Experimental and numerical investigation on under-water friction stir welding of armour grade AA2519-T87 aluminium alloy

S.SREE SABARI，S.MALARVIZHI，V.BALASUBRAMANIAN，*，G.MADUSUDHAN REDDY

aCentre for Materials Joining and Research （CEMAJOR），Department of Manufacturing Engineering，Annamalai University，Annamalai Nagar，608 002 Tamil

Nadu，India

bMetal Joining Section，Defense Metallurgical Research Laboratory （DMRL），Hyderabad，India

Experimental and numerical investigation on under-water friction stir welding of armour grade AA2519-T87 aluminium alloy

S.SREE SABARIa，S.MALARVIZHIa，V.BALASUBRAMANIANa，*，G.MADUSUDHAN REDDYb

aCentre for Materials Joining and Research （CEMAJOR），Department of Manufacturing Engineering，Annamalai University，Annamalai Nagar，608 002 Tamil

Nadu，India

bMetal Joining Section，Defense Metallurgical Research Laboratory （DMRL），Hyderabad，India

Friction stir welding （FSW）is a promising welding process that can join age hardenable aluminium alloys with high joint eff i ciency.However，the thermal cycles experienced by the material to be joined during FSW resulted in the deterioration of mechanical properties due to the coarsening and dissolution of strengthening precipitates in the thermo-mechanical affected zone （TMAZ）and heat affected zone （HAZ）.Under water friction stir welding （UWFSW）is a variant of FSW process which can maintain low heat input as well as constant heat input along the weld line.The heat conduction and dissipation during UWFSW controls the width ofTMAZ and HAZ and also improves the joint properties.In this investigation，an attempt has been made to evaluate the mechanical properties and microstructural characteristics of AA2519-T87 aluminium alloy joints made by FSW and UWFSW processes.Finite element analysis has been used to estimate the temperature distribution and width of TMAZ region in both the joints and the results have been compared with experimental results and subsequently correlated with mechanical properties.

Aluminium alloy；Friction stir welding；Underwater friction stir welding；Mechanical properties；Microstructural characteristics；Finite element

analysis

1.Introduction

In recent days armour grade steels in the military vehicle is replaced by lighter materials for better mobility ［1，2］.Aluminium alloy AA2519-T87 is a potential candidate，to replace steel in a few defence applications due to its high speci fi c strength.In addition，it has excellent tensile properties，fracture toughness properties and ballistic immunity.Researchers confi rmed that friction stir welding （FSW）can produce sound joints in heat-treatable aluminium alloys ［3-7］.However，the thermal cycles exerted during FSW resulted in the reduction of mechanical properties of the joints due to coarsening and dissolution of strengthening precipitates in the thermo mechanically affected zone （TMAZ）and heat affected zone （HAZ）of the joints ［8，9］.

In most of the age hardenable aluminium alloys，TMAZ and HAZ are identif i ed as the weak regions ［10-13］.The strength of the parent metal and weld joints of aluminium alloys depend on the presence of the precipitates.The thermal conditions prevailed during heating and cooling cycle of welding affect the precipitation behaviour.In FSW，the heat generated in the stir zone and heat conducted to TMAZ and HAZ causes the precipitates to be either solutionized or coarsened.The loss of precipitates during solutionizing and coarsening of precipitates resulted in poor joint properties ［14］.In addition，the high thermal conductivity of aluminium creates variation in the thermal gradient and heat build ups along the longitudinal direction，especially in joining of thick plates.It results in non-uniformity along the weld region.In order to overcome the heat related problems created in TMAZ and HAZ and to improve the strength of FSW joints，the heat should be dissipated readily using a cooling method.The water cooling method is more eff i cient than the other cooling method in terms of uniform cooling and high heat transfer coeff i cient.

Liu et al.carried out an investigation on under water friction stir welding （UWFSW）ofAA2219-T6 aluminium alloy and the investigation concluded that the UWFSW improved the strength of the joint by restricting the coarsening and dissolution of strengthening precipitates ［15］.Fu et al.also observed the improvement in the tensile strength of the UWFSW joints of AA7075 T87 aluminium alloy by reducing the width of HAZ ［16］.Fratini et al.simulated the temperature histories of UWFSW process and f i nds that the water cooling resulted in reduction of thermal f l ow adjacent to the tool［17］.Huijie et al. developed mathematical modelling and optimization procedures for underwater friction stir welding of a heat-treatable aluminium alloy.The optimization result indicated that a maximum tensile strength of 360 MPa can be obtained through UWFSW，which is 6%higher than the maximum tensile strength obtained in FSW ［18］.From the literature it is understood that the mechanical properties of the FSW joints could be improved by the water cooling.

