Materials and Design 32 (2011) 3854–3864

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier .com/locate /matdes

Microstructure and mechanical properties of resistance upset butt welded 304austenitic stainless steel joints

M. Sharifitabar a, A. Halvaee b,⇑, S. Khorshahian a

a Metallurgy and Materials Engineering Division, Sistan & Baluchestan University, Zahedan, Iranb School of Metallurgy and Materials Engineering, University College of Engineering, University of Tehran, P.O. Box: 11365-4563, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history:Received 26 October 2010Accepted 3 March 2011Available online 10 March 2011

Keywords:A. Ferrous metals and alloysD. WeldingE. MechanicalF. Microstructure

0261-3069/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.matdes.2011.03.007

⇑ Corresponding author. Tel.: +98 2161114104; faxE-mail address: halvaee@ut.ac.ir (A. Halvaee).

Resistance upset welding (UW) is a widely used process for joining metal parts. In this process, current,time and upset pressure are three parameters that affect the quality of welded products. In the presentresearch, resistance upset butt welding of 304 austenitic stainless steel and effect of welding power andupset pressure on microstructure, tensile strength and fatigue life of the joint were investigated. Micro-structure of welds were studied using scanning electron microscopy (SEM). X-ray diffraction (XRD) anal-ysis was used to distinguish the phase(s) that formed at the joint interface and in heat affected zone(HAZ). Energy dispersive spectroscopy (EDS) linked to the SEM was used to determine chemical compo-sition of phases formed at the joint interface. Fatigue tests were performed using a pull–push fatigue testmachine and the fatigue properties were analyzed drawing stress-number of cycles to failure (S–N)curves. Also tensile strength tests were performed. Finally tensile and fatigue fracture surfaces were stud-ied by SEM. Results showed that there were three different microstructural zones at different distancesfrom the joint interface and delta ferrite phase has formed in these regions. There was no precipitation ofchromium carbide at the joint interface and in the HAZ. Tensile and fatigue strengths of the jointdecreased with welding power. Increasing of upset pressure has also considerable influence on tensilestrength of the joint. Fractography of fractured samples showed that formation of hot spots at high weld-ing powers is the most important factor in decreasing tensile and fatigue strengths.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Resistance upset welding is a solid-state welding process whichinvolves the interaction of electrical, thermal, mechanical and met-allurgical phenomena. In this process, the joining surfaces are keptat a forced contact; followed by a high electric current passingthrough the workpieces. Due to the contact resistance and Jouleheating, a vast amount of heat is generated at the faying surfaces.Before, during and after applying the electric current, force isapplied to maintain the electric current continuity and to providethe pressure necessary to form the weld zone. The metal at thejoint is heated to a temperature where recrystallization can rapidlyoccur across the heated surfaces. In this process, similar to otherresistance welding processes, there is no requirement to any extra-neous material such as filler material or shielding gas [1]. In thiswelding process there are two types of resistances namely contactresistance and bulk resistance. At the earlier stages of the welding,contact resistance plays the main role but gradually it decreasesand the role of bulk resistance becomes more important [2,3].

ll rights reserved.

: +98 2188006076.

Kanne expressed that in comparison with fusion welding pro-cesses, the chemical composition and metallurgical properties arenot significantly changed leading to better mechanical properties.Simplicity, welding speed, capability of remote control and inde-pendence of welding quality from the operator skill are the otheradvantages of this process [4]. Miyazaki et al., Kang, Kanne andSharifitabar and Halvaee stated that resistance upset welding is asuitable welding process for applications such as sealing of atomicwaste containers, welding of automotive parts and joining of stain-less steels, low carbon steels, super alloys, aluminum alloys andparts made of dissimilar materials [4–7]. The general configurationof parts and equipments in upset welding is shown in Fig. 1.

Stainless steels play an important role in the modern world.Austenitic stainless steels represent more than 2/3 of the totalstainless steel production. These stainless steels are preferred morethan other stainless steel types due to their good weldability [8].But there are some negative metallurgical changes during weldingof these steels which should be considered. They are [9,10]:(a) formation of delta ferrite phase, (b) formation of sigma phase,(c) stress corrosion cracking, (d) precipitation of chromium carbideat grain boundaries and (e) formation of hot cracks.

Nikitin et al. and Nikitin and Bses stated that fatigue behavior ofaustenitic stainless steel welds is strongly affected by stress ampli-

Fig. 1. Schematic illustration of resistance upset welding process.

M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864 3855

tude, temperature, frequency and welding conditions [11,12]. Mostof the service failures are expected to occur either in the HAZ or inthe weld metal. These failures are most frequently associated withdefects or microstructural in-homogeneities. But with variations inwelding conditions, changes in the type and the amount of defectsand in-homogeneities lead to variations in fatigue behavior of thejoint [13].

