGZ), andequiaxed grain zone (EGZ), where thecolumnar grain zone is divided intothe columnar grain with large second-ary dendrite arm spacing (LCGZ) andthe columnar grain with small second-ary dendrite arm spacing (SCGZ). Thehardness test indicates that the LCGZhas the lowest hardness. 2) Three failure modes in Types I,II, and III joints, named the interfacial(IF) failure, partial thickness-partialpullout (PT-PP) failure, and pullout(PO) failure, were observed. There isno critical welding parameter ornugget diameter to separate the PT-PPand PO failures. The formation of theLCGZ in the weld nugget contributesto the PT-PP failure. There is a compe-tition between the two interfaces inthe Type III joint, and failure will occuron the weaker one. 3) Three failure modes in the TypeIV joint, named the double interfacial(DIF) failure, one interfacial/one pull-out (IF/PO) failure, and the base metalfracture (BMF) failure were identified.In the case of the DIF and IF/PO fail-ures, the nugget was squeezed and ex-perienced work hardening. In the DIFfailure, one interface failed through theLCGZ first, and then the other inter-face failed through the interior of theweld nugget. In the case of IF/PO fail-ure, the weld nugget experienced lessdeformation due to its larger nuggetsize. In the case of BMF failure, theweld nugget had a very small deforma-tion and the crack formed around theedge of the weld nugget and then prop-agated to the base metal. 4) The LCGZ is the weak area in

DC( )Type III = 3tIDPf

HPMZHEGZ

cos IFcos PO

= 3 1.350.6

7560

cos 3 deg

cos 10 deg8.4 mm

DC( )Type IV 6.0 mm

DC( )Type II = 3tIDPf

HPMZHEGZ

cos IFcos PO

= 3 1.80.6

7560

1

cos 2 deg11.6 mm

DC( )Type I = 3tIDPf

HSCGZHEGZ

cos IFcos PO

= 3 1.30.6

8560

1cos 2 deg

9.2 mm

WELDING RESEARCH



three-sheet aluminum alloy RSWs.Cracks will form and propagate in theinterior of the LCGZ or along the in-terface of SCGZ and LCGZ during thetensile-shear test. 5) The following equations are pro-

posed to predict the critical nugget di-ameter required to ensure POfailure mode during the tensile-shear

tests of three-sheet aluminum alloyspot weld joints

where t is thickness of the middle sheet, tID is the sheet thickness consid-ering the indentation, W is the widthof the sheet, P is the porosity factor, fis a constant coefficient, IF is the rota-tion angle when the joint fails in the IFmode, PO is the rotation angle whenthe joint fails in the PO mode, HFL isthe hardness of the failure location,HLCGZ is the hardness of the columnargrain with a large secondary dendritearm spacing, HBM is the hardness of thebase metal, HPMZ is the hardness of thepartially melted zone, and HEGZ is thehardness of the equiaxed grain zone. 6) The joint design has a significanteffect on the failure mode transition.For three equal-thickness sheet RSWs,the critical weld nugget diameter (DC)required for obtaining a PO failuremode during the tensile-shear test in-creases in order of Type III, Types Iand II, and Type IV. For three unequal-thickness sheet RSWs, the DC may becontrolled by the thickness of the mid-dle sheet for the joint design of pureshear. This paper preliminarily investi-gates the failure behavior of triple-thin-sheet aluminum alloy resistancespot welds under tensile-shear loads.There are many contents that needfurther research. More thickness com-binations should be tested to verifythe proposed analytical mode. Thefailure behaviors of spot welds under

other loading conditions, such ascoach peel and cross tension, are im-portant issues and need to be studied.The failure behaviors of spot welds ofother materials, such as 5000 series al-loys, or dissimilar materials, such as5000 series alloys to 6000 series al-loys, are valuable to pursue.

This research is supported by theNational Nature Science Foundationof China (Grants 51405334 and51275342).