Studying the thermal histories of the weld joints is useful to explore the fundamental aspects of UWFSW.In order to map the temperature distribution and to predict the thermal histories of welding processes，f i nite element （FE）method was used by many researchers ［19，20］.However，very few researchers have carried out experimental investigation and f i nite element analysis of UWFSW of aluminium alloys.Moreover，the previous investigations were focused on the microstructural characterization，mechanical properties evaluation and FE analysis of the conventional FSW process ［21，22］.Very few publications are available on the comparison of experimental and numerical investigations of FSW and UWFSW of aluminium alloys. Hence，this investigation was carried out to evaluate the mechanical properties of FSW and UWFSW joints of high strength，armour grade AA2519-T87 aluminium alloy.Finite element analysis was also used to estimate temperature distribution during FSW and UWFSW of AA2519-T87 aluminium alloy.Further，width of the TMAZ region was estimated for both the joints （FSW and UWFSW）by FE analysis and compared with the experimental results.

2.Experimental work

Fig.1 shows the experimental details employed for the fabrication of FSW and UWFSW joints.Fig.1（a）shows the schematic representation of the UWFSW process.The workpiece was rigidly clamped on the backing plate in the tank.The water is discharged into the tank through the inlet to the level above the tool shoulder.During welding，the inlet and outlet valves are adjusted to control the water f l ow in such a way to maintain the temperature of water below 60 °C near to the welding location. From the previous investigation，the author found that the taper threaded tool was well suited for joining this material and so it is used for this investigation ［23］.The dimension of the tool is shown in Fig.1（b）.The photographs of the FSW and UWFSW setups are shown in Fig.1（c）and 1d respectively.AA2519-T87 aluminium alloy was used as the parent metal.Joint conf i guration of 150 × 150 × 19 mm was used in this investigation（Fig.1（e））.The chemical composition of the parent metal was conformed using vacuum spectrometer and presented in Table 1.The process parameters and the welding conditions used for fabricating FSW and UWFSW joints are shown in the Table 2.Tensile test was carried out using Instron made servo hydraulic controlled universal testing machine.ASTM E8M-04（Standard test methods for tension testing of metallic materials）guidelines were followed for the extraction （Fig.1（e））and testing of the tensile samples.The transverse tensile properties such as yield strength，ultimate tensile strength and elongation were evaluated from the tensile test.Vickers microhardnesstester was used to measure the microhardness at the various region of the weld joint with a load of 0.5 N and dwell time of 15 s.The sample preparations and the testing procedures for the microindentation were followed as per the guidelines from ASTM E384-99 standard （Standard Test Method for Microindentation Hardness of Materials）.

Fig.1.Experimental details （a）Schematic diagram of UWFSW process （b）Tool dimensions （c）Photographs of FSW setup （d）Photographs of UWFSW setup （e）Joint conf i guration and specimen extraction diagram.

Table 1Chemical composition （wt%）of AA 2519 T87 aluminium alloy.

Table 2Process parameters and tool dimensions used in this investigation.

Fig.2.3-Dimensional meshed model.

Microstructural examination was carried out using an optical microscope.The specimens for metallographic examination were initially polished by rough emery and subsequently polished using different grades of emery papers to get mirror polish.The specimens were etched with modif i ed Keller’s reagent as per the ASTM E407 standard （Standard Practice for Microetching Metals andAlloys）.Transmission electron microscope was used to characterize the microstructure ofTMAZ and HAZ.The samples of 3 mm diameter is extracted and polished to 10 μm thick using ion milling process.Line intercept method was employed to measure the average grain size of different regions of the weld joint.

3.Finite element analysis

This paper presents three-dimensional thermal analysis using the f i nite element （FE）code COMSOL.Dissimilar meshed model of varying size and element types were used in the FE analysis （Fig.2）.The meshed model composed of quadrilateral，triangular and square elements.The model is meshed into 6821 f i nite elements.In order to reduce the computational time and achieve accurate results，f i ner mesh was carried out near the tool and coarser mesh was carried out for the region away from the tool.The input parameters used for FE analysis is tabulated in Table 3.

Table 3Input parameters used in FE analysis.

3.1.Thermal boundary condition

The heat generated during welding is lost due to three modes of heat transfer which greatly reduce the availability of heat required to weld the work piece.In welding，the heat is transferred by conduction，convection and radiation modes.The governing differential equation for three dimensional heat conduction equations for a solid in Cartesian coordinate system is given by ［24］

where thermal diffusivity of the material，qgis the heat generation per unit volume in W/m3，ρ is density of the material in kg/m3，Cpis specif i c heat in J/kgK and k are thermal conductivity W/mK.

where hupand hdownare the convection coef fi cient of top and bottom surface respectively，T0is the reference temperature，Tambis the ambient temperature，σ is the Stefan-Boltzmann constant and ε is the surface emissivity.The natural convection between aluminium and air was experimentally found by Choa et al.as 15 W/（m2°C）for FSW ［26］.Similarly Zhang et al. found the heat transfer coef fi cient of UWFSW at top and bottom as 2000 W/（m2°C）and 1000 W/（m2°C）respectively［27］.Temperature based material properties were considered for the fi nite element analysis.In this investigation，the thermal model was considered for the analysis and the material fl ow behaviour was not accounted.The degree of heterogeneity between the advancing side （AS）and retreating side （RS）mainly depends on the material fl ow and the microstructure rather on the temperature distribution and hence symmetric model was considered for the analysis.The mid-plane of the butt joint was assumed to be a plane of symmetry in the analysis.Zero displacementconditions were used for constraining the butt joint which resembling the complete fi xed fi xture.

