Optics & Laser Technology 88 (2017) 24–39

Contents lists available at ScienceDirect

Optics & Laser Technology

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

Full length article

Comparative study of pulsed Nd:YAG laser welding of AISI 304 and AISI
316 stainless steels
Nikhil Kumar a,n, Manidipto Mukherjee b, Asish Bandyopadhyay a
a
Mechanical Engineering Department, Jadavpur University, Kolkata 700032, India
b
Metallurgical and Materials Engineering Department, Jadavpur University, Kolkata 700032, India

art ic l e i nf o a b s t r a c t

Article history: Laser welding is a potentially useful technique for joining two pieces of similar or dissimilar materials
Received 3 April 2016 with high precision. In the present work, comparative studies on laser welding of similar metal of AISI
Accepted 16 August 2016 304SS and AISI 316SS have been conducted forming butt joints. A robotic control 600 W pulsed Nd:YAG
Available online 12 September 2016
laser source has been used for welding purpose. The effects of laser power, scanning speed and pulse
Keywords: width on the ultimate tensile strength and weld width have been investigated using the empirical
Laser welding models developed by RSM. The results of ANOVA indicate that the developed models predict the re-
Response surface methodology sponses adequately within the limits of input parameters. 3-D response surface and contour plots have
Modeling and optimization been developed to find out the combined effects of input parameters on responses. Furthermore, mi-
Microstructure
crostructural analysis as well as hardness and tensile behavior of the selected weld of 304SS and 316SS
Mechanical properties
have been carried out to understand the metallurgical and mechanical behavior of the weld. The se-
lection criteria are based on the maximum and minimum strength achieved by the respective weld. It has
been observed that the current pulsation, base metal composition and variation in heat input have
significant influence on controlling the microstructural constituents (i.e. phase fraction, grain size etc.).
The result suggests that the low energy input pulsation generally produce fine grain structure and im-
proved mechanical properties than the high energy input pulsation irrespective of base material com-
position. However, among the base materials, 304SS depict better microstructural and mechanical
properties than the 316SS for a given parametric condition. Finally, desirability function analysis has been
applied for multi-objective optimization for maximization of ultimate tensile strength and minimization
of weld width simultaneously. Confirmatory tests have been conducted at optimum parametric condi-
tions to validate the optimization techniques.
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction petroleum, petrochemical, food processing and pulp industries as

well as power generating plants due to its high corrosion re-
Laser welding is a high-energy-density welding process. It is sistance. There are large number of applications of AISI 304 and
favored for its deep penetration, quick, narrow heat-affected zone AISI 316 stainless steel in industries, hence laser welding of these
(HAZ), well welding seam quality. This welding technique can be stainless steel have been studied.
applied to weld different thicknesses of sheets/plates, similar/ Wu et al. [1] have analyzed CO2 laser welding of vehicle body
dissimilar materials and different surface materials together so as by joint microstructure and stress-strain plot of specimen. Khan
to make integral structure of large-sized overlay for automobile, et al. [2] have presented an experimental design approach to
microelectronic, marine, aerospace industries. Laser beams can be process parameter optimization for the welding of martensitic AISI
either continuous or pulse type. Continuous laser systems are 416 and AISI 440 FSe stainless steel. Kong et al. [3] have studied
mostly used for very deep welding whereas pulse lasers are used the effect of laser welding on the zinc coated galvanized high
to weld very thin materials. Austenitic stainless steels, known as strength steel. Dongxia et al. [4] have done laser welding with
300 series, have become a cost-effective, stable material for long- filler wire successfully to joining of a new-type Al-Mg Alloy. Liao
term applications in many industrial sectors including gas, and Yu [5] have done experimental study of laser spot welding on
stainless steel sheets. Kim et al. [6] have used robotic control disc
n
laser welding system to enhance the flexibility of laser welding
Corresponding author.
E-mail addresses: [email protected] (N. Kumar),
process. Khan et al. [7] have studied laser beam welding of dis-
[email protected] (M. Mukherjee), similar AISI 304L and AISI 430 stainless steel. Yang and Lee [8]
[email protected] (A. Bandyopadhyay). have presented the strength difference between laser-spot and

http://dx.doi.org/10.1016/j.optlastec.2016.08.018
0030-3992/& 2016 Elsevier Ltd. All rights reserved.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 25