Plastic deformation of meta-stable austenitic steels leads to aphase transformation from paramagnetic austenite to ferromag-netic martensite [14,15]. Smage showed that consequences of thistransformation for the application of these materials can be posi-tive or negative. Increment of the strength, e.g. the transformationinduced plasticity (TRIP) effect and increase in the lifetime in thehigh cycle fatigue (HCF) range are advantages in contrast to localincrease of the hardness and related reduction in ductility [14].

Because of high cooling rate, short welding time and formationof the joint in solid state in resistance upset but welding, there ispossibility for elimination of some of these metallurgical changesin welding of austenitic stainless steels by UW.

The literature in the upset welding field is not very extensive.The first reported work on development of UW was the researchdone at NASA Lewis Research Center. In This project Holko focusedon magnetic resistance upset welding of stainless steel 304 plateswith different thicknesses [16]. Resistance welding of nuclearwaste containers was another application of this technology whichrequired design of new equipment able to deliver currents of up to400,000 A at 64,000 kgf. The same application was further reportedby Kanne [17]. He examined the properties of upset welded cylin-drical and spherical components. He pointed out that advantagesof UW, compared to fusion welding processes, include fewer de-fects and stronger welds with a faster and more reliable process.Cannell used UW for welding canisters made of 304L stainless steel[18]. Bezprozvannyi at Paton Welding Institute reported upsetwelding of high-speed steel to carbon steel with a current regula-tion system for controlling special cyclic welding [19]. The effect ofvariation in upset butt welding parameters, such as current andwelding length on the hardness of different regions of the HAZ,microstructure and toughness of the weld in high strength low-al-loy steel weldment were studied by Ghosh and Gupta [20]. Miya-zaki et al. examined the upset weldability of Nb-bearing highstrength steel of the 600 MPa level. They found that the higherwelding current density requires shorter upset length (the length

that two samples penetrate into each other during welding) forproducing a high quality weld. Also they reported that the requiredupset length can be reduced using lower welding forces [5]. Kannealso reported applicability of the UW process to weld a variety ofstainless steels (including A-286), super alloys (including TD nick-el), refractory metals (including tungsten) and aluminum alloys(including 2024) [4]. Shieh and Chang presented a study of upsetwelding process in wire drawing; obtaining the optimum parame-ters of the operation for a better distribution of hardness in thewire [21]. Further, Cannel et al. wrote on the optimization and reli-ability of UW process [22]. In a study by Kang et al. the upset wel-dability and formability of a particular kind of material (SPCC) wasinvestigated. The results showed that the formability of upsetwelded SPCC steel sheets were slightly lower than that of the par-ent material [11]. Applications of upset welding processes were re-cently extended to cast iron parts by Shakhmatov and Shakhmatovand dissimilar austenitic to martensitic stainless steels by Sharifit-abar and Halvaee [7,23]. They found that a good metallurgical bondcan be produced between austenitic and martensitic stainlesssteels by resistance upset welding. Also in the past decade somework has been carried out on numerical simulation of resistanceupset welding. Recently, Kerstens and Richardson reported anexperimental study of weld development during resistance upsetbutt welding process. They also made a simplified thermal finiteelement model to explore the influence of welding conditions onheating [2]. In a very recent study Hamedi et al. considered numer-ical simulation and experimental investigation of UW processparameters including heating and post-weld heating current andtheir corresponding duration as well as interference of the part fea-tures that form the joint and effect of these parameters on tensilestrength of a low carbon low alloy oil pressure sensor. They foundthat both numerical and experimental results suggest an optimumset of welding parameters, i.e. time and electrical current that yielda maximum value for the tensile strength of the joint. Also the ef-fects of post-weld heating time and current on the tensile strengthshowed that these parameters had a remarkable effect on improv-ing tensile strength of the weldment [24].

In this research, resistance upset butt welding of 304 austeniticstainless steel and effect of welding power and upset pressure onmicrostructure, tensile and fatigue life of the joint were investi-gated in order to correlate the weld quality to the variation of thesetwo parameters and introducing optimum welding conditions.

2. Experimental procedure

Chemical composition of AISI 304 stainless steel used in this re-search was 0.04%C, 0.48%Si, 1.75%Mn, 18.15%Cr, 8.2%Ni, 0.045%P,0.016%S, 0.7%Cu and 0.11Mo. Also yield and tensile strengths ofthe steel used were 242 and 658 MPa respectively.

Start material was heated 10 min at 1060 �C and cooled in air todiminish cold work effects due to mechanical processing prior towelding. The welding machine used in this research had been man-ufactured by Electro-TechnoTak Company (Tehran, Iran) and itsmaximum power was 25 KVA. Then surfaces of samples wereground by 1000 mesh grinding paper to remove oxide layer formedduring heat treatment. Two rods of 50 mm length and diameter of8 mm were clamped in upset welding machine for each state ofwelding. Table 1 shows welding conditions and measured param-eters during welding. Primary and upset pressures were the sameand were applied by a mechanical system. Firstly, heating pressurewas applied on the faying surfaces. Then electrical current waspassed through the bars in contact. During welding, the electricalpotential was measured by an AVO meter and welding powerwas calculated using following equation

P ¼ VI ð1Þ

Table 1Selected conditions and measured parameters during welding.

