Journal of Materials Processing Technology 210 (2010) 14551463

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

journa l homepage: www.e lsev ier .com/ l

Analys owweldin

Shigeki H moTsuyoshia Department o yogo-kb Hitachi Amerc Hitachi Resead Department o shi, Ku

a r t i c l

Article history:Received 25 JaReceived in reAccepted 6 Ap

Keywords:Friction stir spot weldingManufacturing process simulationTemperature distributionPlastic deformationParticle method

lasticicle medictsdownated igood

elds. Tquantied using numerical methods and is expressed as standard deviation of the particle movement. Atriangular pin with a concave shoulder is the preferred tool geometry from the current study that resultsin high strength spot welds.

2010 Elsevier B.V. All rights reserved.

1. Introdu

Frictioncess. A nonfashion intogenerated aaction of a vThe tool is ttermined spmaterial arowith translmain advanduring weldpieces, thusarcweldingof the frictioA non-consbe joined.Uis held in threferred tois then retr

CorresponE-mail add

0924-0136/$ doi:10.1016/j.ction

stir welding is a relatively new solid state joining pro--consumable rotating tool is plunged in a controlledthe workpieces that are being joined. Frictional heat ist the toolwork material interface by the tool under theertical load, thereby reducing the material ow stress.hen traversed along the desired weld path at a prede-eed to realize the weld. The localized heating softensund the tool and the action of tool rotation combined

ation results in a solid state joint being produced. Thetage of the friction stir welding is that the temperatureing is less than the melting temperature of the work-the deformation is signicantly less than conventionaltechnique. Friction stir spotwelding (FSSW) is a variantn stir welding and is schematically illustrated in Fig. 1.

umable rotating tool is plunged into the workpieces topon reaching thedesiredplungedepth, the rotating toolat position for a pre-determined nite time (sometimesas dwell period). Subsequent to that, the rotating toolacted from the welded joint leaving behind a friction

ding author. Tel.: +81 78 803 6153; fax: +81 78 803 6155.ress: hirasawa@kobe-u.ac.jp (S. Hirasawa).

stir spot weld. FSSW can be considered as a transient process dueto its short cycle time (usually a few seconds). During FSSW, toolpenetration and the dwell period essentially determine the heatgeneration, material plasticization around the pin, weld geometryand therefore mechanical properties of the welded joint. The toolhas two distinct parts the shoulder and the pin. The shoulder gen-erates bulk of the frictional or deformational heat. The pin assists inmaterial ow between the workpieces. Features are incorporatedon the shoulder andpin surfaces to further aid these functionalities.If process conditions for the welds are not optimized, the resultingjoint may contain weld defects. Presently, optimum welding con-ditions are determined in many experiments and a numerical sim-ulation code to calculate optimum welding conditions is desired.

Many investigators have tried to simulate friction stir weldingprocess. Langerman andKvalvik (2003) and Colegrove and Shercliff(2004) analyzed the temperature distribution and the ow of met-als by using the computational uid dynamic code FLUENT withnon-Newtonian uid model. Drer (2008) analyzed the metalow during the welding of different materials by using the com-putational uid dynamic code COMSOL with the level set method.When the uid model is used, it is difcult to approximate metalproperties of the plastic deformation behavior. Khandkar and Khan(2003), McCune et al. (2004), Schmidt and Hattel (2005), Zhangand Zhang (2008, 2009), and Rajamanickam et al. (2009) analyzedthe temperature distribution and the plastic deformation by using

see front matter 2010 Elsevier B.V. All rights reserved.jmatprotec.2010.04.003is of effect of tool geometry on plastic g using particle method

irasawaa,, Harsha Badarinarayanb, Kazutaka OkaKawanamia

f Mechanical Engineering, Kobe University, 1-1 Rokkodai-machi, Nada-ku, Kobe-shi, Hica, Ltd., 34500 Grand River Ave., Farmington Hills, MI 48335, USArch Laboratory, Hitachi, Ltd., 7-1-1 Omika, Hitachi, Ibaraki, 319-1292, Japanf Advanced Mechanical Systems, Kumamoto University, 2-39-1 Kurokami, Kumamoto-