resistance-spot welding process. Chen et al. [9] have studied the microstructure and mechanical properties of the welded
joining mechanism of Ti/Al dissimilar alloy laser welding-brazing joints. However, there is little or no information available in the
process with automated wire feed. Balasubramanian et al. [10] literature on the comparative analysis of microstructure and me-
have studied laser beam welding of AISI 304 stainless steel using chanical properties of pulsed Nd:YAG laser welded joints of AISI
2 kW CW Nd:YAG laser. Nemecek et al. [11] have analyzed the 304 and AISI 316 stainless steel sheet by statistical approach.
difference in microstructures and properties of HSS (high speed In this paragraph the state-of-the-art review regarding laser
steel) joint made by MAG (Metal Active Gas) and laser welding. welding has been presented. An investigation has been made by
Um and Stroud et al. [12] have calculated the total energy con- Schweier et al. [32] to study the spatter formation in laser beam
sumption by a robot arm, laser source and cooling system. Cha- welding with superposed beam oscillation. They have derived the
minade at al. [13] have investigated the specific temperature empirical model for the prediction of the number of spatters as a
profile of a high power diode laser soldering process. Zhang et al. function of laser power, welding velocity, oscillation amplitude
[14] have applied sandwich method to observe the keyhole in and frequency. In the research of Chen et al. [33] the spectroscopic
deep penetration laser welding. Sibillano et al. [15] have studied measurements of electron temperature and electron density of the
the correlation between plasma electron temperature and pene- keyhole plasma and plasma plume in deep penetration laser
tration depth in laser welding processes. Abdullah et al. [16] have welding condition has been studied. Colombo et al. [34] have es-
reported the effect of incorporating pre- and post-heating on the tablished solution for the on-line monitoring of the gap in remote
mechanical properties of laser-welded joint of mild steel and fiber laser welding of overlapped zinc-coated steel and the new
stainless steel. Berretta et al. [17] have investigated laser welding monitoring architecture Through Optical Combiner Monitoring
of AISI 304 to AISI 420 stainless steel. Bauchspiess et al. [18] have (TOCM) has been presented and tested. In the work of Erdos et al.
investigated an on-line tracking optimization system for sensor [35] the discussion on technical background of the remote laser
guided robotic manipulators by associating sensor information. welding technology has been done. They have develop a complete
Tsirkas et al. [19] have developed a three dimensional finite ele- off-line programming toolbox for remote laser welding which can
ment model for laser welding process and also predict laser wel- provide an automated method for computing close-to-optimal
ded panel distortions. Sibillano et al. [20] have developed the robot programs. A systematic analysis of the effects of laser power,
spectroscopic monitoring of penetration depth in CO2, Nd:YAG and welding speed and beam diameter on depth of penetration and
fiber laser welding processes. Kim and Ahn [21] have presented weld profile of bead-on-plate laser welds has been presented by
coaxial monitoring of the keyhole generated by an Yb:YAG laser Suder and Williams [36]. A theoretical calculation of the melt
beam during lap welding of steel and Al sheets. Kawahito et al. thickness on the keyhole front is also presented by Eriksson et al.
[22] have developed a relationship of laser absorption to keyhole [37]. The Results indicate that the thickness of the liquid on the
behavior in high power fiber laser welding of stainless steel and keyhole front is similar to that of the resolidified layer found in
aluminum alloy. Zhao et al. [23] have done optimization of laser micrographs of the front if the laser is suddenly turned off. Olabi
welding of thin-gage galvanized steel via response surface meth- et al. [38] have studied the laser-butt welding of low carbon steel
odology. Sathiya et al. [24] have optimized the laser welding and austenitic steel AISI 316 using CW 1.5 kW CO2 laser. The effect
process parameters for super austenitic stainless steel using arti- of laser power, welding speed and focal point position on me-
ficial neural networks and genetic algorithm. Alam et al. [25] have chanical properties (i.e., ultimate tensile strength, UTS and impact
generalized the fatigue stress analysis of different laser weld strength, IS) and on the operating cost has been also investigated.
geometries. Du et al. [26] have studied the metal flow in full pe- Hence, in the present work, an attempt has been made to
netration laser beam welding of titanium alloy. compare the laser welding of 304SS and 316SS. For this purpose
Apart from the bead geometry, microstructure and mechanical 3 factors-5 levels experiments have been planned using response
properties of the laser welded joints are also affected by the var- surface methodology (RSM) design matrix and analyzing the re-
iation in process parameters. There are many reports which deal sponses of interest by developed mathematical models based on
with the shape and solidification structure of the fusion zone of experimental results. The input parameters and their limits are
laser beam welds in relation with different parameters. El-Batahgy same for both 304SS and 316SS. The second order mathematical
[27] studied the fusion zone shape and solidification structure of equations have been developed for predicting the desired weld
304L, 316L and 347 austenitic stainless steels as a function of CO2 quality. In spite of statistical evaluation of the welded joints, me-
laser parameters and found that the heat input as a function of tallurgical and mechanical analyses have been carried out on the
both laser power and welding speed has almost no effect on mi- selected specimens of 304SS and 316SS weld.
crostructure and mechanical properties of welds. However, it has
been also reported that higher laser power and lower welding
speed resulted in coarse dendritic structure in the fusion zone. In 2. Response surface methedology (RSM)
general, the microstructure of laser welded fusion zone is pre-
dominantly austenite with small amount of δ ferrite. Usually the Response surface methodology (RSM), is a useful design of
amount of δ ferrite varies between 3‐8 vol.% to prevent solidifi- experiment method that is gaining popularity. This includes a re-
cation cracking in austenitic stainless steels [28]. However, during view of basic experimental designs for fitting linear response
the non-equilibrium rapid solidification condition in laser welding, surface models, in addition to a description of methods for the
the high cooling rate will result in incomplete δ-γ transformation determination of optimum operating conditions. The steps of re-
and some metastable δ-Fe should be remained unavoidably and sponse surface methodology are:
depending on the heat input of the process ferrite to austenite
ratio varies in the fusion zone [29]. Again, the mode of solidifica- (i) Developing experimental strategy for selecting independent
tion may alter depending on the fusion zone chemical composi- variables.
tion. The chemical composition of the fusion zone mostly depends (ii) Statistical modeling to build an approximate relationship be-
on the four factors i.e. base material composition, joint config- tween the response and process variables.
uration, welding procedure and filler material composition. It is (iii) Optimization for finding values of process variables producing
well known that any fusion zone can experience four different desirable values of the response.
modes of solidification depending on the Creq/Nieq ratio [30,31].
Furthermore, different welding processes can significantly alter When all the independent variables are measurable,
26 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Table 1
Chemical compositions (wt%) of AISI 304 and AISI 316 stainless steel.

Type Chemical composition

C% Si% Mn% P% S% Cr% Ni% Mo% Cu% Nb% Al% N%

304SS 0.079 0.2858 1.8 0.032 0.0194 18.56 8.2002 0.2654 0.2920 0.0281 0.0063 –
316SS 0.081 0.4 1.7 0.025 0.012 18.66 11 2.6 0.08 0.03 0.001 0.07

controllable and continuous during experiments, response surface, most important [39].
y can be expressed with negligible error by:

y = f ′ ( x)β + ϵ (1)
3. Experimental set up and procedure
where
x ¼(x1, x2,.,xk), In this experimental work AISI 304 and AISI 316 stainless steel
f ′ ( x ) ¼a vector function of p elements. work piece of 100 mm  20 mm  1.5 mm thick sheet are chosen.
β ¼a vector of p unknown constant coefficients, The chemical compositions of the materials are shown in Table 1.
ϵ ¼ a random experimental error assumed zero mean. The schematic diagram of laser welding process is shown in Fig. 1.
In RSM, an approximate model is needed to develop for the All the experiments have been conducted on JK600HP Nd:YAG
true response surface. The approximated model is constructed laser generator (GSI, UK) integrated with ABB IRB 1410 robotic
utilizing observed data from the process or system. Multiple re- control. The welding of work piece has been conducted in pulse
gression analysis is commonly used for this. Usually, a second- wave regime and repetition rate of 25 Hz. Experimental set-up is
order polynomial equation is used in RSM, which is given by given in Fig. 2. Samples are butt jointed and during welding
technically zero gap between two sheets is maintained in each
k k
case. An argon gas jet emerges from the side nozzle which makes a
y=β0 + ∑ βixi+ ∑ ∑ βij xi xj+ ∑ βiix2i + ε
i=1 i=1 (2) fixed angle with the laser beam to avoid any external atmospheric
contamination during welding. The laser beam has a spot size
where parameters β0 , βi, βij, βii are called regression coefficient for 0.7 mm. The ranges of input parameters are selected on the basis
i¼0,1,...,k and j¼0, 1...k. of trial experiments conducted by using one factor at a time ap-
proach. The chosen process parameters and their limits are given
2.1. Desirability function analysis in Table 2. Twenty experiments for each stainless steel (304SS and
316SS) have been conducted as per central composite rotatable
It is an approach in which, individual responses are trans- design (CCD). Statistical software Design Expert 10 has been ap-
formed to corresponding desirability values. Desirability value plied to establish the design matrix. Fig. 3 shows a weld sample in
depends on acceptable tolerance range as well as target of the butt joint configuration. Samples for the metallographic ex-
response. Unity is assigned, as the response reaches its target va- aminations have been prepared by polishing successively in 80,
lue, which is most desired situation. Beyond acceptable limit, de- 120, 220, 320, 400, 1200, 1600, 2000 grade emery papers to re-
sirability value assumes zero. In this study, individual desirability move the scratches. Then each of the samples is polished on Velvet
function posses one of the following two characteristics:- cloth. After performing the polishing with Velvet cloth each of the
For goal of maximum, the desirability ( di ) will be defined by samples is ready for etching. The compositions of the etchant are
⎧ 0; if response( y ) ≤ low value( L ) 2.4 g. of CuCl2, 10 ml of 99% C2H5OH and 10 ml of 40% HCl.
i i
⎪ Furthermore, micro-hardness survey was made on flat me-
⎪ ⎛ y − L ⎞ wi
di = ⎨ ⎜ i i
⎟ ; as response( yi ) varies from low( L i) to high( Hi) tallographic specimen across the joints and top to bottom in the
⎪ ⎝ Hi − L i ⎠ weld using Vickers's microhardness testing machine (Make: LECO
⎪
⎩ 1; if response( yi ) ≥ high value( Hi) (3)