Samplename

Weldingcurrent (A)

Weldingpressure (MPa)

Weldingtime (S)

Heat input(V.A.S)

A1B1 1500 1.01 1.5 4500A2B1 2000 1.01 1.42 5680A3B1 2500 1.01 1.33 6650A4B1 3000 1.01 1.21 7260A1B2 1500 1.15 1.51 4530A2B2 2000 1.15 1.41 5640A3B2 2500 1.15 1.33 6650A4B2 3000 1.15 1.19 7140A1B3 1500 1.27 1.5 4500A2B3 2000 1.27 1.40 5600A3B3 2500 1.27 1.31 6550A4B3 3000 1.27 1.20 7200A1B4 1500 1.41 1.49 4470A2B4 2000 1.41 1.42 5680A3B4 2500 1.41 1.32 6600A4B4 3000 1.41 1.21 7260

3856 M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864

where P is the welding power (Volt–Ampere), I is the current inten-sity (Ampere) and V is the electrical voltage (Volt) [2]. Also, weldingtime which is the passing time of electrical current was measuredby the AVO meter.

Tensile test was carried out by MST30/MH machine at 2 mm/min displacement rate on welded samples. This test was per-formed according to ASTM-E8 standard [25] while the joint inter-face was held in the middle of the tension samples and the flashwas removed. The examination method in fatigue investigationwas the single factor method, i.e. for each series of experimentsone factor was varying while the other parameter was kept con-stant at previously optimized levels of tensile strength. The aimof the authors was to investigate the effect of UW process param-eters on fatigue properties of 304 austenitic stainless steel joints. Inother words, the authors wanted to see how variations in weldingparameters can affect the fatigue life of the joint. For this purpose,three different stress amplitudes were selected higher than yieldstrength of the alloy to represent the results of the tests as S–Ncurves. The Pull–push fatigue test was performed according toASTM-E606–92 standard [26] while the joint interfaces were heldin the middle of the samples. These test was carried out by IN-STRON 8502 fatigue testing machine at R = �1, frequency of 2 Hzand 320, 370 and 430 MPa stress amplitudes. The metallographicsamples were cut longitudinally and the interfaces were studiedby SEM Cam Scan MV2300 and Energy dispersive X-ray lined upto the SEM after preparation and etching with Kalling No. 2 agent(5 g CuCl2, 100 ml HCl and 100 ml Ethanol). XRD test was fulfilledby wave length of Coka ¼ 1:7889 nm: Finally fracture surfaces werestudied by SEM.

3. Results and discussion

Leber et al. showed that cold work causes formation of non-homogeneities such as shear bands, mechanical twins and defor-mation induced martensite in the microstructure of austeniticstainless steels. Presence of non-homogeneities in the microstruc-ture leads to decreasing corrosion resistance of these alloys [15].Therefore in many applications, austenitic stainless steels are usedin normalized heat treating condition after cold working. There-fore, in this research normalizing heat treatment was performedon samples before welding to improve corrosion resistance.Fig. 2a and b shows microstructure of base metal before and afternormalizing respectively. It is observable that the microstructureconsists of high density of mechanical micro and macro twins. Alsosome transformation induced martensite is formed in twins and

twin-matrix boundaries (Fig. 2a). After annealing, equiaxe grainsand annealing twins were formed in the microstructure (Fig. 2b).

3.1. Study of microstructure

Fig. 3a shows macrostructure of half of the joint interface insample A1B3. Fig. 3b–d shows the microstructures of differentzones formed at the joint interface. As can be seen, three differentmicrostructural zones have been formed in the interface due tothermal gradient between the joining faces and electrodes. Theyare:

1. Widmanstätten austenite formation zone (WAZ), an ellipticalzone at the center of the joint interface consisted of differentmorphologies of austenite (Fig. 3b). In addition to allotriomor-phic austenite (Ac) localized in grain boundaries and intergran-ular austenite (Ic), unusual austenite morphology is found.According to literatures, this microstructure is Widmanstättenaustenite (Wc) [27,28]. Small amounts of ‘‘lathy’’ d-ferrite couldbe found within Widmanstätten austenite laths. The Wid-manstätten austenite structure is more common in austeniticstainless steels solidifying as d-ferrite [28]. Woollin, expressedthat Like Widmanstätten ferrite found in carbon steels, appear-ance of Widmanstätten austenite is one of narrow wedges ema-nating either directly from a grain boundary or fromallotriomorphic ferrite (allotriomorphic austenite in case ofWidmanstätten austenite) [27].

2. Dynamic recrystallization zone (DRZ) around the WAZ com-posed of fine austenite grains and delta ferrite phase is formedat the grain boundaries (Fig. 3c).