e i n f o

nuary 2010vised form 2 April 2010ril 2010

a b s t r a c t

The effect of tool geometry on the p(FSSW) is investigated using the parttool with at shoulder, the model prdirection and the material is pushedgeometry. Other pin geometries evalushoulder, and concave shoulder. Withto predict thematerial ow for spotwocate / jmatprotec

during friction stir spot

toc, Toshio Tomimurad,

en 657-8501, Japan

mamoto 860-8555, Japan

ow and material mixing during friction stir spot weldingethod approach. For spot welds made with a cylindrical pinthe material ow at the pin periphery to be in the upwardward beneath the shoulder giving rise to the resultant hooknclude taperedpin, inverse taperedpin, triangular pin, convexcorrelation with experimental trials, this model is then usedhematerial ow, and thereby the resultant hook formation, is

1456 S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463

Fig. 1. Schem(b) vertical cro

the nite etion model.very high svoid formaal. (2006) anThe presenthe temperaanalyzed thfriction stiranalyze larrates. Badarresults of teffect of too

In the wthe plastic plastic owusing the ping the effeduring FSSWthe optimu

2. Analytic

A two-sanalysis. Fircondition (oity). Then, umaterial is

In ordersingle sheethickness oder diametlength of 1.other pin gand inversewere also mtool rotatioThe calculadistribution

alculaction.

ereden thpera

nduc

t+ u

x, y,are tg/m3

K)),assl andatione tooheaatic view of friction stir spot welding (FSSW). (a) Isometric view andss-section.

lement code ABAQUS with an elasticplastic deforma-It is necessary to analyze the plastic deformation withtrain rates. Schmidt and Hattel (2005) analyzed thetion in metal at non-successful condition. Zahedul etalyzed the residual thermal stresses in welded metals.

t authors (Hirasawa et al., 2005, 2006, 2009) analyzedturedistributionby thenite elementmethod, andalsoe plastic ow of material by the particle method duringwelding process. The particle method is preferable to

ge plastic deformation behavior with very high straininarayan et al. (2007, 2009a,b) reported experimentalhe temperature distribution, the ow pattern and thel geometry on metallurgical bond for FSSW.ork reported here, the temperature distribution andow of material are analyzed during FSSW process. Theis analyzed with the elasticplastic deformation model

Fig. 2. Ccross-se

considbetwe

Temmal co

c

(T

whereand wsity (k(W/(mHeat isthe toothe locnear thsurfacearticle method. Numerical analysis aids in understand-ct of different tool geometries on the material ow. The outcomeof this researchwill help in determining

m tool geometry for this process.

al method

tep approach is undertaken in the current numericalst, the temperature distribution is computed at a givenf tool geometry, plunge force and material ow veloc-sing this temperature distribution, the plastic ow of

analyzed.to simplify the modeling, the welding was done on a

t and hence called as Spot-on-Plate type welding. Thef the workpiece is 2.5mm. The FSSW tool has a shoul-er of 12mm, cylindrical pin diameter of 5mm and pin25mm. The shoulder prole is at. In addition to this,eometries evaluated were: triangular pin, tapered pin,tapered pin. Also, concave and convex shoulder prolesodeled. The process conditions used for welding were:nal speed 1500 rpm, plunge rate 3.3mm/s for 0.44 s.tion conditions for the numerical model of temperatureare shown in Fig. 2. The dimension of the workpiece

q0 = dpNwhere d is(=0.6 insure (Pa), aforce of thobtained frthe contacttial temperfor temperaperiphery obe 25 C. AOther bounin the air gThe workpi180W/(mK(1) is calculchange is caof 0.44 s.

The plasperature diworkpieceregionof thtion model of temperature distribution. (a) Top view and (b) vertical

is 100mm100mm2.5mm. There is 0.1mm air gape workpiece and the anvil.ture distribution is calculated with the unsteady ther-tion equation.

