Materials and Design 32 (2011) 38543864

Contents lists available at ScienceDirect

Materials and Design

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

Microstructure and mechanical properties of resistance upset butt welded 304

austenitic stainless steel joints
M. Sharitabar a, A. Halvaee b,, S. Khorshahian a
a
b

Metallurgy and Materials Engineering Division, Sistan & Baluchestan University, Zahedan, Iran
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

Article history:
Received 26 October 2010
Accepted 3 March 2011
Available online 10 March 2011
Keywords:
A. Ferrous metals and alloys
D. Welding
E. Mechanical
F. Microstructure

a b s t r a c t
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 present
research, resistance upset butt welding of 304 austenitic stainless steel and effect of welding power and
upset pressure on microstructure, tensile strength and fatigue life of the joint were investigated. Microstructure of welds were studied using scanning electron microscopy (SEM). X-ray diffraction (XRD) analysis 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 composition of phases formed at the joint interface. Fatigue tests were performed using a pullpush fatigue test
machine and the fatigue properties were analyzed drawing stress-number of cycles to failure (SN)
curves. Also tensile strength tests were performed. Finally tensile and fatigue fracture surfaces were studied by SEM. Results showed that there were three different microstructural zones at different distances
from the joint interface and delta ferrite phase has formed in these regions. There was no precipitation of
chromium carbide at the joint interface and in the HAZ. Tensile and fatigue strengths of the joint
decreased with welding power. Increasing of upset pressure has also considerable inuence on tensile
strength of the joint. Fractography of fractured samples showed that formation of hot spots at high welding 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 which
involves the interaction of electrical, thermal, mechanical and metallurgical phenomena. In this process, the joining surfaces are kept
at a forced contact; followed by a high electric current passing
through the workpieces. Due to the contact resistance and Joule
heating, a vast amount of heat is generated at the faying surfaces.
Before, during and after applying the electric current, force is
applied to maintain the electric current continuity and to provide
the pressure necessary to form the weld zone. The metal at the
joint is heated to a temperature where recrystallization can rapidly
occur across the heated surfaces. In this process, similar to other
resistance welding processes, there is no requirement to any extraneous material such as ller material or shielding gas [1]. In this
welding process there are two types of resistances namely contact
resistance and bulk resistance. At the earlier stages of the welding,
contact resistance plays the main role but gradually it decreases
and the role of bulk resistance becomes more important [2,3].

Corresponding author. Tel.: +98 2161114104; fax: +98 2188006076.

E-mail address: halvaee@ut.ac.ir (A. Halvaee).
0261-3069/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2011.03.007

Kanne expressed that in comparison with fusion welding processes, the chemical composition and metallurgical properties are
not signicantly changed leading to better mechanical properties.
Simplicity, welding speed, capability of remote control and independence of welding quality from the operator skill are the other
advantages of this process [4]. Miyazaki et al., Kang, Kanne and
Sharitabar and Halvaee stated that resistance upset welding is a
suitable welding process for applications such as sealing of atomic
waste containers, welding of automotive parts and joining of stainless steels, low carbon steels, super alloys, aluminum alloys and
parts made of dissimilar materials [47]. The general conguration
of 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 total
stainless steel production. These stainless steels are preferred more
than other stainless steel types due to their good weldability [8].
But there are some negative metallurgical changes during welding
of 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 carbide
at grain boundaries and (e) formation of hot cracks.
Nikitin et al. and Nikitin and Bses stated that fatigue behavior of
austenitic stainless steel welds is strongly affected by stress ampli-