3. Partially recrystallization zone (PRZ) which contains recrystalli-zation lines along the drawing direction during manufacturingof the bar and delta ferrite formed along some of these lines(Fig. 3d).

According to statements of Kerstens and Richardson and Song[2,3], because electrical resistance of faying surfaces is higher thanbulk resistance of the materials, during welding contact surfacesare hotter than bulk of the samples and their temperature roseup to liquid + delta ferrite zone in Fe–Cr–Ni phase diagram. Whenthe upset pressure was applied, the liquid metal at the edges ofcontact surfaces of samples was rejected as flash and was replacedwith mushy metal led to formation of upset in contact area. But inthe center of the contact surfaces, the liquid metal was trapped andsolidified. Due to high Cr equivalent to Ni equivalent ratio (Creq/Nieq = 1.91), the solidification microstructure is fully ferrite[28,29]. In austenitic stainless steels, Delta ferrite is not stable atroom temperature. So it transforms to austenite phase during cool-ing. Southwick and Honeycombe concluded that decomposition ofd-ferrite to austenite occurs by two different mechanisms depend-ing upon the transformation temperature. At high temperature, thereaction occurs by a diffusional nucleation and growth processwhereas at low temperature the austenite phase forms by a displa-cive mechanism. It is believed that the Widmanstätten austenitegrows by a displacive mechanism whereas allotriomorphic austen-ite is considered to be a reconstructive transformation product[30]. Menezes et al. have reported that in bead on plate weldingof two-phase ferritic–austenitic stainless steels, residual compres-sive stresses were formed near the ferrite to austenite transforma-tion temperature. These stresses increased the probability offormation of Widmanstätten austenite [31]. During solid statetransformation of ferrite to austenite in resistance upset weldingof this steel, allotriomorphic austenite formed in ferrite grainboundaries. But the transformation across the entire grain wassuppressed by high cooling rate of upset butt welding resultinglow diffusion rate and low driving force due to low transformation

Fig. 2. Microstructure of base metal (a): before and (b): after annealing heat treatment.

Fig. 3. (a): Macrostructure of resistance upset butt welding joint of 304 stainless steel. (b)–(d): different microstructural zones formed in the joint.

M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864 3857

temperature. This caused residual ferrite to transform to Wid-manstätten austenite by displacive mechanism. Compressive stres-ses during transformation in this welding process also encouragedferrite to Widmanstätten austenite transformation (Fig. 3b).

Joining surfaces have the highest temperature during weldingand with increasing distance from the weld interface, temperaturedecreases [2,28]. Around the interface, temperature rose up to aus-tenite + delta ferrite in Fe–Cr–Ni phase diagram resulting to theformation of ferrite at austenite grain boundaries (Fig. 3c). Because

of high cooling rate of the joint, the possibility of ferrite to austen-ite transformation was low and some ferrite was remained in grainboundaries. So microstructure of this region consisted of austeniteand delta ferrite [27]. Fuller et al. concluded that presence of d-fer-rite in grain boundaries prohibits grain growth and so the grainsare fine. Also this phase acts as a crack growth inhibitor and re-duces the possibility of intergranular fracture [32]. But Lippoldand Kotecki expressed that presence of delta ferrite decreasesformability of austenitic stainless steels and increases probability

3858 M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864

of precipitation of carbides [28]. On the other hand, dynamicrecrystallization due to hot deformation may be one of the mostimportant factors in decreasing the grain size in this region asmentioned by Humphreys [33].

Fig. 3d shows how recrystallization is limited to a series of linesalong the drawing direction during manufacturing of the bars anddelta ferrite is formed along some of these lines. It is observed thatdirection of partial recrystallization lies on the direction of theshear bands formed due to mechanical working during manufac-turing of the bars. Cizek stated that these bands are composed offine dislocation containing cells which grow parallel to each otherand pass through the grains. These cells have high angle non-crys-tallographic grain boundaries with the matrix [34]. The possiblereason for formation of this region is that heat treatment in this re-search could not remove these cells due to high density of disloca-tions and low stacking fault energy in austenitic stainless steels.During welding, because of stress concentration in these bands,density of dislocation increased. So, heat and pressure caused dy-namic recrystallization along these bands. But around the bands,possibility of dynamic recrystallization was low due to low densityof dislocations. Also because of the high dislocation density inthese shear bands; diffusion rate of ferrite promoting elementssuch as chromium and segregation of these elements was highleading to formation of high temperature delta ferrite along theshear bands and in austenite grain boundaries [28].