For goal of minimum, the desirability will be defined by

⎧ 1; if response( y ) ≤ low value( L )
i i
⎪
⎪ ⎛ H − y ⎞ wi
di = ⎨ ⎜ i i
⎟ ; as response( yi )varies from low( L i) to high( Hi)
⎪ ⎝ Hi − L i ⎠
⎪
⎩ 0; if response( yi ) ≥ high value( Hi) (4)

A weight ( w ) can be assigned to a goal to emphasize the par-

ticular desirability function. Weights can be varied between
0.1 and 10. A weight greater than 1 gives more emphasis to the
goal, while weights less than 1 give less emphasis.
The simultaneous objective function, D, is a geometric mean of
all transformed responses:
1
1 ⎛ n ⎞ ∑ ri
D = ( d1r1 × d1r 2×……………×d1rn )∑ ri = ⎜⎜ ∏ diri⎟⎟
⎝ i=1 ⎠ (5)

where n is the number of responses in the measure. Each response

can be assigned an importance relative to the other responses.
Importance ( ri ) values varies from 1, the least important, to 5, the Fig. 1. Schematic diagram of laser welding process.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 27

maximum tensile strength for 304SS is observed in sample no. R10

and for 316SS is with sample no. S4. The minimum weld width for
304SS and 316SS is R4 and S16 respectively. The metallurgical
characteristics (metallographic, hardness and tensile strength) of
weld with sample nos. R9, R10, S4 and S5 have been presented in
the following section.

4.1. Microstructural analysis of weld metals

The microstructural analysis of four pulsed Nd:YAG laser wel-

ded specimens (two from 304SS weld i.e. sample no. R9 and
sample no. R10 and two from 316SS weld i.e. S4 and S5) have been
carried out using scanning electron microscope (SEM). The selec-
tion of specimens for microstructural analysis has been governed
by the obtained tensile strength responses of 304SS and 316SS
Fig. 2. Robotic control laser welding set-up.
weld (Table 3). Weld metals having maximum and minimum UTS
(as per DoE for both 304SS and 316SS weld) have been considered
Table 2
Process control parameters and their limits.
for the microstructural analysis and it is also assumed that the rest
should have microstructural properties in between the extreme
Parameters with units Notation Levels limits.
Fig. 5a and Fig. 6a show the typical microstructure of 304SS
2 1 0 þ1 þ2
base metal and 316SS base metal. Both mainly consists of fully
Power, W P 407 425 450 475 492 austenitic structure along with few annealing twins. The macro
Scanning speed, mm/s S 4.16 4.5 5 5.5 5.84 view (at nominal magnification) of typical fusion zone of 304SS
Pulse width, ms W 4.40 4.75 5.25 5.75 6.09 and 316SS after pulsed Nd:YAG laser welding have been presented
in Figs. 5b and 6b which clearly point out three distinguishable
regions marked as fusion zone 1 (FZ1), fusion zone 2 (FZ2) and
fusion zone 3 (FZ3) extended from fusion boundary (FB) to the
weld center. It is likely to be mentioned here that all other weld
metals also have these three distinguishable fusion regions. These
distinguishable fusion zones have been generated probably due to
the periodic pulse oscillation (on time and off time) of the pulsed
Nd:YAG laser power source where the base metal initially melts at
the on time or the high energy pulse duration and get partially
solidified during the off time or low energy pulse duration. The
detail micrographs of different fusion zones of 304SS and 316SS
Fig. 3. Close view of welded sample in butt joint configuration. weld metals have been given in Figs. 5c–h and 6c–h respectively.
Micrographs of FZ1 generally show epitaxial columnar dendritic
Co., USA; Model: LM248AT) at 100 gf load with 10 s dwell time. growth at the initial stages of solidification from the fusion
The measurement of tensile strength of welded samples have been boundary towards the weld center. Furthermore, 304SS weld
conducted on Instron (Model-8801) as per ASTM E8 standard with metals show the presence of vermicular and lacy δ ferrite at the
strain rate of 10  4 mm/mm and the weld bead geometry has been dendrite core surrounded by interdendritic γ-phase (Fig. 5c and d)
measured using LEICA DMLM microscope. The schematic view of and the elongated intercellular δ-ferrite enveloped by primary γ-
the tensile test specimen as per ASTM E8 with dimension is shown phase has been observed in 316SS weld metals (Fig. 6c and d). The
in Fig. 4. presence of different δ-ferrite morphologies could be explained by
the primary solidification modes (PSM) of the weld metals which
are based on the Creq/Nieq ratios of the 304SS base metal (¼ 1.68)
and 316SS base metal ( ¼1.33) because no filler metal has been
4. Results and discussion
used in the study. Depending on the Creq/Nieq ratios, the PSM of
the weld metals can be generally categorized in the following two
The measured responses are listed in Table 3. Design expert 10
modes [30,31,40]:
software has been applied for analyzing the measured responses.
The fitted quadratic polynomial models for responses are statisti- AF mode: L-Lþ γ-L þ γ þ δ-γ þ δ-γ:1.25 oCreq/Nieq o1.48
cally significant for the prediction within range of process para-
meters. Therefore they will be used for further analysis. The
FA mode: L-L þ δ-L þ δ þ γ-δ þ γ-γ:1.48 oCreq/Nieq o1.95

The weld metals of 304SS are generally solidified in FA mode

due to high Creq/Nieq ratio. The FA mode of solidification signifies
the precipitation of primary ferrite, plus three phase reaction
(L þ δ þ γ) at the terminal solidification stage, and δ-γ continuing
below the solidus line [41]. The complexity of FA mode of solidi-
fication arises from the fact that, after a certain amount of primary
solidification as ferrite, austenite precipitation occurs through a
peritectic/eutectic reaction (L þ δ-Lþ δ þ γ) and changes the di-
rection of micro-segregations. Moreover, the subsequent solid-
Fig. 4. Laser welded sample for tensile test as per ASTM E8M. state transformation of δ-γ causes additional partitioning,
28 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Table 3
CCD design for actual factors and measured experimental results.

Power (P), Scanning speed Pulse width AISI 304 AISI 316
W (S), mm/s (W), ms
Sample no. Ultimate tensile Weld width Sample no. Ultimate tensile Weld width
strength (UTS), MPa (WW), lm strength (UTS), MPa (WW), lm

425 4.5 4.75 R1 570.16 848.7 S1 523.89 894.2

425 5.5 4.75 R2 558.83 842.5 S2 593.12 879.8
450 5.0 6.09 R3 651.95 899.6 S3 543.22 881.8
450 5.8 5.25 R4 658.82 805.2 S4 600.12 929.4
450 5.0 4.40 R5 467.66 892.2 S5 361.43 894.2
475 5.5 5.75 R6 642.58 892.2 S6 566.97 919.1
450 5.0 5.25 R7 655.34 927.4 S7 568.36 869.5
450 5.0 5.25 R8 652.38 919.1 S8 584.21 877.7
475 4.5 4.75 R9 388.15 925.6 S9 396.5 956.3
425 5.5 5.75 R10 681.5 854.9 S10 582.26 867.3
450 5.0 5.25 R11 568.12 908.7 S11 573.77 904.6
450 5.0 5.25 R12 586.49 912.9 S12 582.91 956.3
475 4.5 5.75 R13 607.89 970.8 S13 530.87 904.6
407 5.0 5.25 R14 612.3 879.3 S14 590.3 840.4
450 4.1 5.25 R15 621.28 896.3 S15 538.62 871.5
475 5.5 4.75 R16 437.11 857.0 S16 407.35 821.8
425 4.5 5.75 R17 631.14 873.5 S17 530.25 865.3
450 5.0 5.25 R18 536.54 919.1 S18 585.87 925.3
492 5.0 5.25 R19 451.78 960.5 S19 478.6 900.5
450 5.0 5.25 R20 595.58 902.3 S20 588.45 968.8