T

x+ vT

y+ w T

z

)=

(2T

x2+

2T

y2+

2T

z2

)+ q (1)

and z are the coordinates (m), t is the time (s), u, v,he ow velocity (m/s), T is temperature (C), is den-), c is specic heat (J/(kgK)), is thermal conductivityand q is heat generation rate per unit volume (W/m3).umed to be generated at the contact surface between

the workpiece. It has been reported that the effect ofof heat generation region on temperature distributionl is not signicant (Hirasawa et al., 2005). Local contactt generating rate q0 (W/m2) is calculated by Eq. (2).(2)

diameter (m), is the dynamic friction coefcientthis work, Tomimura et al., 2006), p is contact pres-nd N is rotating speed (N= (1500/60) round/s). Plungee tool is assumed to be constant 1400N, which wasom experimental data (Badarinarayan et al., 2007), andpressure p is calculated from the plunge force. The ini-ature of the workpiece is 25 C. Boundary conditionsture analysis are as following. The temperature of thef the workpiece, and top end of the tool are xed tolso, the temperature of the anvil is xed to be 25 C.dary conditions are adiabatic. The thermal conductionap between the workpiece and the anvil is computed.ece is aluminum alloy A6061 with thermal conductivity), density 2700kg/m3 and specic heat 896 J/(kgK). Eq.ated using the nite element method. The temperaturelculatedwith the time step of 0.044 s until thenal time

tic ow of material is analyzed with the obtained tem-stribution. The calculation region of plastic ow in theis assumed to be 18mm in diameter. The calculationeplasticow is smaller than that of the temperaturedis-

S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463 1457

Fig. 3. Initial condition of particles in vertical cross-section.

tribution shown in Fig. 2, because the plastic ow region is smallerthen the temperature diffusion region. The plastic ow is calcu-lated with the elasticplastic deformation equations shown in Eqs.(3) and (4).

2rt2

= x

(E

(1 + )(1 2) +E

(1 + ))

+ K (3)

=12

(ux

+ ux

)(4)

where r isx is the spsity (kg/m3

is the stnal body foelasticplasticle methoassumed toand the momaterial isuij is calcula

uij = (rj riwhere ri ismatrix, andshear stress

nij =E

(1 +

sij =E

(1 + Plastic defostress exceeparticle me(Chikazawamethod is apiece mateas shown inputed. Thes

Fig. 4. S

Fig. 5. Yield stress of workpiece.

These particles are interconnected in the normal direction and thetangential direction.

Fig. 3 shows the initial condition of particles in vertical cross-section. Theworkpiece is comprised of 53,000 particleswith a pitch

mm. Red particles are in the plastic ow region. Blue par-re those on the tool or those that represent xed positions. Blaper spedn ofnvilt surrkpi, andal temf paral timle-lo

culat

anda

stanm inhowrkpieistrin toosymmr, antempiece tthe coordinates of the material (m), t is the time (s),ace coordinates in Cartesian system (m), is the den-), E is the Youngs modulus (Pa), is the Poisson ratio,rain tensor, is the Kronecker delta, K is the exter-rce (N/m3), and u is the deformation vector (m). Thetic deformation equations are analyzed using the par-d. In the particle method, the workpiece material isbe composed of several particles, as shown in Fig. 3,vements of these particles are computed. When the

expressed with many particles, the deformation vectorted by Eq. (5).

) R(r0j r0i ) (5)

the coordinates of the particle (m), R is the rotationsufx 0 is the initial condition. Normal stress n

ij, and

sijis obtained by Eqs. (6) and (7).