Microstructural analysis of samples welded according to differ-ent welding conditions represented in Table 1 showed that increas-ing of welding power at a constant welding pressure (e.g. samplesA1B2–A4B2) caused widening all different zones formed in the jointinterface. As stated before, higher welding power produced higherheat at the joint interface resulting to the formation of consider-able amount of liquid metal at the interface and widening theWAZ. Also, The heat generated by bulk resistance during weldingincreased at higher welding powers leading the wider zone of basemetal temperature to rise to austenite + ferrite region in Fe–Cr–Niphase diagram. This caused widening the zone that can dynami-cally recrystallized during welding according to following equation

Z ¼ e0 expðQ=RTÞ ð2Þ

where Z is Zener–Holloman parameter, e0 is strain rate, Q is activa-tion energy, R is gas constant and T is temperature. According to thisequation, higher temperature during hot deformation reduces Z andincreases the possibility of dynamic recrystallization and thereforewidening DRZ and PRZ [33]. On the other hand, higher heat gener-ated by bulk resistance of samples decreased temperature gradient

Fig. 4. XRD analysis of

between the joint interface and bulk of the samples. This reducedthe cooling rate leading to decreasing the amount of grain boundaryferrite near the interface and grain growth in this region.

By using higher welding pressure at constant welding powers(e.g. samples A2B1–A2B4), width of the WAZ decreased. But itcaused widening DRZ and PRZ. Higher welding pressures lowerthe interface resistance according to Eq. (3) and increase the rollof bulk resistance on heat generation during welding and forma-tion of lower amount of liquid at the interface.

Rc ¼ ðq=2ÞðpHB=FupsÞ1=2 ð3Þ

where Rc is contact resistance, HB is Brinell hardness, q is specificresistivity and Fups is upsetting force [2]. Also, high welding pressurecaused rejection of liquid metal formed at the interface duringupsetting and decreased width of the WAZ at high welding pres-sures. On the other hand, increasing the roll of bulk resistance onheat generation at high pressures led to widening the area thatcan dynamically recrystallize according to following equation

Z ¼ C1 sinhðC2rÞn ð4Þ

where C1, C2 and n are constants and r is stress exerted during hotdeformation [33]. According to this equation, higher welding pres-sures increased the possibility of dynamic recrystallization andwidening DRZ and PRZ.

3.2. Phase analysis

Fig. 4 shows the result of XRD analysis for weld metal and HAZ.Only there are austenite and delta ferrite phases in these regionsand no precipitation of chromium carbides (Cr23C6) phase is ob-served. This is because of high cooling rate of weld interfaces fromthe temperature range of chromium carbide precipitation (450–850 �C).

Fig. 5a and b shows EDX analysis of the black phase in Fig. 3c.Also chemical composition of this phase is shown in Table 2. Itcan be seen that the amount of Cr increased and Ni decreased inthis phase and its chemical composition is in the ferrite region ofthe Fe–Cr–Ni phase diagram at room temperature [28].

3.3. Mechanical properties

3.3.1. Tensile propertiesFig. 6 shows that tensile strength of the joint decreases with

welding power. Welding power of 3 KVA maintained enough heat

the joint interface.

Fig. 5. (a): Black phase formed in the microstructure and (b): EDX line scan analysis of the black phase in the microstructure.

Table 2EDX Chemical composition of black phase in the microstructure.

Element Cr Ni Si Mn Fe

Wt% 23.93 4.71 0.46 1.61 Bal

M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864 3859

to produce a mushy zone and a complete metallurgical joint. Butwith increase in welding power, the grains within the HAZ grawand the joint strength decreased according to Hall–Petch equation.On the other hand, higher heat input produced at high weldingpowers increased residual liquid at the joint face and producedconsiderable amount of Widmanstätten austenite phase. Plateform of this phase increased stress concentration in this regionand therefore decreased joint strength.

Formation of hot spots at the joint interface also reduced jointstrength at high welding powers. Fig. 7 shows a hot spot in the cen-ter of the joint interface of sample A4B1 formed due to heteroge-neous distribution of electrical current and heat.

Using finite element method (FEM) and experimental investiga-tions, Kerstens and Richardson [2] showed that heterogeneousheating and formation of hot spots occur because of non-uniformcurrent density passing through the material. This heterogeneity

Fig. 6. Effect of welding power on

can be formed by in-homogeneities in the material and local vari-ations in the interface resistance resulting from contaminations,non-uniform deformation or surface imperfections. A non-uniformupsetting pressure distribution over the joint may also play a role.If the upsetting pressure is not uniform, then according to Eq. (3)there will be a difference in contact resistance over the joint area.

Non-uniformity in current distribution also may arise as a resultof contact resistance variations at the electrode/sample interfacedue to the electrode surface condition and contaminations or vari-ations of the clamping force. With increasing of welding power, theprobability of hot spot formation rises due to increase in the heatinput to the joint interface.

Fracture surface of sample A4B2 (welded at high welding power)is shown in Fig. 8. The fracture mode is completely ductile in thissteel. It can be seen that formation of hot spot at the interface ofthis sample caused crack initiation and reduced the joint strength.