3.2.Heat source modelling

The heat generated is concentrated locally and propagates rapidly into subsequent regions of the plates by heat conduction according to Eq.（1）as well as convection and radiation through the boundary.Constant heat fl uxes were applied as the heat source in the tool shoulder-workpiece interface and tool pinworkpiece interface.

Heat generation in shoulder-work piece interface is the function of major elements such as axial load，area subjected to friction，coef fi cient of friction and angular velocity.The heat fl ux can be mathematically expressed by Song and Kovacevic［28］

Fig.3.Stress strain curves.

Fig.4. （a）Macrostructure of FSW joint（b）Macrostructure of UWFSW joint.

where Fnis the normal force （kN），Asis the shoulder’s surface area （mm2），μ is the friction coeff i cient，R is the distance from the center axis of the tool（mm）and ω is the tool rotation speed（rpm）.The mathematical expression of heat generated at the interface of pin and work piece is adapted from Colegrove ［29］

where μ is the friction coeff i cient，rpis the pin radius （mm），ω is to the angular velocity （rad/s），and Y（T）is the average shear stress of the material（N/mm2）.The input parameter used for the FEM simulation of FSW and UWFSW are presented in Table 3.

4.Results

4.1.Tensile properties

Fig.3 shows the superimposed stress strain curves of FSW and UWFSW joints.From the stress strain curves，the transverse tensile properties like yield strength，tensile strength，percentage of elongation and joint eff i ciency of FSW and UWFSW joints were calculated and presented in Table 4.The tensile strength was measured as 452 MPa for parent material，271 MPa for UWFSW joint and 248 MPa for FSW joint.The yield strength was measured as 427 MPa for parent material，223 MPa for UWFSW joint and 198 MPa for FSW joint. UWFSW joint exhibited higher tensile and yield strength than FSW joint.However，both the joints showed lower strength values than the parent material.The joint eff i ciency was calculated by f i nding the ratio of tensile strength of welded joint and tensile strength of the unwelded parent metal.The UWFSW joint yielded a higher joint eff i ciency of 60%and the FSW jointexhibited lower joint eff i ciency of 55%.From the percentage of elongation value，the degree of ductility of the weld joint can be understood.The elongation was measured as 11.2%for parent material and 4.56%for UWFSW joint.

Table 4Transverse tensile properties of parent metal and welded joints.

4.2.Macro and microstructure

Fig.4 shows the cross-sectional macrographs of FSW and UWFSW joints.Defect free trapezoidal shape stir zone is observed in FSW joint （Fig.4（a））.Fig.4（b）shows the macrograph of UWFSW joint consisting defect free stir zone characterized by large sized onion ring formation extended from the mid thickness region （MTR）to the pin inf l uenced region （PIR）.FSW joint reveals a wider TMAZ in bothAS and RS compared to UWFSW joint.The width of RS-TMAZ is measured as 4.1 mm and 1.7 mm for FSW and UWFSW joints，respectively.Fig.5 shows the optical microstructure of parent metal.It comprises of elongated grains of 49 μm （average grain diameter）oriented towards the rolling direction.The stir zone is classif i ed as upper shoulder inf l uenced region （SIR），middle mid-thickness region （MTR）and lower pin inf l uenced region（PIR）and their respective optical micrographs are shown in Fig.6.All the three regions of stir zone reveal f i ne recrystallized equi-axed grains in both the joints.The average grain size of various regions are measured and presented in the Table 5.Thegrain size variation in all the three regions of stir zone is not so signif i cant in FSW joint（Fig.6（a），（c）and （e））.But in UWFSW joint，the grain size is larger in the SIR，relatively smaller in MTR and very f i ne in PIR.Comparing the average grain size of stir zone，UWFSW joint consists of marginally larger grains than FSW joint.

Fig.5.Optical micrograph of parent metal.

Fig.6.Optical micrographs of stir zone at various locations （a）and （b）Shoulder inf l uenced region （SIR）（c）and （d）Mid-thickness region （MTR）（e）and （f）Pin inf l uence region （PIR）.

Fig.7 shows the micrographs ofTMAZ and HAZ of advancing side （AS）and retreating side （RS）.InTMAZ，the grains are pulled upward towards the stir zone in both FSW and UWFSW joints （Fig.7（a）-（d））.But the grains are more lineated towards stir zone in UWFSW than FSW joints.The extent of straining of TMAZ is higher in UWFSW than FSW joint.Figs.7（e）and（f）show the HAZ micrographs in which FSW joint exhibit slightly larger grains than the UWFSW joint.The average grain size of HAZ is measured as 55 μm and 49 μm for FSW and UWFSW joints respectively.The grain size of HAZ of UWFSW is equal to the grain size of the parent metal.