)nij (6)

)sij (7)

rmation is assumed tooccurwhen themaximumtensileds the yield stress. These equations are analyzed by thethod with the moving-particle semi-implicit methodand Koshizuka, 2001; Kondo et al., 2006). The particlemeshless method. In the particle method, the work-

rial is assumed to be composed of several particles,Fig. 3, and the movement of these particles is com-

eparticles are essential springmodels as shown inFig. 4.

of 0.25ticles aregionthe upoverlappositioof the acontacThe wo65GPathe locment othe nthe tab

3. Cal

3.1. St

The1.25mFig. 6 sthe woature drotatioareaxicirculametricworkppring model. (a) Normal direction and (b) tangential direction. Fig. 6. Teck particles show the horizontal position at 1mm fromurface, which will be compared with contact surface oftwo workpieces in experiment. In the calculation, the

the moving tool is given at each time step. The positionsand the periphery of the workpiece are xed. Slip at thefaces between the tool and the workpiece is neglected.ece is aluminum alloy A6061 with the Youngs modulusthe Poisson ratio 0.3. The yield stress is a function ofperature as shown in Fig. 5 (Chao andQi, 1998).Move-

ticles is calculated with the time step of 0.00002 s untile of 0.44 s. The computation time is shortened by usingok-up procedure (Aratani, 1995).

ion results

rd cylindrical pin

dard cylindrical pin of the tool is 5mm in diameter andlength. The shoulder diameter of the tool is 12mm.

s the calculation result of temperature distributions ince at time 0.44 s for a standard cylindrical pin. Temper-

bution is axisymmetric and the temperature below thel is 340 C at 0.44 s. Heat transfer phenomena in FSSWetric.Materialowcausedby the tool rotation isnearly

d the effect of the circular ow velocity on the axisym-eraturedistribution is small. Fig. 7 shows thecalculatedemperature prole just below the tool and at a distancemperature distribution of workpiece at 0.44 s for cylindrical pin.

1458 S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463

Fig

of 2.5mm (circumferenin temperatthe shouldelocal pressuin that regiothermal conthat instantregionof thbe high enotemperaturthe obtaine

Fig. 8 shodifferent timparticles arwas 1mmFig. 3. The mpredicts thedirection. N beneath tthe force acperiphery tof the matematerial show is simi(Badarinaravisual of th

Fig. 10 sparticles shat 1mm frosponds to taverage poPartial metthe overlap2009a). Theresulting horesult.

Fig. 11 shparticles atsignies mathe pin.

3.2. Tapere

Plastic (Fig. 12). That the free e

hange of positions of particles (left) and ow pattern (right) for cylindrical.13 s; (b) 0.26 s; (c) 0.35 s and (d) 0.44 s.

Vertical cross-section of FSSW weld specimen for cylindrical pinarayan et al., 2007).. 7. Temperature change of workpiece for cylindrical pin.

near the pin circumference), 6mm (near the shoulderce), and 8mm from the tool axis. There is a slight dropure below the shoulder surface at 0.35 s, this is becauser comes in contact with the workpiece and hence there decreases to about one-seventh of its original valuen for a constant plunge force of the tool. However, theduction heat ux in the workpiece is almost same at. This drop in temperature is not observed outside theeshoulderdiameter. Temperaturebelowthe tool shouldugh to soften the workpiece but less than the meltinge. Plastic owandmixing ofmaterial are analyzedusingd temperature distribution.ws the instantaneous snap-shot of the material ow ate instances during the tool plunging phase. The black

e an indication of the original particle position thatfrom the upper surface of the workpiece, as shown inaterial ow shows particle velocity vectors. The modelmaterial ow at the pin periphery is in the upward

ear the shoulder, there are two ow patterns observedhe shoulder, the material is pushed downward due toting from the shoulder face, whereas on the shoulderhematerial ows upward and outward due to extrusionrial that is caused by the shoulder plunge. This extrudedows up on the specimen surface as burr. The materiallar to that observed in experimental trial shown in Fig. 9yan et al., 2007). Thus, this modeling technique gives ae complex material mixing phenomena during FSSW.hows the numerical results of the average position ofown in black particles in Fig. 8(d) (that were originallym the upper surface of the workpiece), and it corre-he contact surface of overlapped two workpieces. Thesition has a round upward bending (called as hook).