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

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

tude, temperature, frequency and welding conditions [11,12]. Most

of the service failures are expected to occur either in the HAZ or in
the weld metal. These failures are most frequently associated with
defects or microstructural in-homogeneities. But with variations in
welding conditions, changes in the type and the amount of defects
and in-homogeneities lead to variations in fatigue behavior of the
joint [13].
Plastic deformation of meta-stable austenitic steels leads to a
phase transformation from paramagnetic austenite to ferromagnetic martensite [14,15]. Smage showed that consequences of this
transformation for the application of these materials can be positive or negative. Increment of the strength, e.g. the transformation
induced plasticity (TRIP) effect and increase in the lifetime in the
high cycle fatigue (HCF) range are advantages in contrast to local
increase of the hardness and related reduction in ductility [14].
Because of high cooling rate, short welding time and formation
of the joint in solid state in resistance upset but welding, there is
possibility for elimination of some of these metallurgical changes
in welding of austenitic stainless steels by UW.
The literature in the upset welding eld is not very extensive.
The rst reported work on development of UW was the research
done at NASA Lewis Research Center. In This project Holko focused
on magnetic resistance upset welding of stainless steel 304 plates
with different thicknesses [16]. Resistance welding of nuclear
waste containers was another application of this technology which
required design of new equipment able to deliver currents of up to
400,000 A at 64,000 kgf. The same application was further reported
by Kanne [17]. He examined the properties of upset welded cylindrical and spherical components. He pointed out that advantages
of UW, compared to fusion welding processes, include fewer defects 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 upset
welding of high-speed steel to carbon steel with a current regulation system for controlling special cyclic welding [19]. The effect of
variation in upset butt welding parameters, such as current and
welding length on the hardness of different regions of the HAZ,
microstructure and toughness of the weld in high strength low-alloy steel weldment were studied by Ghosh and Gupta [20]. Miyazaki et al. examined the upset weldability of Nb-bearing high
strength steel of the 600 MPa level. They found that the higher
welding current density requires shorter upset length (the length

3855

that two samples penetrate into each other during welding) for
producing a high quality weld. Also they reported that the required
upset length can be reduced using lower welding forces [5]. Kanne
also reported applicability of the UW process to weld a variety of
stainless steels (including A-286), super alloys (including TD nickel), refractory metals (including tungsten) and aluminum alloys
(including 2024) [4]. Shieh and Chang presented a study of upset
welding process in wire drawing; obtaining the optimum parameters of the operation for a better distribution of hardness in the
wire [21]. Further, Cannel et al. wrote on the optimization and reliability of UW process [22]. In a study by Kang et al. the upset weldability and formability of a particular kind of material (SPCC) was
investigated. The results showed that the formability of upset
welded SPCC steel sheets were slightly lower than that of the parent material [11]. Applications of upset welding processes were recently extended to cast iron parts by Shakhmatov and Shakhmatov
and dissimilar austenitic to martensitic stainless steels by Sharitabar and Halvaee [7,23]. They found that a good metallurgical bond
can be produced between austenitic and martensitic stainless
steels by resistance upset welding. Also in the past decade some
work has been carried out on numerical simulation of resistance
upset welding. Recently, Kerstens and Richardson reported an
experimental study of weld development during resistance upset
butt welding process. They also made a simplied thermal nite
element model to explore the inuence of welding conditions on
heating [2]. In a very recent study Hamedi et al. considered numerical simulation and experimental investigation of UW process
parameters including heating and post-weld heating current and
their corresponding duration as well as interference of the part features that form the joint and effect of these parameters on tensile
strength of a low carbon low alloy oil pressure sensor. They found
that both numerical and experimental results suggest an optimum
set of welding parameters, i.e. time and electrical current that yield
a maximum value for the tensile strength of the joint. Also the effects of post-weld heating time and current on the tensile strength
showed that these parameters had a remarkable effect on improving tensile strength of the weldment [24].
In this research, resistance upset butt welding of 304 austenitic
stainless steel and effect of welding power and upset pressure on
microstructure, tensile and fatigue life of the joint were investigated in order to correlate the weld quality to the variation of these
two parameters and introducing optimum welding conditions.
2. Experimental procedure
Chemical composition of AISI 304 stainless steel used in this research 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 of
the steel used were 242 and 658 MPa respectively.
Start material was heated 10 min at 1060 C and cooled in air to
diminish cold work effects due to mechanical processing prior to
welding. The welding machine used in this research had been manufactured by Electro-TechnoTak Company (Tehran, Iran) and its
maximum power was 25 KVA. Then surfaces of samples were
ground by 1000 mesh grinding paper to remove oxide layer formed
during heat treatment. Two rods of 50 mm length and diameter of
8 mm were clamped in upset welding machine for each state of
welding. Table 1 shows welding conditions and measured parameters during welding. Primary and upset pressures were the same
and were applied by a mechanical system. Firstly, heating pressure
was applied on the faying surfaces. Then electrical current was
passed through the bars in contact. During welding, the electrical
potential was measured by an AVO meter and welding power
was calculated using following equation

P VI

3856

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

twin-matrix boundaries (Fig. 2a). After annealing, equiaxe grains

and annealing twins were formed in the microstructure (Fig. 2b).