Fig. 9 shows that with increasing of the upset pressure, strengthof the joint increases firstly and then decreases. Contact resistancerises with decrease in upset pressure according Eq. (3). Experimen-tal investigations of Song et al. [3] also showed that contact resis-tance increases with decreasing of welding pressure. So, lowwelding pressures leads to widening the area for formation of Wid-manstätten austenite and decreasing joint strength [35]. Also at

tensile strength of the joint.

Fig. 7. SEM macrostructure of a hot spot in sample A4B1 and formation of cracks inthis region.

3860 M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864

low welding pressures, good metallurgical bond between samplesdid not happen and therefore the strength of the joint is low. Withincreasing of upset pressure up to 1.27 MPa, amount of liquidmetal formed at the joint interface decreased. Also, large amount

Fig. 8. Fracture surface of sample A4B2, a: lower magnification

Fig. 9. Effect of upset pressure on

of liquid was rejected in upsetting stage due to high welding pres-sure. This caused thinning the area for formation of Widmanstät-ten austenite and increasing joint strength (sample A1B3). Alsoincrease in upset pressure led to formation of good metallurgicaljoint. By using welding pressure higher than 1.27 MPa; all liquidand mushy metals were rejected from the joint interface as flashand an incomplete joint was formed which reduced the jointstrength (sample A1B4). On the other hand, effect of welding pres-sure on tensile strength decreased with increasing of weldingpower (Fig. 9). This may be due to the formation of hot spots athigh welding powers. Because of the presence of welding defectat the joint area, effect of microstructure on mechanical propertieswas reduced led to decreasing the effect of welding pressure ontensile strength at high welding powers.

Fig. 10 shows fracture surface of sample A3B1 (welded at lowwelding pressure). In Fig. 10a there are different crack initiationsites at the fracture surface. The higher magnifications of zones1–3 are shown in Fig. 10b–d respectively. It is observed that pres-ence of plate form Widmanstätten austenite phase at the jointinterface caused formation of large voids due to stress concentra-tion and reduced the joint strength at low welding pressures.

3.3.2. Fatigue propertiesFig. 11 show the effect of welding power on fatigue life of the

joint at different stress amplitudes in samples welded with1.27 MPa welding pressure. It is observed that fatigue life ofwelded samples is lower than base metal and decreases slightly

b: higher magnification showing hot spot on this surface.

tensile strength of the joint.

Fig. 10. (a): Fracture surface of sample A3B1 (b)–(d) higher magnification of crack initiation sites on the fracture surface.

M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864 3861

with increasing of welding power from 3000 to 5000 V.A. But insamples welded with welding power of 6000 V.A, fatigue life de-creases remarkably in all stress amplitudes. Slight decrease in fati-gue strength with increasing of welding power from 3000 to5000 V.A may be due to grain growth in the HAZ of welded sam-ples. As stated by Hertzberg, according to Hall–Petch equation,coarse grain size reduces tensile, fatigue strength and consequentlyfatigue life [34].

As stated before, the probability of hot spot formation increasesat higher welding power due to higher heat input to the joint area.

Fig. 11. Effect of welding power on fatigue strength of the joint.

Because of the melted and solidified microstructure and presenceof cracks into the hot spots, stress concentration decreases the fa-tigue strength. Fig. 12a shows the main crack initiation site on thefatigue fracture surface of the sample welded at high weldingpower (6000 V.A) and tested at 320 MPa stress amplitude. In theFig. 12b and c the higher magnification of this site and dendritesarisen from melting and solidifying in the hot spots are shownrespectively. Therefore it can be concluded that formation of hotspots is the main reason in decreasing of the fatigue life at weldingpower of 6000 V.A.

Variation in welding pressure at welding power of 3 KW had noconsiderable influence on fatigue life of the joints and therefore itis not represented here.

Fig. 13a and b show the fatigue fracture surfaces at the stage II offatigue crack propagation in samples A2B3 tested at 320 and430 MPa stress amplitudes respectively. There are secondary crackson the fracture surface of the sample tested at 320 MPa (Fig. 13a).But no secondary cracks are observed at the fracture surface of thesample tested at 430 MPa (Fig. 13b). Observation of cracks couldbe associated with the partially transformed martensite phase inaustenite matrix. Because of non-uniform microstructure, localstresses may be concentrated at these locations causing secondarycracks to initiate [36]. But at the stress amplitude of 430 MPa, thephenomenon of the self-heating of the specimens was much morepronounced and affected deformation behavior of the sample. Sur-face temperature of the samples tested at 320, 370 and 430 MPastress amplitudes was measured by thermocouples. It is observedthat it rose up to 40, 69 and 85 �C respectively (the Md(30/50)(�C)temperature for investigated steel is 47 �C where Md(30/50)(�C) isthe temperature at which 50 vol% a-martensite is formed in this

Fig. 12. Fatigue fracture surface of the sample welded at high welding power.

Fig. 13. Fatigue fracture surface of the sample A2B3 tested at two different stress amplitudes (a): 320 MPa and (b): 430 MPa.