Fig. 8. Cpin. (a) 0

Fig. 9.(Badarinallurgical bond observed in the weld region betweenped metal sheets in experimental (Badarinarayan et al.,numerical result shown in Fig. 10 explicitly shows thisok geometry which closely matches the experimental

ows the standarddeviationofmovement of these blacktime 0.44 s. The standard deviation of the movementterial mixing. The mixing is large near the periphery of

d pin

ow of material is calculated for a tool with tapered pine pin diameter at the root of the pin is 5mm whereasnd it is 4mm. Pin length is 1.25mm.

Fig. 10. Average position of particles that were originally at 1mm from upper sur-face of workpiece at 0.44 s for cylindrical pin.

S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463 1459

Fig. 11. Standard deviation of movement of particles at 0.44 s for cylindrical pin.

Fig.

Fig. 13 sat 0.44 s. FigThe ow isaverage posupper surfais smaller tstandard deof the stand

Fig

Fig. 15. Average position of particles that were originally at 1mm from upper sur-face of workpiece at 0.44 s for tapered pin.Fig. 12. Tapered pin tool geometry.

13. Vertical positions of particles at 0.44 s for tapered pin.

hows the calculation result of the positions of particles. 14 shows total material ow pattern from 0s to 0.44 s.along the tapered surface of the pin. Fig. 15 shows theition of particles (that were originally at 1mm from thece of theworkpiece) at time 0.44 s. The hook formationhan that for the standard cylindrical pin (Fig. 10). Theviation of these particles at time 0.44 s is similar to thatard cylindrical pin (Fig. 11).

. 14. Total ow pattern from 0s to 0.44 s for tapered pin.

Fig. 17. Vertic

3.3. Inverse

Plastic tapered pinwhereas at

Fig. 17 sh0.44 s after

Fig. 18Fig. 16. Inverse tapered pin tool geometry.

al positions of particles at 0.44 s after plunge for inverse tapered pin.

tapered pin

ow of material is calculated for a tool with inverse(Fig. 16). The pin diameter at the root of the pin is 5mmthe free end it is 6mm. Pin length is 1.25mm.ows the calculation result of thepositionsofparticles atplunge. Fig. 18 shows total material ow pattern from

. Total ow pattern from 0s to 0.44 s for inverse tapered pin.

1460 S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463

Fig. 19. Averaface of workpi

0 s to 0.44 swere origintime 0.44 s.cylindrical0.44 s is sim

3.4. Convex

Plasticder (Fig. 20angle of thethe pin leng

Fig. 21 sh0.44 s afterpto 0.44 s. Figoriginally a0.44 s. Thereto this toolparticles atthat the resof these spostandard cy

Fig. 21. Vertic

Fig. 22. Total ow pattern from 0s to 0.44 s for convex shoulder.

Average position of particles that were originally at 1mm from upper sur-orkpiece at 0.44 s for convex shoulder.

ncave shoulderge position of particles that were originally at 1mm from upper sur-ece at 0.44 s for inverse tapered pin.

Fig. 20. Convex shoulder tool geometry.

. Fig. 19 shows the average position of particles (thatally at 1mmfrom theupper surface of theworkpiece) atThe hook formation is similar to that for the standardpin. The standard deviation of these particles at timeilar to that of the standard cylindrical pin (Fig. 11).

Fig. 23.face of w

3.5. Coshoulder

owofmaterial is calculated for a toolwith convex shoul-). The length of the convex shoulder is 0.5mm (inclinedface is 10), the diameter of cylindrical pin is 5mm, andth is 0.75mm.ows the calculation result of thepositionsofparticles atlunge. Fig. 22 shows totalmaterial owpattern from0s. 23 shows the average position of particles (that were

t 1mm from the upper surface of theworkpiece) at timeisno hook formation in theweld regiongenerateddue

geometry. Fig. 24 shows the standard deviation of thesetime 0.44 s. This magnitude is quite small indicatingultant material mixing is poor. Therefore the strengtht welds is expected to be lower than those made withlindrical pin.

al positions of particles at 0.44 s after plunge for convex shoulder.