Table 1
Selected conditions and measured parameters during welding.
Sample
name

Welding
current (A)

Welding
pressure (MPa)

Welding
time (S)

Heat input
(V.A.S)

A1B1
A2B1
A3B1
A4B1
A1B2
A2B2
A3B2
A4B2
A1B3
A2B3
A3B3
A4B3
A1B4
A2B4
A3B4
A4B4

1500
2000
2500
3000
1500
2000
2500
3000
1500
2000
2500
3000
1500
2000
2500
3000

1.01
1.01
1.01
1.01
1.15
1.15
1.15
1.15
1.27
1.27
1.27
1.27
1.41
1.41
1.41
1.41

1.5
1.42
1.33
1.21
1.51
1.41
1.33
1.19
1.5
1.40
1.31
1.20
1.49
1.42
1.32
1.21

4500
5680
6650
7260
4530
5640
6650
7140
4500
5600
6550
7200
4470
5680
6600
7260

where P is the welding power (VoltAmpere), I is the current intensity (Ampere) and V is the electrical voltage (Volt) [2]. Also, welding
time which is the passing time of electrical current was measured
by the AVO meter.
Tensile test was carried out by MST30/MH machine at 2 mm/
min displacement rate on welded samples. This test was performed according to ASTM-E8 standard [25] while the joint interface was held in the middle of the tension samples and the ash
was removed. The examination method in fatigue investigation
was the single factor method, i.e. for each series of experiments
one factor was varying while the other parameter was kept constant at previously optimized levels of tensile strength. The aim
of the authors was to investigate the effect of UW process parameters on fatigue properties of 304 austenitic stainless steel joints. In
other words, the authors wanted to see how variations in welding
parameters can affect the fatigue life of the joint. For this purpose,
three different stress amplitudes were selected higher than yield
strength of the alloy to represent the results of the tests as SN
curves. The Pullpush fatigue test was performed according to
ASTM-E60692 standard [26] while the joint interfaces were held
in the middle of the samples. These test was carried out by INSTRON 8502 fatigue testing machine at R = 1, frequency of 2 Hz
and 320, 370 and 430 MPa stress amplitudes. The metallographic
samples were cut longitudinally and the interfaces were studied
by SEM Cam Scan MV2300 and Energy dispersive X-ray lined up
to the SEM after preparation and etching with Kalling No. 2 agent
(5 g CuCl2, 100 ml HCl and 100 ml Ethanol). XRD test was fullled
by wave length of Coka 1:7889 nm: Finally fracture surfaces were
studied by SEM.

3. Results and discussion

Leber et al. showed that cold work causes formation of nonhomogeneities such as shear bands, mechanical twins and deformation induced martensite in the microstructure of austenitic
stainless steels. Presence of non-homogeneities in the microstructure leads to decreasing corrosion resistance of these alloys [15].
Therefore in many applications, austenitic stainless steels are used
in normalized heat treating condition after cold working. Therefore, in this research normalizing heat treatment was performed
on samples before welding to improve corrosion resistance.
Fig. 2a and b shows microstructure of base metal before and after
normalizing respectively. It is observable that the microstructure
consists of high density of mechanical micro and macro twins. Also
some transformation induced martensite is formed in twins and