1. Chao, Y. J. 2003. Ultimate strengthand failure mechanism of resistance spotweld subjected to tensile, shear, or com-bined tensile/shear loads. Journal of Engi-neering Materials and Technology 125:125–132. 2. Pouranvari, M., and Marashi, S. P. H.2012. Failure behavior of three-steelsheets resistance spot welds: effect of jointdesign. Journal of Materials Engineering andPerformance 21(8): 1669–1675. 3. Shen, J., Zhang, Y. S., Lai, X. M., andWang, P. C. 2011. Modeling of resistancespot welding of multiple stacks of steelsheets. Materials & Design 32: 550–560. 4. Harlin, N., Jones, T. B., and Parker, J.D. 2002. Weld growth mechanisms duringresistance spot welding of two and threethickness lap joints. Science and Technologyof Welding and Joining 7(1): 35–41. 5. Harlin, N., Jones, T. B., and Parker, J.D. 2003. Weld growth mechanism of resist-ance spot welds in zinc coated steel. Jour-nal of Materials Processing Technology143–144: 448–453. 6. Nielsen, C. V., Friis, K. S., Zhang, W.,and Bay, N. 2011. Three-sheet spot weld-ing of advanced high-strength steels. Weld-ing Journal 90(2): 32-s to 40-s. 7. Pouranvari, M., and Marashi, S. P. H.2011. Critical sheet thickness for weldnugget growth during resistance spotwelding of three-steel sheets. Science andTechnology of Welding and Joining 16(2):162–165. 8. Lei, Z. Z., Kang, H. T., and Liu, Y. G.2011. Finite element analysis for transientthermal characteristics of resistance spotwelding process with three sheets assem-blies. Procedia Engineering 16: 622–631. 9 Ma, N., and Murakawa, H. 2010. Nu-merical and experimental study on nuggetformation in resistance spot welding forthree pieces of high strength steel sheets.Journal of Materials Processing Technology210: 2045–2052. 10. Katayama, S., and Kawahito, Y.2010. Evolution of laser welding to dissim-ilar materials joining. Transactions of JWRI39(2): 268–269.

11. Li, Y., Yan, F. Y., Luo, Z., Chao, Y. J.,Ao, S. S., and Cui, X. T. 2015. Weld growthmechanisms and failure behavior of three-sheet resistance spot welds made of 5052aluminum alloy. Journal of Materials Engi-neering and Performance 24(6): 2546–2555. 12. Hongyan, Z., and Jacek, S. 2012.Resistance Welding: Fundamental and Ap-plications, 2nd ed. Boca Raton, CRC Press. 13. Kou, S. 2003. Welding Metallurgy,2nd ed. New Jersey, John Wiley & Sons, Inc. 14. Drebushchak, V. A. 2008. The Pelti-er Effect. Journal of Thermal Analysis andCalorimetry 91(1): 311–315. 15. Li, B. Q. 2002. Research on the nu-merical Simulation of the Process for Alu-minum Alloy Resistance Spot Welding andEnergy Analysis. PhD Dissertation. Tian-jin, Tianjin University. 16. Pouranvari, M., Marashi, S. P. H.,and Mousavizadeh, S. M. 2010. Failuremode transition and mechanical propertiesof similar and dissimilar resistance spotwelds of DP600 and low carbon steels. Sci-ence and Technology of Welding and Joining15(7): 625–631. 17. Marya, M., Wang, K., Hector, L. G.,and Gayden, X. H. 2006. Tensile-shearforces and fracture modes in single andmultiple weld specimens in dual-phasesteels. Transactions of the ASME Journal ofManufacturing Science and Engineering128(1): 287–298. 18. Han, L., Thornton, M., Boomer, D.,and Shergold, M. 2011. A correlation studyof mechanical strength of resistance spotwelding of AA5754 aluminium alloy. Jour-nal of Materials Processing Technology 211:513–521. 19. Sun, X., Stephens, E. V., Davies, R.W., Khaleel, M. A., and Spinella, D. J. 2004.Effects of fusion zone size on failuremodes and static strength of aluminum re-sistance spot welds. Welding Journal83(11): 308-s to 318-s. 20. Pouranvari, M., and Marashi, S. P.H. 2011. Failure mode transition in AHSSresistance spot welds. Part I. Controllingfactors. Materials Science and Engineering A528(29–30): 8337–8343. 21. Pouranvari, M., Asgari, H. R.,Mosavizadch, S. M., Marashi, P. H., andGoodarzi, M. 2007. Effect of weld nuggetsize on overload failure mode of resistancespot welds. Science and Technology of Weld-ing and Joining 12(3): 217–225. 22. Pouranvari, M., and Marashi, S. P.H. 2012. Failure mode transition in AISI304 resistance spot welds. Welding Journal91(11): 303-s to 309-s. 23. VandenBossche, D. J. 1977. Ulti-mate strength and failure mode of spotwelds in high strength steels. SAE TechnicalPaper 770214: 955–966. 24. Pouranvari, M., and Marashi, S. P.H. 2013. Critical review of automotivesteels spot welding: Process, structure andproperties. Science and Technology of Weld-ing and Joining 18(5): 361–403. 25. Kissell, J. R., and Ferry, R. L. 2002.Aluminum Structures: A Guide to Their Speci-fications and Design, 2nd ed. New York,John Wiley & Sons, Inc.

DC( )Type IV =

34

tHPMZ tHBM( )

+ 32

12tHBM 2

tHPMZ

2

+ 4+3fHEGZ + 4+

3fHLCGZ

W2tHBM + 3W

4tfHBM

fHEGZ + fHLCGZ

DC( )Types I&II&III = 3tIDPf

HFLHLCGZ

cos IFcos PO

WELDING RESEARCH


References

Acknowledgments


failure mode transition of triplethinsheet aluminum ... · ture is partially melted zone (pmz),...

Documents