altering the existing concentration profiles [42]. Thus, the final cooling rate, less time is available for segregation of ferrite stabi-
microstructure of FZ1 after complete solidification consists of lizing and austenite stabilizing elements at the solidifying inter-
vermicular or lacy δ-ferrite at the dendrite core enveloped by an face and thus only partial alteration of the epitaxially grown pri-
interdendritic layer of austenite. On the other hand, 316SS weld mary ferrite (due to FA mode) to austenite is possible during
metals are generally solidified in AF mode due to low Creq/Nieq cooling. On the other hand, slower cooling rate allows additional
ratio; where the ferrite forms at the end of the primary austenite time for alloy segregation at the interface and accelerated the δ-γ
solidification process via a eutectic reaction (L þ γ-Lþ γ þ δ) in solid state transformation during cooling. Hence, weld metals
FZ1. Sufficient ferrite-promoting elements (primarily Cr and Mo) samples R10 and S4 should have slightly higher δ-ferrite content
partition to the solidification subgrain boundaries during solidifi- in the final microstructure owing to the faster cooling.
cation to promote the formation of ferrite as a terminal solidifi- Furthermore, the presence of separate fusion regions in the
cation product. The ferrite that forms along the boundary is rela- weld metals of 304SS and 316SS can be explained by the shift in
tively stable and resists transformation to austenite during cooling solidification behavior due to the local undercooling at the solid-
since it is already enriched in ferrite-promoting elements. The δ- liquid interface typically associated with the pulsed-laser welding
ferrite morphology here is not vermicular-like, as in the FA mode [44]. The undercooling is a function of the dendrite tip radius and,
shown in Fig. 5c, because the interfacial energy at the solid-liquid thus, as solidification growth velocity increases (tip radius de-
interface during primary γ-phase solidification is greater than the creases) the local undercooling of the tip also increases. The local
primary δ-ferrite solidification. The dendrite does not develop any undercooling can influence both segregation and solidification
secondary arms and more easily becomes cell like, and thus the mode and at high solidification growth rates (critical growth rate),
elongated intercellular δ-ferrite has been formed (Fig. 6c). It can be sufficient dendrite tip undercooling occurs to promote a shift in
also assumed here that the 304SS weld metals having FA mode solidification mode from primary ferrite to primary austenite [45].
should produce relatively higher amount of δ-ferrite compare to The shift is also dependent on the composition in terms of Creq/
316SS weld metals after complete solidification. Nieq ratio [44]. Weld metals having low Creq/Nieq ratio values (i.e.
The solidification behavior among the weld metals of 304SS 316SS weld metals) may aggravate the shift even in lower critical
(i.e. sample nos. R9 and R10) and 316SS (i.e. sample nos. S4 and S5) growth rate; whereas weld metals with higher Creq/Nieq ratio va-
lues should required higher critical growth rate to initiate the shift.
have been altered due to the variation in continuous cooling as
Despite the degree of growth rate among the weld metals, it can
shown in Fig. 7. Cooling rates of four weld metals have been cal-
be postulated that the pulsed Nd:YAG laser welding is generally a
culated using the following empirical formula [43]:
rapid cooling process and thus all the weld metals should ex-
∂T 2πλρCp(νd)2 perience a certain degree of local undercooling and shift in soli-
(K.s−1) = (T − T0)3
∂t q02 (6) dification which ultimately creates the separate solidification
boundaries in the fusion zone (Figs. 5b and 6b).
where λ ¼thermal conductivity: 16.2 W m K , ρ ¼material
1 1
Again, periodic pulse oscillation (on time and off time) results
density: 8000 kg m  3, Cp ¼ specific heat at constant pressure: in higher turbulence in weld pool which may reduce the tem-
500 J Kg  1 K  1, ν ¼welding speed (m s  1) (Table 3), d ¼plate perature in front of the solidifying interface. The temperature
thickness: 0.0015 m, q0 ¼net power (W) (Table 3), T ¼reference fluctuation inherent in pulsed Nd:YAG laser welding also leads to a
temperature: 1523.2 K and T0 ¼room temperature: 300.15 K. continuous change in the weld pool size and shape, which de-
Calculated cooling rates of samples R9, R10, S4 and S5 are generates dendritic structure growing from the fusion boundary
150.42 K s  1, 280.68 K s  1, 284.01 K s  1 and 208.18 K s  1 re- and favors the growth of new grains at the weld center in FZ2 and
spectively which clearly shows that sample R9 weld metal has the FZ3 (Figs. 5e–h and 6e–h). The new grains have been found to be
longest cooling time (  8.13 s) followed by weld sample S5 regularly oriented toward the direction of higher thermal gradient
(  5.87 s), R10 (  4.36 s) and S4 (  4.3 s). Therefore, at faster (from FZ2 to FZ3) which may yield the shift in preferred growth
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 29

Fig. 5. SEM micrographs of (a) AISI 304SS base metal; (b) three different fusion zones (FZ) in the weld of 304SS; (c) fusion zone 1 (FZ1) of sample no. R9 (d) fusion zone 1
(FZ1) of sample no. R10 (e) fusion zone 2 (FZ2) of sample no. R9 (f) fusion zone 2 (FZ2) of sample no. R10 (g) fusion zone 3 (FZ3) of sample no. R9 and (h) fusion zone 3 (FZ3)
of sample no. R10 after pulsed Nd:YAG laser welding.
30 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Fig. 6. SEM micrographs of (a) AISI 316SS base metal; (b) three different fusion zones (FZ) in the weld of 316SS; (c) fusion zone 1 (FZ1) of sample no. S4 (d) fusion zone 1
(FZ1) of Sample no. S5 (e) fusion zone 2 (FZ2) of sample no. S4 (f) fusion zone 2 (FZ2) of sample no. S5 (g) fusion zone 3 (FZ3) of sample no. S4 and (h) fusion zone 3 (FZ3) of
sample no. S5 after pulsed Nd:YAG laser welding.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 31

Fig. 7. Calculated continuous cooling curve of (a) 304SS weld samples (R9 and R10) and (b) 316SS weld samples (S4 and S5) show the variation in cooling rate.