3.1. Study of microstructure

Fig. 3a shows macrostructure of half of the joint interface in
sample A1B3. Fig. 3bd shows the microstructures of different
zones formed at the joint interface. As can be seen, three different
microstructural zones have been formed in the interface due to
thermal gradient between the joining faces and electrodes. They
are:
1. Widmansttten austenite formation zone (WAZ), an elliptical
zone at the center of the joint interface consisted of different
morphologies of austenite (Fig. 3b). In addition to allotriomorphic austenite (Ac) localized in grain boundaries and intergranular austenite (Ic), unusual austenite morphology is found.
According to literatures, this microstructure is Widmansttten
austenite (Wc) [27,28]. Small amounts of lathy d-ferrite could
be found within Widmansttten austenite laths. The Widmansttten austenite structure is more common in austenitic
stainless steels solidifying as d-ferrite [28]. Woollin, expressed
that Like Widmansttten ferrite found in carbon steels, appearance of Widmansttten austenite is one of narrow wedges emanating either directly from a grain boundary or from
allotriomorphic ferrite (allotriomorphic austenite in case of
Widmansttten austenite) [27].
2. Dynamic recrystallization zone (DRZ) around the WAZ composed of ne austenite grains and delta ferrite phase is formed
at the grain boundaries (Fig. 3c).
3. Partially recrystallization zone (PRZ) which contains recrystallization lines along the drawing direction during manufacturing
of 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 than
bulk resistance of the materials, during welding contact surfaces
are hotter than bulk of the samples and their temperature rose
up to liquid + delta ferrite zone in FeCrNi phase diagram. When
the upset pressure was applied, the liquid metal at the edges of
contact surfaces of samples was rejected as ash and was replaced
with mushy metal led to formation of upset in contact area. But in
the center of the contact surfaces, the liquid metal was trapped and
solidied. Due to high Cr equivalent to Ni equivalent ratio (Creq/
Nieq = 1.91), the solidication microstructure is fully ferrite
[28,29]. In austenitic stainless steels, Delta ferrite is not stable at
room temperature. So it transforms to austenite phase during cooling. Southwick and Honeycombe concluded that decomposition of
d-ferrite to austenite occurs by two different mechanisms depending upon the transformation temperature. At high temperature, the
reaction occurs by a diffusional nucleation and growth process
whereas at low temperature the austenite phase forms by a displacive mechanism. It is believed that the Widmansttten austenite
grows by a displacive mechanism whereas allotriomorphic austenite is considered to be a reconstructive transformation product
[30]. Menezes et al. have reported that in bead on plate welding
of two-phase ferriticaustenitic stainless steels, residual compressive stresses were formed near the ferrite to austenite transformation temperature. These stresses increased the probability of
formation of Widmansttten austenite [31]. During solid state
transformation of ferrite to austenite in resistance upset welding
of this steel, allotriomorphic austenite formed in ferrite grain
boundaries. But the transformation across the entire grain was
suppressed by high cooling rate of upset butt welding resulting
low diffusion rate and low driving force due to low transformation

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

3857

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.

temperature. This caused residual ferrite to transform to Widmansttten austenite by displacive mechanism. Compressive stresses during transformation in this welding process also encouraged
ferrite to Widmansttten austenite transformation (Fig. 3b).
Joining surfaces have the highest temperature during welding
and with increasing distance from the weld interface, temperature
decreases [2,28]. Around the interface, temperature rose up to austenite + delta ferrite in FeCrNi phase diagram resulting to the
formation of ferrite at austenite grain boundaries (Fig. 3c). Because

of high cooling rate of the joint, the possibility of ferrite to austenite transformation was low and some ferrite was remained in grain
boundaries. So microstructure of this region consisted of austenite
and delta ferrite [27]. Fuller et al. concluded that presence of d-ferrite in grain boundaries prohibits grain growth and so the grains
are ne. Also this phase acts as a crack growth inhibitor and reduces the possibility of intergranular fracture [32]. But Lippold
and Kotecki expressed that presence of delta ferrite decreases
formability of austenitic stainless steels and increases probability

3858

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

of precipitation of carbides [28]. On the other hand, dynamic

recrystallization due to hot deformation may be one of the most
important factors in decreasing the grain size in this region as
mentioned by Humphreys [33].
Fig. 3d shows how recrystallization is limited to a series of lines
along the drawing direction during manufacturing of the bars and
delta ferrite is formed along some of these lines. It is observed that
direction of partial recrystallization lies on the direction of the
shear bands formed due to mechanical working during manufacturing of the bars. Cizek stated that these bands are composed of
ne dislocation containing cells which grow parallel to each other
and pass through the grains. These cells have high angle non-crystallographic grain boundaries with the matrix [34]. The possible
reason for formation of this region is that heat treatment in this research could not remove these cells due to high density of dislocations 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 dynamic recrystallization along these bands. But around the bands,
possibility of dynamic recrystallization was low due to low density
of dislocations. Also because of the high dislocation density in
these shear bands; diffusion rate of ferrite promoting elements
such as chromium and segregation of these elements was high
leading to formation of high temperature delta ferrite along the
shear bands and in austenite grain boundaries [28].
Microstructural analysis of samples welded according to different welding conditions represented in Table 1 showed that increasing of welding power at a constant welding pressure (e.g. samples
A1B2A4B2) caused widening all different zones formed in the joint
interface. As stated before, higher welding power produced higher
heat at the joint interface resulting to the formation of considerable amount of liquid metal at the interface and widening the
WAZ. Also, The heat generated by bulk resistance during welding
increased at higher welding powers leading the wider zone of base
metal temperature to rise to austenite + ferrite region in FeCrNi
phase diagram. This caused widening the zone that can dynamically recrystallized during welding according to following equation