Plastic shoulder (F(inclined an5mm, and

Fig. 26 scles at 0.44material oUpward mabeneath th

Fig. 24. Standow of material is calculated for a tool with concaveig. 25). The length of the concave shoulder is 0.5mmgle of the face is 10), the diameter of cylindrical pin is

the pin length is 1.75mm.hows the calculation result of the positions of parti-s after plunge. Fig. 27 shows positions of particles andw pattern at 0.35 s during the tool plunging phase.terial ow at the pin periphery and downward ow

e shoulder are stronger than the standard cylindrical

ard deviation of movement of particles at 0.44 s for convex shoulder.

S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463 1461

Fig. 25. Concave shoulder tool geometry.

Fig. 26. Vertical positions of particles at 0.44 s after plunge for concave shoulder.

Fig. 27. Positishoulder.

pin (Fig. 8)were originat time 0.4sharp hooktact surfacestrength (Bdeviation oupper surfastandard decylindrical pa signicanstatic stren

3.6. Triangu

Fig. 30 sgeometry. Ta concave sthe calculat

Fig. 28. Averaface of workpi

Fig. 29. Stand

Fig. 30. Tool geometry of triangular pin with concave shoulder.ons of particles (left) and ow pattern (right) at 0.35 s for concave

. Fig. 28 shows the average position of particles (thatally at 1mm from the upper surface of the workpiece)4 s. There is a sharp hook formation observed. The formation prevents failure propagation along the con-of overlapped two workpieces and increases welding

adarinarayan et al., 2009a). Fig. 29 shows the standardf the particles (that were originally at 1mm from thece of the workpiece) at time 0.44 s. The magnitude ofviation is smaller than that predicted for the standardin indicating that less mixing. Tool shoulder prole hast inuence on the material mixing and eventually thegth.

lar pin with concave shoulderhows the schematic illustration of the triangular pinhe circumscribed pin diameter is 5mm. The tool hashoulder with a 10 angle of concavity. Fig. 31 showsion result of temperature changes of the workpiece just

ge position of particles that were originally at 1mm from upper sur-ece at 0.44 s for concave shoulder.

Fig

below the taxis. Theretriangular ppin (Fig. 7).

Fig. 32 sthe verticalresult of th

Fig. 32. Verticder.ard deviation of movement of particles at 0.44 s for concave shoulder.. 31. Temperature change of workpiece for triangular pin.

ool and at a distance of 2.5, 6, and 8mm from the toolis more heat generation by plastic deformation near thein and temperature is larger than that of the cylindrical

hows the calculation result of the particle positions incross section at 0.44 s. Fig. 33 shows the calculation

e particle positions in the horizontal cross-section and

al positions of particles at 0.44 s for triangular pinwith concave shoul-

1462 S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463

Fig. 33. Changtriangular pin.

material o0.436 s. Thepin. At theoutside andcorners of tupward dirhorizontal cpin (markeAt the advaas B), theof the corninward dirematerial.

Fig

Fig. 35. Average position of particles that were originally at 1mm from upper sur-face of workpiece at 0.44 s for triangular pin with concave shoulder.

Standard deviation of movement of particles at 0.44 s for triangular pin.

35 shows the average position of particles (that were orig-e of horizontal positions of particles (left) and ow pattern (right) for(a) 0.429 s and (b) 0.436 s.

w pattern in the vertical cross section at time 0.429 andhorizontal particle positions are shown at the tip of thecorner of the triangular pin (Fig. 33(b)), the ow is thedownward direction. At the cut portion between the

he triangular pin (Fig. 33(a)), the ow is the inside andection. Fig. 34 shows the material ow pattern in theross-section at 0.436 s. At the corner of the triangular

d as A in Fig. 34), the ow is in the outward direction.