direction and generates finer grain structure at the FZ3 [46]. Al- results of different weld metals and these are given through Fig. 9a
though FZ2 and FZ3 of both the weld metals generally have fine and b. Tensile test of each welded joint has been carried out at
grain structure, yet they persists significant variation in grain size constant cross-head speed of 0.007 mm/min (strain rate of
among them. Relatively large austenite grains are visualized in the 10  4 mm/mm) up to the fracture and the fracture is generally
FZ3 of sample R9 and sample S5 (Figs. 5g and 6g) and fine localized at the HAZ-BM interface regardless of the base material
equiaxed grains are obtained in the FZ3 of samples R10 and S4 and welding conditions except sample R9 and sample S5. Weld
(Figs. 5h and 6h). Among these weld metals, the significant dif- samples R9 and S5 directly fail from the fusion zone. Yield strength
ference in net energy input may have lead to the grain structure (sys), ultimate tensile strength (sUTS) and uniform elongation (eu)
variation in the fusion zones. From Table 3 it is clear that the of weld metals show different behavior in comparison with base
sample no. R9 and sample no. S5 have higher energy input of metal (Table 4). From Table 4 two major observations are deduced:
105.5 J mm  1 and 90 J mm  1 respectively than the sample no. weld of 304SS (samples R9 and R10) have lower sys and 316SS
R10 and sample no. S4 where the energy input are  77.3 J mm  1 (samples S4 and S5) have higher sys values than their respective
and  77.58 J mm  1 respectively. Higher net energy input extends base metals. Secondly, among the weld samples R10 and S4 have
the cooling time (Fig. 7) and allow the austenite grains to grow up comparable sUTS and eu values with the respective base metals.
to a certain extent. On the other hand, relatively lower energy Although, weld sample R10 has slightly higher sUTS value than the
input reduces the cooling time which restricts grain growth of base metal 304SS. The true stress-true strain relationship (Fig. 9a
austenite and hence generates relatively fine grain structure. and b) also reveals the variation in strain hardening behavior
among the weld metals which take place during uniform plastic
4.2. Micro-hardness analysis deformation (equivalent to true plastic strain). Strain hardening
behavior can be understood from strain-hardening exponent (n) of
Micro-hardness profiles across the fusion zone (horizontal di- different weld metals which is expressed by a power law equation
rection), top to bottom (vertical direction) and the average hard- s¼ Kεn [47], where s is the true stress, ε is the true strain, K is
ness values of 304SS and 316SS weld metals are shown in Fig. 8. strength coefficient and n is the strain-hardening exponent [48].
The hardness of weld sample R9 (  180 HV) and R10 ( 263 HV) Strain-hardening exponent (n) for each weld metals have been
are higher than that of the 304SS base metal (  165 HV) (Fig. 8a– measured from the log-log plots of s-ε and is presented in Table 4.
c). Weld metal sample S4 has slightly higher micro-hardness value Furthermore, to understand the effect of n-value on tensile prop-
(  209 HV) and weld metal sample no. S5 has slightly lower mi- erties, measured n-values of the weld metals are plotted against
cro-hardness value (  160 HV) than the 316SS base metal ( 177 sys and sUTS (Fig. 9c). The results show that sUTS increases with
HV) (Fig. 8a–c). The variation in hardness values among the weld increase in n-value; whereas sys does not vary significantly with
metals can be explained from the relative microstructural differ- n-value and ultimately demonstrates large strain hardening effect
ences. Lower hardness of sample R9 and sample S5 weld metals during uniform plastic deformation. Again, the stress increment
resulted from stable and higher fraction of γ-phase formation (ΔsUTS-YS), expressed as the difference between the values of sUTS
along with lower δ-ferrite content and relatively coarse grain and sys, significantly varies among the weld metals (Table 4). The
structure. The weld samples R10 and S4 having relatively higher δ- correlation between ΔsUTS-YS, sUTS and sys as shown in Fig. 9d
ferrite (due to faster cooling) at the dendrite core and relatively reveals significant stress increment after yield during plastic de-
fine grain structure, results in higher hardness values. Again, formation of the weld metals due to negligible variation in sys.
Fig. 8d and e shows the micro-hardness distribution within the Weld metals having higher n-value show relatively higher ΔsUTS-
different fusion zones (namely FZ1, FZ2 and FZ3) of both the weld YS and eu (Table 4) which indicate large amount of energy ab-
metals. FZ3 hardness is slightly higher followed by FZ2 and FZ1. sorption during plastic deformation up to sUTS and enhancement
Comparatively fine grain structure of FZ3 (due to shift in solidifi- of the strength. The energy absorption during plastic deformation
cation and periodic pulse oscillation) probably is responsible and significant strength enhancement in weld samples R10 and S4
higher hardness. affect the fracture path and shifted the crack propagation towards
the HAZ-BM interface (Fig. 9e) unlike weld samples R9 and S5.
4.3. Tensile behavior Weld samples R9 and S5 show drastic reduction in sUTS and ac-
cording to the Fig. 9a and b they have instantaneous fracture after
Uniaxial tensile test results of base metals and different welded yield through the fusion zone. Weld samples R9 and S5 have
joints are presented in Table 4. Furthermore, true stress vs. true natural notch (reduced fusion zone area) at the fusion zone as
strain curves have been considered to understand the tensile test shown in Fig. 10 where local stress concentration is significantly
32 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Fig. 8. Micro-hardness plots across the fusion zone of (a) 304SS weld and (b) 316SS weld; (c) the average micro-hardness values of weld and the micro-hardness plots taken
from top to bottom of (d) 304SS weld and (e) 316SS weld through different fusion zones.

Table 4
Tensile test results of base metal and welded joints.

Sample Specification rYS (MPa) rUTS (MPa) eu (%) ΔrUTS-YS (MPa) n K (MPa) Location of fracture

BM 304SS 302.06 632.67 49.77 330.61 0.338 253.9 BM

R9 270.87 388.15 5.045 117.28 0.210 294.32 WM
R10 284.41 681.5 44.83 402.67 0.410 218.52 HAZ/BM
BM 316SS 235.56 617.05 31.81 381.49 0.356 257.46 BM
S4 324.51 600.21 34.27 275.7 0.387 236.2 HAZ/BM
S5 274.35 361.43 2.61 87.08 0.267 278.34 WM

sYS ¼yield strength; sUTS ¼ ultimate tensile strength; eu ¼uniform elongation; ΔsUTS-YS ¼stress increment; n ¼strain-hardening exponent; K¼ strength coefficient.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 33

Fig. 9. True stress-true strain curves of (a) 304SS base metal and weld and (b) 316SS base metal and weld; evolution of (c) strain-hardening coefficient vs. strength, (d)
ΔsUTS-YS vs. strength; (e) tensile specimen sample (R10) after fracture shows the fracture path.

high enough and may be the reason of sudden failure unlike other models (p o0.05) at 95% confidence level.
weld samples. It is likely to be mentioned here that the notch at
sample R9 and sample S5 has been created due to the higher 4.4.1. ANOVA output
energy input and, is associated with them resulting in excessive The quadratic model has been found to be the best fit model
melt along with large weld pool during welding which is unable to when compared to linear, 2FI and cubic model on basic of highly
fill the weld crater during solidification. significant p-value and no lack-of-fit. The Fstatistics and p value have
been used in analysis to test the significant factors. The ANOVA
4.4. Analysis of variance tables also show the other adequacy measure R2, adjusted R2,
adequacy precision R2 and predicted R2 for each response. The
The purpose of the ANOVA is to investigate which process entire adequacy measures are close to 1, which are responsible and
parameter/s significantly affects the quality characteristic. The test indicate adequate models. The adequate precision compares the
for significance of the regression model, the test for significance on range of predicted value at the design points to the average pre-
individual model coefficients and the lack-of-fit test have been dicted error [49]. In this study, the values of adequate precision are
performed using Design Expert 10 software. Table 5–8 show the significantly greater than 4. The adequate precision ratio above
analysis of variance of the responses and show the significant 4 indicates adequate model discrimination.
34 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Fig. 10. Macrographs show the fusion zone profiles of samples (a) R9, (b) R10, (c) S4 and (d) S5.