Z e0 expQ =RT

between the joint interface and bulk of the samples. This reduced
the cooling rate leading to decreasing the amount of grain boundary
ferrite near the interface and grain growth in this region.
By using higher welding pressure at constant welding powers
(e.g. samples A2B1A2B4), width of the WAZ decreased. But it
caused widening DRZ and PRZ. Higher welding pressures lower
the interface resistance according to Eq. (3) and increase the roll
of bulk resistance on heat generation during welding and formation of lower amount of liquid at the interface.

Rc q=2pHB=F ups 1=2

where Rc is contact resistance, HB is Brinell hardness, q is specic

resistivity and Fups is upsetting force [2]. Also, high welding pressure
caused rejection of liquid metal formed at the interface during
upsetting and decreased width of the WAZ at high welding pressures. On the other hand, increasing the roll of bulk resistance on
heat generation at high pressures led to widening the area that
can dynamically recrystallize according to following equation

Z C 1 sinhC 2 rn

where C1, C2 and n are constants and r is stress exerted during hot
deformation [33]. According to this equation, higher welding pressures increased the possibility of dynamic recrystallization and
widening 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 regions
and no precipitation of chromium carbides (Cr23C6) phase is observed. This is because of high cooling rate of weld interfaces from
the 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. It
can be seen that the amount of Cr increased and Ni decreased in
this phase and its chemical composition is in the ferrite region of
the FeCrNi phase diagram at room temperature [28].

where Z is ZenerHolloman parameter, e is strain rate, Q is activation energy, R is gas constant and T is temperature. According to this
equation, higher temperature during hot deformation reduces Z and
increases the possibility of dynamic recrystallization and therefore
widening DRZ and PRZ [33]. On the other hand, higher heat generated by bulk resistance of samples decreased temperature gradient

3.3. Mechanical properties

3.3.1. Tensile properties
Fig. 6 shows that tensile strength of the joint decreases with
welding power. Welding power of 3 KVA maintained enough heat

Fig. 4. XRD analysis of the joint interface.

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

3859

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

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

Cr

Ni

Si

Mn

Fe

Wt%

23.93

4.71

0.46

1.61

Bal

to produce a mushy zone and a complete metallurgical joint. But

with increase in welding power, the grains within the HAZ graw
and the joint strength decreased according to HallPetch equation.
On the other hand, higher heat input produced at high welding
powers increased residual liquid at the joint face and produced
considerable amount of Widmansttten austenite phase. Plate
form of this phase increased stress concentration in this region
and therefore decreased joint strength.
Formation of hot spots at the joint interface also reduced joint
strength at high welding powers. Fig. 7 shows a hot spot in the center of the joint interface of sample A4B1 formed due to heterogeneous distribution of electrical current and heat.
Using nite element method (FEM) and experimental investigations, Kerstens and Richardson [2] showed that heterogeneous
heating and formation of hot spots occur because of non-uniform
current density passing through the material. This heterogeneity

can be formed by in-homogeneities in the material and local variations in the interface resistance resulting from contaminations,
non-uniform deformation or surface imperfections. A non-uniform
upsetting 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 result
of contact resistance variations at the electrode/sample interface
due to the electrode surface condition and contaminations or variations of the clamping force. With increasing of welding power, the
probability of hot spot formation rises due to increase in the heat
input 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 this
steel. It can be seen that formation of hot spot at the interface of
this sample caused crack initiation and reduced the joint strength.
Fig. 9 shows that with increasing of the upset pressure, strength
of the joint increases rstly and then decreases. Contact resistance
rises with decrease in upset pressure according Eq. (3). Experimental investigations of Song et al. [3] also showed that contact resistance increases with decreasing of welding pressure. So, low
welding pressures leads to widening the area for formation of Widmansttten austenite and decreasing joint strength [35]. Also at

Fig. 6. Effect of welding power on tensile strength of the joint.