Table 5Average grain size （μm）of various regions of welded joints.

Fig.7.Optical micrographs ofTMAZ and HAZ regions （a）and （b）Advancing side TMAZ （c）and （d）Retreating side TMAZ （e）and （f）Advancing side HAZ（g）and （h）Retreating side HAZ.

Fig.8.Transmission electron micrographs of various regions of weld joints （a）Parent metal（b）RS-TMAZ of FSW joint （c）RS-TMAZ of UWFSW joint（d）RS-HAZ of FSW joint （e）RS-HAZ of UWFSW joint.

The precipitation behaviour of TMAZ and HAZ are alone analysed with the transmission electron microscopy （TEM）. Fig.8 shows the TEM images of various regions of FSW and UWFSW joints.Fig.8（a）shows the TEM image of parent metal，comparing needle like precipitates normally oriented to each other.The precipitates are densely distributed and f i ner in size （of 50 nm in diameter）.These Al-Cu alloys are strengthened by the presence of θ‘-CuAl2precipitates ［30-32］.TMAZ of FSW joint exhibits precipitate free zone （PFZ）and very few coarsened precipitates of 500 nm in size （Fig.8（b））.Similarly，TMAZ of UWFSW joint exhibits precipitate free zone （PFZ）and dislocation cell structures （Fig.8（c））.The presence of dense dislocation refers the extent of work hardening in theTMAZ.In FSW joint，HAZ has undergone coarsening of precipitates and reduction in the number of precipitates （Fig.8（d））.Compared to the parent metal，the precipitates in HAZ are transformed from θ’precipitates to large sized equilibrium θ precipitates of 200 nm in size.In UWFSW joint，the size and distribution of precipitates are almost similar to the parent metal but very few precipitates are coarsened.The XRD analysis was made to identify the evolution of the precipitates during FSW and UWFSW joints （Fig.9）.TheAl and CuAl2phases are observed in the parent metal，F(xiàn)SW and UWFSW joints.The new phase Al4Cu9is observed in the FSW and UWFSW joints with signif i cant intensities.

4.3.Microhardness

Fig.9.XRD results.

Fig.10 shows the microhardness plot for FSW and UWFSW joints.The stir zone shows f l uctuating hardness values ranging from 108 HV to 131 HV in UWFSW joint.Similarly the hardness values are varying from 108 HV to 120 HV in the stir zone of FSW joint.A drop in hardness value is observed in the TMAZ.In both joints the RS-TMAZ exhibit lower hardness and therefore this region is termed as lower hardness distribution region （LHDR）.The LHDR of FSW and UWFSW joints exhibited lower hardness of 90 HV and 94 HV respectively.The microhardness plots shows increasing trend from TMAZ to HAZ.The increment in the hardness value is gradual from theTMAZ to HAZ in FSW joint.But，an abrupt increment is observed in UWFSW joint.The HAZ exhibit hardness values of 132 HV and 144 HV in FSW and UWFSW joints respectively. The parent metal region recorded maximum hardness of 158 HV.The higher microhardness values are recorded in the stir zone and HAZ of UWFSW joint than the FSW joint.

Fig.10.Microhardness plots of the joints （a）FSW joint（b）UWFSW joint.

Fig.11.Fractographs of the joints （a）FSW joint（b）UWFSW joint.

4.4.Fracture surface analysis

The fractograph of UWFSW joint shows f i ne populated dimples with uniform size oriented towards the loading direction （Fig.11）.In contrast，F(xiàn)SW joint has both small and large size dimples.The secondary cracks are observed inside the dimples.From the fractographs，it could be inferred that both the joints failed under ductile mode.The TMAZ near weld periphery undergone high grain deformation，lower hardness and fracture localization.

4.5.Finite element （FE）analysis

Fig.12.Simulated temperature contour at top surface of the joints （a）FSW joint （b）UWFSW joint.

Fig.12 illustrates the top surface temperature contours for FSW and UWFSW joints.In FSW joint，the temperature distributed to the leading and trailing edge of the tool along the longitudinal direction seemed to be asymmetrical.Symmetric temperature distribution is observed along the longitudinal direction for the UWFSW joint.Fig.13 illustrates the thermal contours in the cross-section of the weld.In FSW joint，the temperature experienced by SIR，MTR and PIR are almost same around 525 °C （Fig.13（a））.But in UWFSW joint，there exists a variation in SIR，MTR and PIR.Maximum temperature of 545 °C is witnessed in the SIR and gradually reduced towards PIR of 512 °C （Fig.13（b））.This is clearly evident from the colour contours which show intense blue colour in the SIR and dull blue colour in the PIR.Similar to the longitudinal temperature contour map，the isothermals of UWFSW joint also remarkably move towards the tool axis in contrast to the FSW joint.The entire stir zone of FSW experienced same amount of heat.In UWFSW joint，the maximum temperature is at the SIR and the temperature is decreasing towards the PIR. TMAZ，theregion next to stirzoneexhibitsbandoftemperature values f l uctuating around 400 °C as shown in light blue colour contour.The material experienced heat in this region is softened and deformed by the stirring action of tool.The FSW joint have wider range of temperature values than the UWFSW.Fig.13（c）shows the temperature plot along the transverse direction.The maximum temperature is not recorded at the weld center but it is observed 25 mm away from the center line，the periphery of shoulder-work piece interface region.