3862 M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864

steel after a true tensile strain of 30%) [8]. This relatively high tem-perature in sample tested at 430 MPa stress amplitude inhibitedmartensite transformation. In addition, plastic deformation bymigration of dislocations was facilitated at these relatively hightemperatures (thermal activation).

Fig. 14 shows the sub-surface fatigue crack formed in the Wid-manstätten austenite formation zone. Fatigue process and itsmechanisms are largely influenced by the presence of the materialin-homogeneities. Since d-ferrite is basically different from austen-ite matrix in crystallography and chemical composition, it is likelyto provide crack nucleation sites. Goyal et al. [13] conducted a

series of experiments to investigate the effect of d-ferrite on thecontinuous cycling fatigue properties. Using finite element model(FEM) calculation and transmission electron microscopy (TEM)studies, they showed that stress concentration at the delta/gammainterface occurs due to incompatibility and consequently acts ascrack initiation site. Therefore, presence of d-ferrite between theWidmanstätten plates is one of the reasons for formation ofsub-surface cracks. Also plate form of Widmanstätten phase ledto stress concentration in this region and increased probability ofsub-surface crack formation. Fatigue crack formed in the dynamicrecrystallized zone is shown in Fig. 15. It is observed that there is

Fig. 14. Fatigue sub-surface crack formed in the Widmanstatten austenite forma-tion zone in sample A2B1 tested at 320 MPa stress amplitude.

Fig. 15. Fatigue crack growth mode in the dynamic recrystallization zone in sampleA3B2 tested at 320 MPa stress amplitude.

M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864 3863

no fatigue crack deflection by d-ferrite phase in this region. There-fore it can be concluded that presence of d-ferrite in austeniticstainless steel welds causes fatigue crack initiation and does nothave any considerable effect on crack path at high stressamplitudes.

4. Conclusions

In the present investigation, resistance upset butt welding of304 austenitic stainless steel and effect of welding power and up-set pressure on microstructure, tensile strength and fatigue prop-erties of the joint were investigated. The obtained results can besummarized as follows:

1. In resistance upset butt welding of 304 stainless steel, three dif-ferent microstructural zones were formed at the joint interfacedue to thermal gradient between the joining faces and elec-trodes. These zones are: Widmanstätten austenite formationzone, dynamic recrystallization zone and partially recrystallizedzone. Also delta ferrite phase formed in these regions.

2. Increase of welding power raises heat input to the joint areaand widens all different microstructural zones at the joint inter-face. Also higher heat input increases probability of hot spot for-mation. Tension tests results showed that tensile strength of thejoint decreases with increase of welding power and hot spotsformed at high welding powers are the most important factorsin decreasing of the joint strength.

3. X-ray diffraction analysis showed that there is no precipitationof chromium carbide in the HAZ due to high cooling rate of thejoint area from the chromium carbide formation temperature.

4. With increasing in welding pressure, area for formation of Wid-manstätten austenite decreases leading to higher tensilestrength of the joint. But effect of welding pressure on tensilestrength decrease at high welding powers due to formation ofhot spots.

5. Fatigue test results indicated that fatigue life of the jointdecreases with welding power due to grain growth in theHAZ. Formation of hot spots is the other reason for decreasingfatigue strength at high welding powers. Microstructural analy-sis of the fatigue samples showed that presence of d-ferritebetween Widmanstätten austenite plates can initiate fatiguecrack and dose not have any considerable effect on crack pathat high stress amplitudes.

Acknowledgments

The authors gratefully acknowledge the extensive support ofthe Electro-Techno Tak Company and University of Tehran forexperimental and financial supports.

References

[1] Brien RL. Editor. Welding Handbook. Miami (FL, USA): American WeldingSociety: 1991.

[2] Kerstens NFH, Richardson IM. Heat distribution in resistance upset buttwelding. J Mat Proc Tech 2009;209:2715–872.

[3] Song Q, Zhang W, Bay N. An experimental study determines the electricalcontact resistance in resistance welding. Weld J 2005:73–6.

[4] Kanne Jr WR. Solid state resistance upset welding: a process with uniqueadvantages for advanced materials. In: The 2nd advanced joining technologiesfor new materials conference, Cocoa Beach: FL; 1994

[5] Miyazaki Y, Ichikawa M, Saito T. Upset weldability of niobium-bearing highstrength 600 MPa steel for wheel rims and mechanism of upset weldability.Weld World 1993;31(5):348–57.

[6] Kang SS. A study of resistance welding in steel sheet, using a tailor-weldedblank, (Report1), evaluation of upset weldability and formability. J Mat ProcTech 2000;101(1–2):186–92.

[7] Sharifitabar M, Halvaee A. Resistance upset butt welding of austenitic tomartensitic stainless steels. J Mater Des 2010;31(6):3044–50.