3860

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

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

low welding pressures, good metallurgical bond between samples

did not happen and therefore the strength of the joint is low. With
increasing of upset pressure up to 1.27 MPa, amount of liquid
metal formed at the joint interface decreased. Also, large amount

of liquid was rejected in upsetting stage due to high welding pressure. This caused thinning the area for formation of Widmansttten austenite and increasing joint strength (sample A1B3). Also
increase in upset pressure led to formation of good metallurgical
joint. By using welding pressure higher than 1.27 MPa; all liquid
and mushy metals were rejected from the joint interface as ash
and an incomplete joint was formed which reduced the joint
strength (sample A1B4). On the other hand, effect of welding pressure on tensile strength decreased with increasing of welding
power (Fig. 9). This may be due to the formation of hot spots at
high welding powers. Because of the presence of welding defect
at the joint area, effect of microstructure on mechanical properties
was reduced led to decreasing the effect of welding pressure on
tensile strength at high welding powers.
Fig. 10 shows fracture surface of sample A3B1 (welded at low
welding pressure). In Fig. 10a there are different crack initiation
sites at the fracture surface. The higher magnications of zones
13 are shown in Fig. 10bd respectively. It is observed that presence of plate form Widmansttten austenite phase at the joint
interface caused formation of large voids due to stress concentration and reduced the joint strength at low welding pressures.
3.3.2. Fatigue properties
Fig. 11 show the effect of welding power on fatigue life of the
joint at different stress amplitudes in samples welded with
1.27 MPa welding pressure. It is observed that fatigue life of
welded samples is lower than base metal and decreases slightly

Fig. 8. Fracture surface of sample A4B2, a: lower magnication b: higher magnication showing hot spot on this surface.

Fig. 9. Effect of upset pressure on tensile strength of the joint.

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

3861

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

with increasing of welding power from 3000 to 5000 V.A. But in

samples welded with welding power of 6000 V.A, fatigue life decreases remarkably in all stress amplitudes. Slight decrease in fatigue strength with increasing of welding power from 3000 to
5000 V.A may be due to grain growth in the HAZ of welded samples. As stated by Hertzberg, according to HallPetch equation,
coarse grain size reduces tensile, fatigue strength and consequently
fatigue life [34].
As stated before, the probability of hot spot formation increases
at 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 solidied microstructure and presence

of cracks into the hot spots, stress concentration decreases the fatigue strength. Fig. 12a shows the main crack initiation site on the
fatigue fracture surface of the sample welded at high welding
power (6000 V.A) and tested at 320 MPa stress amplitude. In the
Fig. 12b and c the higher magnication of this site and dendrites
arisen from melting and solidifying in the hot spots are shown
respectively. Therefore it can be concluded that formation of hot
spots is the main reason in decreasing of the fatigue life at welding
power of 6000 V.A.
Variation in welding pressure at welding power of 3 KW had no
considerable inuence on fatigue life of the joints and therefore it
is not represented here.
Fig. 13a and b show the fatigue fracture surfaces at the stage II of
fatigue crack propagation in samples A2B3 tested at 320 and
430 MPa stress amplitudes respectively. There are secondary cracks
on the fracture surface of the sample tested at 320 MPa (Fig. 13a).
But no secondary cracks are observed at the fracture surface of the
sample tested at 430 MPa (Fig. 13b). Observation of cracks could
be associated with the partially transformed martensite phase in
austenite matrix. Because of non-uniform microstructure, local
stresses may be concentrated at these locations causing secondary
cracks to initiate [36]. But at the stress amplitude of 430 MPa, the
phenomenon of the self-heating of the specimens was much more
pronounced and affected deformation behavior of the sample. Surface temperature of the samples tested at 320, 370 and 430 MPa
stress amplitudes was measured by thermocouples. It is observed
that 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) is
the temperature at which 50 vol% a-martensite is formed in this

3862

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

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.