5.Discussion

5.1.Thermal analysis

During tool traverse，the leading edge of the tool gets contacted with the cold zone which was generally referred as preheat zone.The high yield strength of the cold zone offers high resistance against the material f l ow.Due to tool rotation，intense frictional heat was generated in the tool-work piece contact area （Fig.13）.The cold zone or preheat zone conducts the heat from the interface.In UWFSW process，the heat conducted to preheat zone was readily dissipated by the convective heat transfer of water.The temperature attained in the preheat zone was too low for UWFSW joint than FSW joint.Therefore，high yield stress was created which in turn results high heat f l ux in the shoulder-work piece contact area.This was one of the reasons for the higher heat f l ux created in the shoulder-work piece contact area of UWFSW.

Fig.13.Temperature prof i les （a）Cross sectional temperature contour of FSW joint （b）Cross sectional temperature contour of UWFSW joint （c）Thermal plots.

The material in the trailing edge of the tool experienced gradual drop in the temperature.Because of thermal softening，the material in the leading edge got dropped in yield strength and got easily transported to the rear end.The plasticized material offers poor friction between the tool and work piece and therefore relatively low temperature is observed in the rear end. As the tool traverses，cooling cycle takes place in the welded region.Hence，low thermal gradient was observed at the rear side of the tool in FSW joint.The isothermal was larger at the rear and smaller at front of the tool.This was because of the heat build-up behind the tool during tool traversing.But in UWFSW process，the heat was readily dissipated by the water and hence high thermal gradient was observed at the rear end. Moreover，similar isothermals were observed at both the front and rear end of the tool during UWFSW （Fig.12）.

The temperature of isothermals can be correlated with the degree of thermal softening of the material.High temperature isothermal results in high degree of thermal softening and viceversa.The isothermal was conf i ned around the tool in UWFSW joint but for FSW joint，widespread isothermal was noted. Consequently thermal softening was also conf i ned near the tool in UWFSW joint and the extent of thermal softening is higher for FSW joint.The softened band of material around the tool gets deformed to form stir zone and TMAZ.Because of the narrowed softening，the extent of region undergoing plastic deformation was limited in UWFSW joint.So a narrowTMAZ was formed in the UWFSW.All these facts are even more clearly evident from the cross sectional temperature mapping（Fig.13）.

From the literature it was identif i ed that the TMAZ of aluminium alloys experiences the temperature range of around 400 °C ［33］.The cross sectional thermal contour map in the Fig.13 shows the peak temperature attained in the various region.FSW joint shows a wider TMAZ region and the temperature range was around 400 °C.At this temperature，the material gets softened and plasticized to form wider TMAZ. But in UWFSW joint，only a narrowed region experiences the same temperature.This was the main reason for achieving constricted TMAZ in the water cooled-FSW joint.The adjacent region was identif i ed as HAZ region，which has peak temperature of around 300 °C.The temperature value was enough to alter the grain size and precipitates in the material.In FSW joint，a wider region experienced this temperature.But，the same range of temperature values cannot be observed in the contour map of UWFSW joint.Hence from the FEM results，it could inferred that the temperature range affecting the microstructure in HAZ was controlled signif i cantly due to the high convection heat transfer nature of water.

5.2.Microstructure

Microstructural investigation shows the grain size variations along different region.Comparing，the grain size in the SIR of UWFSW joint is higher than the FSW joint.In UWFSW，the higher tool rotation speed of 800 rpm results high friction force between the tool shoulder and work piece interface.Hence，due to the high intense frictional heat，the grains are large sized in SIR.Nearly 80%of total heat is generated at the tool shoulder and work piece interface.But in FSW joint，due to high thermal conductivity the heat generated is evenly distributed over the stir zone.The low tool travel speed of FSW process also allows time to conduct the temperature over stir zone.Hence，the grain size over SIR，MTR and PIR are similar in size.The rate of heat transfer to the backing plate is minimal because it previously experienced heat due to conduction.So heat is build up in the PIR of FSW joint.In contrast，the backing plate experienced minimal amount of heat in UWFSW.The high rate of heat transfer to cold backing plate does not allow heat to build up in PIR.So regardless of high temperature in SIR，PIR experiences low temperature.Therefore，f i ne grains are formed in the PIR as like FSW joint.Despite of the difference in temperature distribution and grain size，the stir zone microstructure exhibit f i ne equi-axed grains in the entire region due to the dynamic recrystallization aided by the interaction effect of plastic deformation and heat generation during FSW and UWFSW process. According to the Hall Petch relationship，the f i ner grain size in the stir zone exhibits higher hardness than the other.It isinteresting to observe that the peak temperature is noted in the weld periphery and not in the weld center for both the joints. This is because at the weld center the material experiences less amount of tool rubbing than the region near to the periphery region of shoulder.