The ANOVA indicates that for the ultimate tensile strength quadratic effect of the laser power ( P 2 ). For WW model of the
(UTS) model of 304SS, the laser power (P), pulse width (W), in- 316SS, laser power (P), along with all interaction and quadratic
teraction effect of laser power and pulse width ( P × W ), the terms are significant process parameters.
quadratic effect of the laser power ( P 2 ) are the significant model The mathematical models for responses, which can be used for
terms. The other model terms are not significant. Secondly for the prediction within same design space, in terms of actual factors are
weld width (WW) model of 304SS, laser power (P), scanning speed shown below:
(S), pulse width (W), interaction effect of laser power and scanning
speed ( P × S ), quadratic effect of the scanning speed ( S2) and pulse
width ( W2) are the significant process parameters. Also the UTS
UTS(for 304SS ) = − 1107. 53 + 20. 7894P−782. 085S−387. 668W
model of the 316SS shows that the main effects are the laser +0. 4462PS+2. 4156PW +23. 71SW
power (P), scanning speed (S), pulse width (W), along with the −0. 0417385P 2+48. 4029S 2−65. 0807W 2 (6)
interaction effect of the laser power and pulse width ( P × W ), the

Table 5
ANOVA for the fitted quadratic polynomial model for UTS of 304SS welded samples.

Source Sum of squares df Mean square F value p-Value Prob4 F

Model 1.170Eþ 005 9 12995.30 9.57 0.0008 Significant

P 29031.10 1 29031.10 21.37 0.0009 Significant
S 2362.28 1 2362.28 1.74 0.2167 Not significant
W 60982.89 1 60982.89 44.90 o 0.0001 Significant
PS 322.83 1 322.83 0.24 0.6364 Not significant
PW 7673.13 1 7673.13 5.65 0.0388 Significant
SW 212.39 1 212.39 0.16 0.7008 Not significant
P2 9930.37 1 9930.37 7.31 0.0222 Significant
S2 2053.56 1 2053.56 1.51 0.2470 Not significant
W2 3892.01 1 3892.01 2.87 0.1214 Not significant
Residual 13583.09 10 1358.31
Lack of Fit 2536.48 5 507.30 0.23 0.9339 Not significant
Pure Error 11046.61 5 2209.32
Cor Total 1.305E þ005 19

R2 ¼ 0.8956, Adjusted R2 ¼ 0.8017.

Predicted R2 ¼0.7149, Adequate Precision¼ 9.720.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 35

Table 6 UTS(for 316SS ) = − 4419. 09 + 9. 52728P +500. 043S

ANOVA for the fitted quadratic polynomial model for WW of 304SS welded
samples. +616. 309W −0. 7429PS+2. 9849PW +4. 015SW
−0. 0255504P 2−14. 7039S 2−180. 231W 2 (8)
Source Sum of df Mean F value p-Value
squares square Prob4F
WW (for 316SS ) = − 7070. 47 + 37. 9346P +1146. 06S
Model 29376.16 9 3264.02 29.02 o 0.0001 Significant
P 9625.27 1 9625.27 85.58 o 0.0001 Significant −1360. 25W −2. 997PS+2. 791PW +178. 75SW
S 7744.27 1 7744.27 68.86 o 0.0001 Significant −0. 040756P 2−72. 8278S 2−77. 0705W 2 (9)
W 1238.34 1 1238.34 11.01 0.0078 Significant
PS 1872.72 1 1872.72 16.65 0.0022 Significant
PW 233.28 1 233.28 2.07 0.1804 Not significant
SW 62.72 1 62.72 0.56 0.4724 Not significant 4.5. Model validation
P2 11.48 1 11.48 0.10 0.7559 Not significant
S2 7996.28 1 7996.28 71.10 o 0.0001 Not significant Fig. 11a–d shows the relationship between the actual and pre-
W2 830.77 1 830.77 7.39 0.0216 Not significant dicted values of responses. These figures indicate that the devel-
Residual 1124.68 10 112.47
Lack of Fit 731.95 5 146.39 1.86 0.2555 Not significant
oped models are adequate and predicted results are in good
Pure Error 392.73 5 78.55 agreement with measured data.
Cor Total 30500.84 19

R2 ¼ 0.9631, Adjusted R2 ¼0.9299. 4.6. Effect of the parameters on responses

Predicted R2 ¼ 0.7992, Adequate Precision¼ 21.310.
The 3-D response surface and contour plots for the effect of
Table 7 process parameters on responses have been constructed according
ANOVA for the fitted quadratic polynomial model for UTS of 316SS welded samples. to the fitted quadratic models (Eqs. (6)–(9)). Figs. 12–15 show the
plots with one variable kept constant at their respective center
Source Sum of df Mean square F value p-Value
squares Prob 4F
value and other two within the working range. It is seen from
Fig. 12a and b that with increase in pulse width, UTS continuously
Model 93920.46 9 10435.61 57.55 o 0.0001 Significant increasing. The tensile properties (UTS) of welded joint affected by
P 19510.56 1 19510.56 107.60 o 0.0001 Significant pulse width can be explained by bead geometry (molten pool).
S 5402.24 1 5402.24 29.79 0.0003 Significant
W 25942.33 1 25942.33 143.07 o 0.0001 Significant
When the pulse width is too low, the welded molten pool has not
PS 689.88 1 689.88 3.80 0.0797 Not significant enough time to form the full penetration, rather incomplete pe-
PW 11137.04 1 11137.04 61.42 o 0.0001 Significant netration occurs. From Fig. 12b and c, it is found that UTS increases
SW 8.06 1 8.06 0.044 0.8373 Not significant
with laser power upto a threshold value and thereafter it start
P2 3675.02 1 3675.02 20.27 0.0011 Significant
S2 194.74 1 194.74 1.07 0.3245 Not significant decreasing. The threshold value of laser power is related to the
W2 29257.66 1 29257.66 161.36 o 0.0001 Significant decomposition temperature (i.e., the temperature of a substance at
Residual 1813.24 10 181.32 which the substance chemically decomposes) of the base metal.
Lack of Fit 1509.01 5 301.80 4.96 0.0518 Not significant
This indicates that increase in the laser power, increases the UTS
Pure Error 304.23 5 60.85
Cor Total 95733.70 19 until the critical temperature of decomposition is reached. It is
evident from Fig. 12a and c that increasing the scanning speed,
R2 ¼ 0.9911, Adjusted R2 ¼ 0.9640. UTS remains almost constant. It indicates that the speed is a non-
Predicted R2 ¼ 0.8720, Adequate Precision ¼23.504.
significant process parameter and it also seen from ANOVA test
(Table 5). Fig. 13a–c shows the weld width increases with laser
Table 8 power and pulse width. This is due to the fact that larger amount
ANOVA for the fitted quadratic polynomial model for WW of 316SS welded
samples.
of the base metal is melted with increase of laser power and pulse
width with increase in power density. With increase in scanning
Source Sum of df Mean square F value p-Value speed weld width gradually decreases (Fig. 13a and c). The line
squares Prob 4F
energy (i.e., the ratio of laser power and scanning speed, defined as
Model 62102.13 9 6900.24 17.75 o 0.0001 Significant the input energy per unit length) decreases as the welding speed
P 7255.24 1 7255.24 18.67 0.0015 Significant increases and less heat is delivered to the weld joint. Fig. 14b and c
S 195.64 1 195.64 0.50 0.4942 Not significant indicate the increase in laser power, UTS continuously decreasing.
W 1337.55 1 1337.55 3.44 0.0933 Not significant
Because lower value of laser power (407 W) is the threshold value,
PS 11227.51 1 11227.51 28.89 0.0003 Significant
PW 9737.10 1 9737.10 25.05 0.0005 Significant which is related to the decomposition temperature of the base
SW 15975.78 1 15975.78 41.10 o 0.0001 Significant metal. It is found from Fig. 14a and b that increase the pulse width,
P2 9350.73 1 9350.73 24.06 0.0006 Significant up to some certain value, increase the UTS. A further increase in
S2 4777.25 1 4777.25 12.29 0.0057 Significant
the pulse width, decrease in the UTS is observed. When the pulse
W2 5350.06 1 5350.06 13.76 0.0040 Significant
Residual 3886.86 10 388.69 width is very low, incomplete penetration appears. But at higher
Lack of Fit 2020.36 5 404.07 1.08 0.4664 Not significant pulse width, an excess of melt material is obtained (underfilling
Pure Error 1866.50 5 373.30 and burnthrough) and consequently less UTS is achieved. Fig. 14a
Cor Total 65988.99 19
shows the increase in scanning speed UTS value remains almost
R2 ¼ 0.9411, Adjusted R2 ¼ 0.8881. constant. Fig. 14c indicates the increase in scanning speed UTS has
Predicted R2 ¼ 0.7267, Adequate Precision ¼16.512. a slightly decreasing trend, because line energy decreases with the
increase of scanning speed. Fig. 15b and (c) shows that the weld
WW (for 304SS ) = − 4355. 46 + 3. 62842P +1504. 19S width increases with laser power. Weld width decreases with the
increase in scanning speed (Fig. 15a and c). Whereas weld width
+199. 533W −1. 224PS+0. 432PW −11. 2SW
increases as the pulse width increases, and then decreases at the
+0. 00142834P 2−94. 222S 2−30. 3703W 2 (7) end because of burnthrough [50].
36 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Fig. 11. Experimental result and predicted plot of (a) UTS of 304SS (b) WW of 304SS (c) UTS of 316SS and (d) WW of 316SS.