steel after a true tensile strain of 30%) [8]. This relatively high temperature in sample tested at 430 MPa stress amplitude inhibited
martensite transformation. In addition, plastic deformation by
migration of dislocations was facilitated at these relatively high
temperatures (thermal activation).
Fig. 14 shows the sub-surface fatigue crack formed in the Widmansttten austenite formation zone. Fatigue process and its
mechanisms are largely inuenced by the presence of the material
in-homogeneities. Since d-ferrite is basically different from austenite matrix in crystallography and chemical composition, it is likely
to provide crack nucleation sites. Goyal et al. [13] conducted a

series of experiments to investigate the effect of d-ferrite on the

continuous cycling fatigue properties. Using nite element model
(FEM) calculation and transmission electron microscopy (TEM)
studies, they showed that stress concentration at the delta/gamma
interface occurs due to incompatibility and consequently acts as
crack initiation site. Therefore, presence of d-ferrite between the
Widmansttten plates is one of the reasons for formation of
sub-surface cracks. Also plate form of Widmansttten phase led
to stress concentration in this region and increased probability of
sub-surface crack formation. Fatigue crack formed in the dynamic
recrystallized zone is shown in Fig. 15. It is observed that there is

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

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

3863

2. Increase of welding power raises heat input to the joint area

and widens all different microstructural zones at the joint interface. Also higher heat input increases probability of hot spot formation. Tension tests results showed that tensile strength of the
joint decreases with increase of welding power and hot spots
formed at high welding powers are the most important factors
in decreasing of the joint strength.
3. X-ray diffraction analysis showed that there is no precipitation
of chromium carbide in the HAZ due to high cooling rate of the
joint area from the chromium carbide formation temperature.
4. With increasing in welding pressure, area for formation of Widmansttten austenite decreases leading to higher tensile
strength of the joint. But effect of welding pressure on tensile
strength decrease at high welding powers due to formation of
hot spots.
5. Fatigue test results indicated that fatigue life of the joint
decreases with welding power due to grain growth in the
HAZ. Formation of hot spots is the other reason for decreasing
fatigue strength at high welding powers. Microstructural analysis of the fatigue samples showed that presence of d-ferrite
between Widmansttten austenite plates can initiate fatigue
crack and dose not have any considerable effect on crack path
at high stress amplitudes.

Acknowledgments
The authors gratefully acknowledge the extensive support of
the Electro-Techno Tak Company and University of Tehran for
experimental and nancial supports.
References

Fig. 15. Fatigue crack growth mode in the dynamic recrystallization zone in sample
A3B2 tested at 320 MPa stress amplitude.

no fatigue crack deection by d-ferrite phase in this region. Therefore it can be concluded that presence of d-ferrite in austenitic
stainless steel welds causes fatigue crack initiation and does not
have any considerable effect on crack path at high stress
amplitudes.
4. Conclusions
In the present investigation, resistance upset butt welding of
304 austenitic stainless steel and effect of welding power and upset pressure on microstructure, tensile strength and fatigue properties of the joint were investigated. The obtained results can be
summarized as follows:
1. In resistance upset butt welding of 304 stainless steel, three different microstructural zones were formed at the joint interface
due to thermal gradient between the joining faces and electrodes. These zones are: Widmansttten austenite formation
zone, dynamic recrystallization zone and partially recrystallized
zone. Also delta ferrite phase formed in these regions.