5.3.Precipitation behaviour

The precipitation sequence of Al-Cu alloy is as follows: supersaturated solid solution （SSSS）→ Guinier-Preston zone（GPZ）→ θ”→ θ’→ θ （stable）［34］.In both FSW and UWFSW process，the temperature attained in the stir zone was higher than the solutionization temperature.Therefore the precipitates get dissolved to form SSSS.From the precipitation sequence，it is understood that after solutionization，GPZ was formed in the aluminium matrix.During the cooling cycle of FSW and UWFSW process，GPZ is formed in the stir zone following the solutionization.The formation of GPZ cannot be observed in TEM microstructures because it was few Angstrom in size. Hence，it was diff i cult to differentiate GPZ from PFZ in the TEM image.

From the thermal analysis，it can conf i rm that the heat available in theTMAZ was not suff i cient to reprecipitate under both cooling conditions.By the sequence of precipitation，it was again conf i rmed that there was no possible formation of GPZ and only solutionization occurred in the TMAZ irrespective of the cooling medium.The dissolution of precipitations results in solid solution strengthening ofTMAZ.It was also observed that the thermal cycles prevailed in FSW enables the diffusion of Cu atom to form Cu rich Al4Cu9phase in the weld joint.Moreover the presence of dislocations in the TMAZ results in strain hardening.From the OM，it was noticed that the TMAZ of UWFSW underwent severe plastic deformation than the FSW joint；therefore，more dense dislocations are observed （Fig.8）.

The HAZ region experiences the peak temperature of 200 °C to 400 °C，which results in over aging.The coarsened of θ’precipitates to θ precipitates conf i rms the over aging of HAZ. Here，the heat conducted is insuff i cient to solutionize or reprecipitate.In the HAZ of UWFSW joint，the transformation from θ’to θ precipitate was greatly reduced and the precipitate distribution was more or less similar to the parent metal.This indicates that the over aging was greatly limited due to water cooling.The over aging of HAZ and the decrement in the hardness was also reported by Zhang et al.［35］.

The presence of θ’precipitate was preferable for the better mechanical properties.During loading，the dislocations movement is hindered by the θ’precipitate and thus the hardening of material takes place.The stable θ precipitate do not create suff i cient strain misf i t around the precipitate and also allows the dislocations to bypass through during loading.Similarly，PFZ deteriorates the mechanical properties because no precipitates are available to hinder the dislocations.The PFZ are strengthened by solid solution strengthening because of dissolution of solute particle in the matrix.However，the strengthening effect is lower than the precipitation strengthening.In external water cooling FSW process，the coarsening of precipitates and narrowing of LHDR were limited and thereby results in superior properties than the air cooled FSW joints.

5.4.Mechanical properties

The region of fracture can be correlated with the LHDR of microhardness plot.In both the joints，the location of failure was found to be exactly in the TMAZ-stir zone interface.From this，it can infer that the extent of softening in LHDR of the joint was a vital factor deciding the tensile properties.From the TEM analysis，it can understand that the softening of TMAZ is attributed to the presence of coarsened precipitates and precipitate free zone （Fig.8）.However，the hardness of the LHDR of UWFSW is higher because of the high extent of precipitate hardening and strain hardening than the FSW joint.A similar strengthening effect was previously reported by Liu et al.［36］. The lower hardness of HAZ of FSW joint was attributed to the presence of high volume fraction of coarse θ’precipitates.The over aging of HAZ was greatly reduced in the UWFSW which results in higher hardness than FSW joint.This was attributed to the high cooling rate and the attainment of lower temperature in the HAZ.During hardness indentation，the hardening precipitates were not available in the TMAZ to limit the plastic deformation.During tensile loading，the strain gets localized at the softerTMAZ among the various regions of weld joint.Because of the strain localization，both FSW and UWFSW joints exhibit lower elongation than the parent metal （Table 2）.Moreover，absence of necking formation was seen due to the severe strain localization.The SEM characterization of fracture surfaces revealed dimples in both the UWFSW and FSW joints which indicates that both the joints undergone ductile mode of failure. However UWFSW joint exhibits f i ne dimples than FSW joint. During tensile loading，the second phase precipitates act as the nucleation sites for the voids formation.The voids were coalesced to form the f i nal fracture.Relatively less coarse and high volume fraction of precipitates in the UWFSW joints provides more nucleation sites than FSW joints and thereby f i ne populated dimples were found.