Fig. 36.

Fig.ncing side of the corner of the triangular pin (markedow is along the direction of rotation. At the backsideer of the triangular pin (marked as C), the ow is thection. This ow pattern enhances the mixing of metal

. 34. Horizontal ow pattern at 0.436 s for triangular pin.

inally at 1m0.44 s. Therthe standarface at timepin is greatat the cornsection of eal., 2009a),et al. (2009pin with cocylindrical

Fig. 37. Vert(Badarinarayam from the upper surface of the workpiece) at timee is a sharp hook formation near the pin. Fig. 36 showsd deviation of movement of particles at the contact sur-0.44 s. This results shows that the mixing around the

ly enhanced. The strong material ow and mixing occurer of the triangular pin. Fig. 37 shows vertical cross-xperiment result for a triangular pin (Badarinarayan etshowing the suppressed hook geometry. Badarinarayanb) reported the weld strength for a tool with triangularncave shoulder is stronger than that for the standardpin. Experimental results showed that welds made with

ical cross-section of FSSW weld specimen for triangular pinn et al., 2009a).

S. Hirasawa et al. / Journal of Materials Processing Technology 210 (2010) 14551463 1463

the cylindrical pin have a continuous hook which bypasses the stirzone and terminates close to the keyhole. By contrast, for weldsmadewith the triangular pin, the hook is directed upward and thenarrested at the periphery of the stir zone. The difference in the hookshape can be attributed to thematerial ow. The calculation resultspresented here is in line with experimental trials. A triangular pinwith a concave shoulder is preferred for friction stir spot weld-ing as this conguration results in welds with a suppressed hookgeometry yielding high strength welds.

4. Summary

The effect of tool geometry on plastic owandmixing ofmateri-als during friction stir spot welding process (FSSW) is investigatednumerically using the particle method.

1. Temperature distribution is axisymmetric and the tempera-ture below the rotating tool is 340 C at 0.44 s for the standardtool with a cylindrical pin of 5mm diameter and pin length of1.25mm rotating at 1500 rpm.

2. For a cylindrical pin tool, the material ow at the pin periph-ery is in the upward direction. Near the shoulder, there are twoow patterns observed beneath the shoulder, the material ispusheddownwarddue to the force acting fromthe shoulder face,whereas on the shoulder periphery the material ows upwardandoutwarddue toextrusionof thematerial that is causedby theshoulderthe hookon the ho

3. The trianenhancedpressingmade wresults in

4. The friction themstrength

References

Aratani, T., 199tion. In: Co

Badarinarayanduring fricof the 200

Badarinarayan, H., Yang, Q., Zhu, S., 2009a. Effect of tool geometry on static strengthof friction stir spot-welded aluminum alloy. International Journal of MachineTools & Manufacture 49, 142148.

Badarinarayan, H., Shi, Y., Li, X., Okamoto, K., 2009b. Effect of tool geometry on hookformation and static strength of friction stir spot-welded aluminum 5754-Osheet. International Journal of Machine Tools & Manufacture 49, 814823.

Chao, Y.J., Qi, X., 1998. Thermal and thermo-mechanical modeling of friction stirwelding of aluminum alloy 6061-T6. Journal of Materials Processing & Manu-facturing Science 7 (10), 215233.

Chikazawa, S., Koshizuka, S., 2001. A particle method for elastic and visco-plasticstructures anduid-structure interactions. ComputationMechanics 27, 97106.

Colegrove, P.A., Shercliff, H.R., 2004. Modeling the friction stir welding of aerospacealloys. In: Proceedings of the 5th International FSW Symposium.

Drer, S.M., 2008. Advanced modeling of friction stir welding improved materialmodel for aluminum alloys and modeling of different materials with differ-ent properties by using the level set method. In: Proceedings of the COMSOLConference 2008, Hannover.