[1] Brien RL. Editor. Welding Handbook. Miami (FL, USA): American Welding
Society: 1991.
[2] Kerstens NFH, Richardson IM. Heat distribution in resistance upset butt
welding. J Mat Proc Tech 2009;209:2715872.
[3] Song Q, Zhang W, Bay N. An experimental study determines the electrical
contact resistance in resistance welding. Weld J 2005:736.
[4] Kanne Jr WR. Solid state resistance upset welding: a process with unique
advantages for advanced materials. In: The 2nd advanced joining technologies
for new materials conference, Cocoa Beach: FL; 1994
[5] Miyazaki Y, Ichikawa M, Saito T. Upset weldability of niobium-bearing high
strength 600 MPa steel for wheel rims and mechanism of upset weldability.
Weld World 1993;31(5):34857.
[6] Kang SS. A study of resistance welding in steel sheet, using a tailor-welded
blank, (Report1), evaluation of upset weldability and formability. J Mat Proc
Tech 2000;101(12):18692.
[7] Sharitabar M, Halvaee A. Resistance upset butt welding of austenitic to
martensitic stainless steels. J Mater Des 2010;31(6):304450.
[8] Padilha AF. Decomposition of austenite in austenitic stainless steels. ISI J Int
2002;42(4):32537.
[9] Mumin S. Evaluation of joint interface properties of austenitic stainless steels
(AISI 304) joint by friction welding. J Mat Des 2007;28:224450.
[10] Kou S. Solidication and liquation cracking issues in welding. Jom 2003:3742.
[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;50:13504.
[12] Nikitin I, Bses M. Effect of low frequency on fatigue behavior of austenitic steel
AISI 304 at room temperature and 25 C. Int J Fat 2008;30:20449.
[13] Goyal S, Sondhya R, Vaslon M, Rao BSK. The effect of thermal aging on low
cycle fatigue behavior of 316 stainless steel welds. Int J Fat 2009;31:44754.
[14] Smage M, Walter F, Eier D. Deformation induced martensite transformation
in metasteble austenitic steel. J Mat Sci Eng 2008;483484:3947.
[15] Leber HJ, Niffeneger M, Tribonod B. Microstructural aspects of low cycle
fatigued Austenitic stainless tube and pipe steels. J Mat Charact
2007;58:100615.
[16] Holko KH. Magnetic force upset welding dissimilar thickness stainless steel tee
joints. Weld J Suppl 1970;49(9):42739.
[17] Kanne Jr WR. Solid state resistance welding of cylinders and spheres. Weld J
1986;65(5):338.
[18] Cannell GR, Sessions CE. Proper procedures are the key to welding radioactive
waste canisters. Weld J 1997;76:617.
[19] Bezprozvannyi IA, Shirokovskii RM, Masalov YUA. Technology of resistance
butt welding of metal-cutting tool billets and their heat treatment. Automat
Weld 1989;11:626.

3864

M. Sharitabar et al. / Materials and Design 32 (2011) 38543864

[20] Ghosh PK, Gupta PC, Goswami TK. Inuence of some upset butt welding
parameters on the weld properties of HSLA steel. Indian Weld J
1989;21(1):42836.
[21] Shieh JK, Chang TL. The optimum upset welding parameters of alloy steel wire.
Technol Train 1996;21(3):12938 [Taiwan].
[22] Cannel GR. Upset welding assures a leak-tight, structurally sound seal. Weld J
1999;78:414.
[23] Shakhmatov MV, Shakhmatov DM. Features of resistance butt welding of cast
iron. Weld Int 2004;18(9):73741.
[24] Hamedi M, Eisazadeh H, Esmailzadeh M. Numerical simulation of tensile
strength of upset welded joints with experimental verication. J Mater Des
2010;31:2296304.
[25] ASTM E8. Standard Test Methods of Tension Testing Metallic Materials. Annual
Book 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 supermartensitic
stainless steels. In: Proceedings of supermartensitic stainless steels
conference, Brussels, Belgium; 2002. p. 199204.
[28] Lippold JC, Kotecki DJ. Welding metallurgy and weldability of stainless
steels. John Wily & Sons; 2005.

[29] Shankar V, Gill TPS, Mannan SL, Sundarlsan S. Solidication cracking in

austenitic stainless steel welds. Sadhana 2003;28(24):35982.
[30] Southwich PD, Honeycomb RWK. In: International conference of martensitic
transformation, Boston: MIT Press; 1979. p. 18994 [ICOMAT79].
[31] Menezes JWA, Abreu H, Kundo S, Bhadeshia HKDH, Kelly PM. Crystallography
of Widmansttten austenite in duplex stainless steel weld metals. J Sci Tech
Weld Joi 2009;14(1):410.
[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 Analysis
2008;15:83546.
[33] Humphreys FJ, Hatherly M. Recrystalization and related annealing
phenomena. University of Manchester, Institute of Science and Technology
UK: Elsevier Ltd; 2004.
[34] Cizek P. Characteristic of shear bands formed in austenitic stainless steels
during hot deformation. J Mat Sci Eng A 2002;324:2148.
[35] Hertzberg RW. Deformation and fracture mechanics of engineering
materials. John Wiley& Sons; 1996.
[36] Haskalik A, Unal W, Ozdemir N. Fatigue behavior of AISI 304 steel to AISI 4340
steel welded by friction welding. J Mat Sci 2006;41:32339.