6.Conclusions

The experimental and numerical investigation on underwater friction stir welding （UWFSW）and friction stir welding（FSW）of armour grade，high strengthAA2519-T87 aluminium alloy was conducted and following important conclusions are derived:

1）The peak temperature experienced by the UWFSW joint is 547 °C which is higher compared to the peak temperature experienced by FSW joint.However，UWFSW joint resulted in higher cooling rate and higher temperature gradient than FSW joint due to severe and even heat absorption capacity of the water cooling system.

2）The coarsening and dissolution of precipitates resulted in TMAZ as the weaker zone.The thermal gradient along the transverse and longitudinal axis of the joint is controlled by the water cooling and thereby the weaker TMAZ and HAZ are appreciably narrowed.Ultimately reduction of width of weaker zone and reduction of over aging of HAZ are the reasons for the marginal increase in tensile properties.

3）Among the various regions，RS-TMAZ exhibited lower hardness of 90 HV and 94 HV in FSW and UWFSW joints respectively.The difference in the microhardness cannot be attributed with the grain size rather it is attributed to the presence of heterogeneous precipitates across the weld joints.

4）Under water FSW joint exhibited higher tensile strength of 271 MPa and higher joint eff i ciency of 60%than conventional FSW joint.The controlling of thermal histories and its subsequent effect on precipitation behaviour are found to be the main reasons for the enhancement in the strength of UWFSW joints.

Acknowledgement

The authors gratefully acknowledge the f i nancial support of the Directorate of Extramural Research&Intellectual Property Rights （ER&IPR），Defense Research Development Organization （DRDO），New Delhi through a R&D project no. DRDO-ERIPER/ERIP/ER/0903821/M/01/1404 to carry out this investigation.The authors also wish to record their sincere thanks to M/S Aleris Aluminium，Germany for supplying the material to carry out this investigation

［1］James JF，Lawrence SK，Joseph RP.Aluminum alloy 2519 in military vehicles.Adv Mater Process 2002；160:43-56.

［2］Sudhakar I，Madhu V，Madhusudhan Reddy G，Srinivasa Rao K. Enhancement of wear and ballistic resistance of armour grade AA7075 aluminium alloy using friction stir processing.Def Technol 2008；11: 10-17.

［3］Balasubramanian V，Ravisankar V，Madhusudhan Reddy G.Effect of post weld aging treatment on fatigue behaviour of pulsed current welded AA7075 aluminium alloy joints.J Mater Engg Perform 2008；7:224-33.

［4］Kadaganchi R，Gankidi MR，Gokhale H.Optimization of process parametersofaluminiumalloyAA2014-T6frictionstirweldsbyresponse surface methodology.Def Technol 2015；11:209-19.

［5］Mehta M，Reddy GM，Rao AV，De A.Numerical modeling of friction stirwelding using the toolswith polygonalpins.Def Technol 2015；11:229-36.

［6］Mishra RS，Ma ZY.Friction stir welding and processing.Mater Sci Eng R 2005；50:1-78.

［7］Rambabu G，Balaji naik D，Venkata Rao CH，Madhusudan Reddy G，Srinivasa Rao K.Optimization of friction stir welding parameters for improved corrosion resistance of AA2219 aluminum alloy joints.Def Technol 2015；11:330-7.

［8］Threadgill PL，Leonard AJ，Shercliff HR，Withers PJ.Friction stir welding of aluminum alloys.Int Mater Rev 2009；54:49-93.

［9］Liang XP，Li HZ，Li Z，Hong T，Ma B，Liu SD，et al.Study on the microstructure in a friction stir welded 2519-T87 Al alloy.Mater Des 2012；35:603-8.

［10］SimarA，BechetY， DeMeesterB，DenquinA，PardoenT. Microstructure，local and global mechanical properties of friction stir welds in aluminium alloy 6005A-T6.Mater Sci Eng A 2008；486:85-95.

［11］Ayd?n H，Bayram A，Uguz A，Akay KS.Tensile properties of friction stir welded joints of 2024 aluminum alloys in different heat treated state. Mater Des 2009；30:2211-21.

［12］Sivaraj P，Kanagarajan D，Balasubramanian V.Effect of post weld heat treatment on tensile properties and microstructure characteristic of friction stir welded armour grade AA7075-T651 aluminium alloy.Def Technol 2014；10:1-8.

［13］Dixit V，Mishra RS，Lederich RJ，Talwar R.Inf l uence of process parameters on microstructural evolution and mechanical properties in friction stirred Al-2024（T3）alloy.Sci Tech Weld Join 2009；14:346-55.

［14］Lee WB，Yeon YM，Jung SB.The improvement of mechanical properties of friction stir welded A356 Al alloy.Mater Sci Eng A 2003；355:154-9.

［15］Liu HJ，Zhang HJ，Yu L.Effect of welding speed on microstructures and mechanical properties of underwater friction stir welded 2219 aluminum alloy.Mater Des 2011；32:1548-53.