[8] Padilha AF. Decomposition of austenite in austenitic stainless steels. ISI J Int2002;42(4):325–37.

[9] Mumin S. Evaluation of joint interface properties of austenitic stainless steels(AISI 304) joint by friction welding. J Mat Des 2007;28:2244–50.

[10] Kou S. Solidification and liquation cracking issues in welding. Jom 2003:37–42.[11] Nikitin I, Scholtes B, Maier HJ. High temperature fatigue behavior and residual

stress stability of laser-shock peened and deep rolled austenitic steel AISI 304.J Scripta Mater 2004;50:1350–4.

[12] Nikitin I, Bses M. Effect of low frequency on fatigue behavior of austenitic steelAISI 304 at room temperature and 25 �C. Int J Fat 2008;30:2044–9.

[13] Goyal S, Sondhya R, Vaslon M, Rao BSK. The effect of thermal aging on lowcycle fatigue behavior of 316 stainless steel welds. Int J Fat 2009;31:447–54.

[14] Smage M, Walter F, Eifler D. Deformation induced martensite transformationin metasteble austenitic steel. J Mat Sci Eng 2008;483–484:394–7.

[15] Leber HJ, Niffeneger M, Tribonod B. Microstructural aspects of low cyclefatigued Austenitic stainless tube and pipe steels. J Mat Charact2007;58:1006–15.

[16] Holko KH. Magnetic force upset welding dissimilar thickness stainless steel teejoints. Weld J Suppl 1970;49(9):427–39.

[17] Kanne Jr WR. Solid state resistance welding of cylinders and spheres. Weld J1986;65(5):33–8.

[18] Cannell GR, Sessions CE. Proper procedures are the key to welding radioactivewaste canisters. Weld J 1997;76:61–7.

[19] Bezprozvannyi IA, Shirokovskii RM, Masalov YUA. Technology of resistancebutt welding of metal-cutting tool billets and their heat treatment. AutomatWeld 1989;11:62–6.

3864 M. Sharifitabar et al. / Materials and Design 32 (2011) 3854–3864

[20] Ghosh PK, Gupta PC, Goswami TK. Influence of some upset butt weldingparameters on the weld properties of HSLA steel. Indian Weld J1989;21(1):428–36.

[21] Shieh JK, Chang TL. The optimum upset welding parameters of alloy steel wire.Technol Train 1996;21(3):129–38 [Taiwan].

[22] Cannel GR. Upset welding assures a leak-tight, structurally sound seal. Weld J1999;78:41–4.

[23] Shakhmatov MV, Shakhmatov DM. Features of resistance butt welding of castiron. Weld Int 2004;18(9):737–41.

[24] Hamedi M, Eisazadeh H, Esmailzadeh M. Numerical simulation of tensilestrength of upset welded joints with experimental verification. J Mater Des2010;31:2296–304.

[25] ASTM E8. Standard Test Methods of Tension Testing Metallic Materials. AnnualBook of ASTM Standards. ASTM Int, 2004.

[26] ASTM E606-92: Standard Practice for Strain-Controlled Fatigue Testing.Annual Book of ASTM Standards, ASTM Int 2004.

[27] Woollin P, Carrouge D. Heat-affected zone microstructures in supermartensiticstainless steels. In: Proceedings of supermartensitic stainless steelsconference, Brussels, Belgium; 2002. p. 199–204.

[28] Lippold JC, Kotecki DJ. Welding metallurgy and weldability of stainlesssteels. John Wily & Sons; 2005.

[29] Shankar V, Gill TPS, Mannan SL, Sundarlsan S. Solidification cracking inaustenitic stainless steel welds. Sadhana 2003;28(2–4):359–82.

[30] Southwich PD, Honeycomb RWK. In: International conference of martensitictransformation, Boston: MIT Press; 1979. p. 189–94 [ICOMAT79].

[31] Menezes JWA, Abreu H, Kundo S, Bhadeshia HKDH, Kelly PM. Crystallographyof Widmanstätten austenite in duplex stainless steel weld metals. J Sci TechWeld Joi 2009;14(1):4–10.

[32] Fuller RW, Ehrgott Jr JQ, Heard WF, Robert SD, Stinson RD, Solanki K, et al.Failure analysis of 304 stainless steel shaft. J Engineering Failure Analysis2008;15:835–46.

[33] Humphreys FJ, Hatherly M. Recrystalization and related annealingphenomena. University of Manchester, Institute of Science and TechnologyUK: Elsevier Ltd; 2004.

[34] Cizek P. Characteristic of shear bands formed in austenitic stainless steelsduring hot deformation. J Mat Sci Eng A 2002;324:214–8.

[35] Hertzberg RW. Deformation and fracture mechanics of engineeringmaterials. John Wiley& Sons; 1996.

[36] Haskalik A, Unal W, Ozdemir N. Fatigue behavior of AISI 304 steel to AISI 4340steel welded by friction welding. J Mat Sci 2006;41:3233–9.