Fig. 12. Response surface plots showing the interaction effects of (a) S and W, (b) P and W, (c) P and S on the UTS of 304SS, while the third parameter is at their respective
center value.

Fig. 13. Response surface plots showing the interaction effects of (a) S and W, (b) P and W, (c) P and S on the WW of 304SS, while the third parameter is at their respective
center value.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 37

Fig. 14. Response surface plots showing the interaction effects of (a) S and W, (b) P and W, (c) P and S on the UTS of 316SS, while the third parameter is at their respective
center value.

Fig. 15. Response surface plots showing the interaction effects of (a) S and W, (b) P and W, (c) P and S on the WW of 316SS, while the third parameter is at their respective
center value.

Table 9 Table 11
The selected criteria for numerical optimization for 304SS and 316SS. Optimization validation test results.

Type Parameter or response Limit Goals Importance Optimum condition

Lower Upper Criterion Type P (W) S (mm/s) W (ms) UTS (MPa) WW (lm)

P (W) 407 492 Within range 3 304SS 425 5.83 5.95 Avg. Actual 708.45 791.5
S (mm/s) 4.16 5.84 Within range 3 Predicted 700.97 794.82
W (ms) 4.4 6.09 Within range 3 Error% 1.05 0.41
316SS 407 5.84 4.618 Avg. Actual 605.32 821.56
304SS UTS (MPa) 388.15 681.5 Maximize 4
Predicted 600.12 832.76
WW (mm) 805.2 970.8 Minimize 3
Error% 0.85 1.36
316SS UTS (MPa) 361.43 600.12 Maximize 4
WW (mm) 821.8 968.8 Minimize 3

Table 10
Optimal welding conditions based on the selected criteria for 304SS and 316SS.

Type Sol. No. P (W) S (mm/s) W (ms) UTS (MPa) WW (lm) Desirability

304SS 1 425.260 5.830 5.950 700.975 794.824 1.000 Selected

2 442.149 5.834 5.802 719.883 803.721 1.000
3 420.140 5.836 5.593 683.802 804.869 1.000
4 430.454 5.811 5.987 709.937 798.233 1.000
5 432.060 5.815 5.880 711.069 802.210 1.000

316SS 1 407.000 5.840 4.618 600.120 832.761 0.967 Selected

2 407.000 5.838 4.619 600.120 832.913 0.967
3 407.000 5.839 4.621 600.612 833.004 0.967
4 407.151 5.840 4.620 600.120 833.065 0.966
5 407.000 5.832 4.620 600.120 833.253 0.966
38 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39

Fig. 16. Comparison between cooling curves of experimental values and optimum values of (a) 304SS and (b) 316SS.

5. Numerical optimization maximized response of UTS. Even the cooling curves of the opti-
mum conditions generally have faster cooling than the experi-
The optimization section in statistical software Design Expert mental results, is eventually signifies the validity of the optimi-
10 searches for a combination of factor levels that simultaneously zation process.
satisfy the requirements placed on each of the responses and
factors. This optimization technique searches the design space,
using the developed mathematical model to find the factor set- 6. Conclusions
tings that optimize any combination of one or more goals. The
goals are combined into an overall desirability function (D). This From the foregoing analysis and discussion the following con-
optimization finds a point that maximizes this desirability func- clusion have been drawn:
tion. In the present work common optimization criteria has been
implemented for both 304SS and 316SS. In the criterion, the goals 1. The developed second order equation can predict the values of
are set to maximize the ultimate tensile strength and to minimize the weld quality characteristics with significant accuracy. The
the weld width simultaneously; whereas, the process parameters developed empirical model has been tested by analysis-of-
are kept within the range of studied design space. Table 9 illus- variance (ANOVA) with a confidence level of 95%.
trates the criteria, representing the goals, upper and lower limits, 2. The multi-objective optimized process parameters (laser
as well as the importance of each goal for each response and power, scanning speed and pulse width) for responses have
factor. Table 10 shows the results of best five set of pareto optimal been validated using confirmatory test.
parameter combinations for 304SS and 316SS for selected criteria. 3. The percentage error between the optimization (model) pre-
It is evident from the results that to achieve the maximum UTS diction and the confirmatory result are found to be approxi-
and minimum WW for 304SS and 316SS, the optimum parametric mately  1.5%, which validates the applied optimization
range for the laser power has to be 425–442 W, 407–407.15 W, method.
scanning speeds has to be 5.81–5.83 mm/s, 5.832–5.84 mm/s and 4. The higher line energy is required for 304SS compare to 316SS
pulse width within the range of 5.59–5.95 ms, 4.618–4.62 ms re- to get maximum ultimate tensile strength and minimum weld
spectively. It is observed from Table 10 that scanning speed re- width.
quired for welding is almost same for both 304SS and 316SS, but 5. Pulse width is the most significant factor affecting the ultimate
there is a large difference in laser power. Hence it is concluded tensile strength of both 304SS and 316SS, and it is followed by
that the higher line energy is required for 304SS compare to 316SS laser power and scanning speed.
to get maximum ultimate tensile strength and minimum weld 6. Laser power is most significant factor affecting the weld width
width. The 304SS base material having higher melting point of 304SS, and it is followed by scanning speed and pulse width.
(1400–1450 °C) may require higher line energy to produce the 7. Laser power is the only significant parameter affecting the
melt compared to 316SS (1350–1400 °C) [51]. weld width of 316SS. Scanning speed and pulse width is sta-
tistically insignificant effect on weld width.
5.1. Confirmatory test 8. Microstructure of the weld metals generally influenced by the
net energy input and the composition of the base metals. The
The results of optimization obtained by desirability function weld metals associated with lower energy input produce fine
analysis, have been validated by conducting confirmatory tests. grain structure along with higher δ-ferrite fraction (due to
Three confirmatory experiments for each of the stainless steel faster cooling) compared to higher energy input weld metals
sheets have been conducted. Welding conditions have been cho- irrespective of base metals. Among the base metals, 304SS and
sen randomly from the Table 10. The tested results of experiments 316SS weld solidified in FA mode and AF mode of solidification
at optimum conditions are presented in Table 11. It is obtained respectively. Thus, higher δ-ferrite content is present in the
from Table 11 that there is a small error percentage between op- weld metals of FA mode (i.e., in 304SS).
timum and the experimental values, which validate the applied 9. Average microhardness of the 304SS weld is observed to be
optimization technique. Another confirmation of the optimized higher than the base metal. However, 316SS weld metals show
values for 304SS and 316SS (Table 11) can be visualized form the significant variation where sample no. S4 has higher hardness
comparison between cooling curves of experimental values and and sample S5 has lower hardness values than the base metal.
optimized values as shown in Fig. 16. The cooling rate of optimum Higher hardness of the weld metals are the resultant of grain
parameters are very close to the experimental values of the refinement and δ-ferrite formation in the weld structure.
 N. Kumar et al. / Optics & Laser Technology 88 (2017) 24–39 39