［16］Fu RD，Sun ZQ，Sun RC，Li Y，Liu HJ，Liu L.Improvement of weld temperature distribution and mechanical properties of 7050 aluminum alloy buttjointsbysubmergedfrictionstirwelding.MaterDes2011；32:4825-31.

［17］Fratini L，Buffa G，Shivpuri R.Mechanical and metallurgical effects of in process cooling during friction stir welding of AA7075-T6 butt joints. Acta Mater 2010；58:2056-67.

［18］Jie LH，Jie ZH，Lei Y.Homogeneity of mechanical properties of underwater friction stir welded 2219-T6 aluminum alloy.J Mater Engg Perform 2011；20:1419-22.

［19］Grujicic M，He T，Arakere G，Yalavarthy HV，Yen CF，Cheeseman BA. Fully coupled thermomechanical f i nite element analysis of material evolution during friction stir welding ofAA5083.Proc Inst Mech Eng Part B:J Eng Manufacture 2010；224（4）:609-25.

［20］Heurtier P，Jones MJ，Desrayaud C，Driver JH，Montheillet F，Allehaux D. Mechanical and thermal modelling of Friction Stir Welding.J Mater Process Tech 2006；171（3）:348-57.

［21］Jie ZH，Jie LH，Lei Y.Microstructure and mechanical properties as a function of rotation speed in underwater friction stir welded aluminum alloy joints.Mater Des 2011；32:4402-7.

［22］Zhang H，Liu H.Mathematical model and optimization for underwater friction stir welding of a heat-treatable aluminum alloy.Mater Des 2013；45:206-11.

［23］Sree Sabari S，Balasubramanian V，Malarvizhi S，Madusudhan Reddy G. Inf l uence of post weld heat treatment on tensile properties of friction stir welded AA2519-T87 aluminium alloy joints.J Mech Behav Mater 2015；24（5-6）:195-205.

［24］Radaj D.Heat effects of welding-temperature f i eld，residual stress，distortion.Berlin:Springer Verlag；1992.

［25］Zhu XK，Chao YJ.Numerical simulation of transient temperature and residual stresses in friction stir welding of 304L stainless steel.J Mater Process Tech 2004；146:263-72.

［26］Chao YJ，Qi X，Tang W.Heat transfer in friction stir weldingexperimental and numerical studies.Trans ASME 2003；125:138-45.

［27］Zhang HJ，Liu HJ，Yu L.Thermal modeling of underwater friction stir welding of high strength aluminum alloy.Trans Nonferrous Met Soc China 2013；23:1114-22.

［28］Song M，Kovacevic R.Thermal modeling of friction stir welding in a moving coordinate system and its validation.Int J Mach Tools Manu 2003；43:605-15.

［29］Colegrove P.3-dimensional f l ow and thermal modelling of the friction stir welding process.Proceedings of the 2nd international symposium on friction stir welding.Gothenburg，Sweden:2000.

［30］Chen YC，F(xiàn)eng JC，Liu HJ.Precipitate evolution in friction stir welding of 2219T6 aluminum alloys.Mater Charact 2009；60:476-81.

［31］Srinivasan PB，Arora KS，Dietzel W，Pandey S，Schaper MK. Characterisation of microstructure，mechanical properties and corrosion behaviour of an AA2219 friction stir weldment.J Alloy Comp 2010；492: 631-7.

［32］Paglia CS， BuchheitRG.Microstructure， microchemistry and environmental cracking susceptibility of friction stir welded 2219-T87. Mater Sci Eng A 2006；429:107-14.

［33］Chen CM，Kovacevic R.Finite element modeling of friction stir welding-thermal and thermomechanical analysis.Int J Mach Tools Manu 2003；43:1319-26.

［34］Son SK，Takeda M，Mitome M，Bando Y，Endo T.Precipitation behavior of anAl-Cu alloy during isothermal aging at low temperatures.Mater Lett 2005；59:629-32.

［35］Zhang Z，Xiao BL，Ma ZY.Inf l uence of water cooling on microstructure and mechanical properties of friction stir welded 2014Al-T6 joints.Mater Sci Eng A 2014；614:6-15.

［36］Jie LH，Jie ZH，Xian HY，Lei Y.Mechanical properties of underwater friction stir welded 2219 aluminum alloy.Trans Nonferrous Met Soc China 2010；20:1387-91.

Received 7 January 2016；revised 2 February 2016；accepted 15 February 2016 Available online 28 March 2016

Peer review under responsibility of China Ordnance Society.

*Corresponding author.Tel.:+91 04144 239734.

E-mail address:visvabalu@yahoo.com （V.BALASUBRAMANIAN）.

http://dx.doi.org/10.1016/j.dt.2016.02.003

2214-9147/? 2016 China Ordnance Society.Production and hosting by Elsevier B.V.All rights reserved.

? 2016 China Ordnance Society.Production and hosting by Elsevier B.V.All rights reserved.