Hirasawa, S., Okamoto, K., Hirano, S., Tomimura, T., 2005. Combined analysis ofplastic deformation ow and temperature distribution during friction stir weld-ing. In: Proceedings of the 2005 ASME International Mechanical EngineeringCongress and Exposition, IMECE2005-79328.

Hirasawa, S., Okamoto, K., Hirano, S., Tomimura, T., 2006. Analysis of temperaturedistribution and plastic deformation ow during spot friction stir welding. In:Proceedings of the 13th International Heat Transfer Conference, MNF-03.

Hirasawa, S., Badarinarayan, H., Okamoto, K., Tomimura, T., Kawanami, T., Hirano,S., 2009. Analysis of temperature and plastic ow during friction stir spot weld-ing using particle method. Journal of Thermal Science and Technology 4 (2),260271.

Khandkar, M.Z.H., Khan, J.A., 2003. Predicting residual thermal stresses in frictionstir welding. In: Proceedings of the 2003 ASME International Mechanical Engi-neering Congress and Exposition, IMECE2003-55048.

Kondo,M., Koshizuka, S., Suzuki, Y., 2006. Application of simplectic scheme to three-dimensional elastic analysis using MPS method. Transactions of JSME 72 (716),425431 (in Japanese).

an, Me eldmal E, R.W.the

rnatioickamt of pion sti, H., Hion stiing 13ra, T.,mic cings o, M., Kesiduessing., Zhas in fr270.., Zhanion stiplunge. Thismaterial owattributes to the formationof geometry. Numerical results showing the predictionok formation agree well with the experimental data.gular pin tool, due to its inherent geometry, showsmaterial ow. This enhances ow also helps in sup-

the upward rising hook geometry (as seen in weldsith cylindrical pin). A triangular pin with a concavespot welds with high strength.

on stirwelding tool geometryhas a signicant inuenceaterialmixingandhenceeventually inuences the staticof resultant spot welds.

5. High-speed computing algorithm for molecular dynamic simula-mputational Mechanics95, vol. 2, pp. 18521857., H., Hunt, F., Okamoto, K., Hirasawa, S., 2007. Study of plunge motiontion stir spotwelding temperature andowpattern. In: Proceedings7 The Minerals, Metals & Materials Society Annual Meeting.

LangermaturTher

McCunewithInte

RajamanEffecfrict

Schmidtfrictneer

Tomimudynaceed

Zaheduling rProc

Zhang, Zeter241

Zhang, Zfrict., Kvalvik, E., 2003. Modeling plasticised aluminum ow and temper-s during friction stir welding. In: Proceedings of the 6th ASMEJSMEngineering Joint Conference, TED-AJ03-133., Ou, H., Armstrong, G., Price, M., 2004. Modeling friction stir weldingnite element method, a comparative study. In: Proceedings of the 5thnal FSW Symposium., N., Balusamy, V., Madhusudhann, Reddy, G., Natarajan, K., 2009.rocess parameters on thermal history and mechanical properties ofr welds. Materials and Design 30 (7), 27262731.attel, J., 2005. A local model for the thermomechanical condition inr welding. Modelling and Simulation in Materials Science and Engi-, 7793.Okita, K., Hirasawa, S., 2006. Experimental evaluation of apparentoefcient of friction between rotating rod and metal plate. In: Pro-f the 2006 Materials Science & Technology.handkar, H., Khan, J.A., Reynolds, A.P., Sutton, M.A., 2006. Predict-

al thermal stresses in friction stir welded metals. Journal of MaterialsTechnology 174, 195203.

ng, H.W., 2008. Numerical studies on controlling of process param-iction stir welding. Journal of Materials Processing Technology 209,

g, H.W., 2009. Numerical studies on the effect of transverse speed inr welding. Materials and Design 30, 900907.

Analysis of effect of tool geometry on plastic flow during friction stir spot welding using particle methodIntroductionAnalytical methodCalculation resultsStandard cylindrical pinTapered pinInverse tapered pinConvex shoulderConcave shoulderTriangular pin with concave shoulder

SummaryReferences