10. Tensile behavior of the weld metals indicate significant varia- [21] C.H. Kim, D.C. Ahn, Coaxial monitoring of keyhole during Yb:YAG laser
tion and the weld metals having fine grain structure along with welding, Opt. Laser Technol. 44 (2012) 1874–1880.
[22] Y. Kawahito, N. Matsumoto, Y. Abe, S. Katayama, Relationship of laser ab-
higher hardness provide better tensile properties. Again, sig- sorption to keyhole behavior in high power fiber laser welding of stainless
nificantly lower strength achieved in the weld of sample no. R9 steel and aluminum alloy, J. Mater. Process. Technol. 211 (2011) 1563–1568.
and sample no. S5 is due to the presence of natural notch [23] Y. Zhao, Y. Zhang, W. Hu, X. Lia, Optimization of laser welding thin-gage gal-
vanized steel via response surface methodology, Opt. Laser Eng. 50 (2012)
created by the higher energy input. 1267–1273.
[24] P. Sathiya, K. Pannerselvam, M.Y.A. Jaleel, Optimization of laser welding pro-
cess parameters for super austenitic stainless steel using artificial neural
networks and genetic algorithm, Mater. Des. 36 (2012) 490–498.
Acknowledgements
[25] M.M. Alam, J. Karlsson, A.F.H. Kaplan, Generalising fatigue stress analysis of
different laser welds geometries, Mater. Des. 32 (2011) 1814–1823.
The authors are grateful to the CoE-TEQIP (Phase-II), Jadavpur [26] H. Du, L. Hu, J. Liu, X. Hu, A study on the metal flow in full penetration laser
beam welding for titanium alloy, Comp. Mater. Sci. 29 (2004) 419–427.
University (Grant no. P-1/RS/137/16) Kolkata, India for providing
[27] A.M. El-Batahgy, Effect of laser welding parameters on fusion zone shape and
the financial assistance for this research work. solidification structure of austenitic stainless steels, Mater. Lett. 32 (1997)
155–163.
[28] M. Mukherjee, T.K. Pal, Role of microstructural constituents on surface crack
formation during hot rolling of standard and low nickel austenitic stainless
References steels, Acta Metall. Sin. (Engl. Lett. ) 26 (2) (2013) 206–216.
[29] J. Yan, M. Gao, X. Zeng, Study on microstructure and mechanical properties of
[1] Q. Wu, J. Gong, G. Chen, L. Xu, Research on laser welding of vehicle body, Opt. 304 stainless steel joints by TIG, laser and laser-TIG hybrid welding, Opt. Laser
Laser Technol. 40 (2008) 420–426. Eng. 48 (2010) 512–517.
[2] M.M.A. Khan, L. Romoli, M. Fiaschi, G. Dini, F. Sarri, Experimental design ap- [30] J.W. Elmer, S.M. Allen, T.W. Eagar, Microstructural development during soli-
proach to the process parameter optimization for laser welding of martensitic dification of stainless steel alloys, Metall. Mater. Trans. A 20A (1989)
stainless steels in a constrained overlap configuration, Opt. Laser Technol. 43 2117–2127.
(2011) 158–172. [31] N.H. Pryds, X. Huang, The effect of cooling rate on the microstructure formed
[3] F. Kong, J. Ma, B. Carlson, R. Kovacevic, Real-time monitoring of laser welding during solidification of ferritic steel, Metall. Mater. Trans. A 31A (2000)
of galvanized high strength steel in lap joint configuration, Opt. Laser Technol. 3155–3166.
44 (2012) 2186–2196. [32] M. Schweier, J.F. Heins, M.W. Haubold, M.F. Zaeh, Spatter formation in laser
[4] Y. Dongxia, L. Xiaoyan, H. Dingyong, N. Zuoren, H. Hui, Optimization of weld welding with beam oscillation, Phys. Proc. 41 (2013) 20–30.
bead geometry in laser welding with filler wire process using Taguchi's ap- [33] G. Chen, M. Zhang, Z. Zhao, Y. Zhang, S. Li, Measurements of laser-induced
proach, Opt. Laser Technol. 44 (2012) 2020–2025. plasma temperature field in deep penetration laser welding, Opt. Laser
[5] Y.C. Liao, M.H. Yu, Effects of laser beam energy and incident angle on the pulse Technol. 45 (2013) 551–567.
laser welding of stainless steel thin sheet, J. Mater. Process. Technol. 190 [34] D. Colombo, B.M. Colosimo, B. Previtali, Comparison of methods for data
(2007) 102–108. analysis in the remote monitoring of remote laser welding, Opt. Laser Eng. 51
[6] C. Kim, J. Kim, H. Lim, Investigation of laser remote welding using disc laser, J. (2013) 34–46.
Mater. Process. Technol. 201 (2008) 521–525. [35] G. Erdos, Z. Kemeny, A. Kovacs, J. Vancza, Planning of remote laser welding
[7] M.M.A. Khan, L. Romoli, M. Fiaschi, G. Dini, F. Sarri, Laser beam welding of processes, Proc. CIRP 7 (2013) 222–227.
dissimilar stainless steels in a fillet joint configuration, J. Mater. Process. [36] W.J. Suder, S. Williams, Power factor model for selection of welding para-
Technol. 212 (2012) 856–867. meters in CW laser welding, Opt. Laser Technol. 56 (2014) 223–229.
[8] Y.S. Yang, S.H. Lee, A study on the joining strength of laser spot welding for [37] I. Eriksson, J. Powell, A.F.H. Kaplan, Melt behavior on the keyhole front during
automotive applications, J. Mater. Process. Technol. 94 (1999) 151–156. high speed laser welding, Opt. Laser Eng. 51 (2013) 735–740.
[9] S. Chen, L. Li, Y. Chen, J. Huang, Joining mechanism of Ti/Al dissimilar alloys [38] A.G. Olabi, F.O. Alsinani, A.A. Alabdulkarim, A. Ruggiero, L. Tricarico, K.
during laser welding-brazing process, J. Alloy. Compd. 509 (2011) 891–898. Y. Benyounis, Optimizing the CO2 laser welding process for dissimilar mate-
[10] K.R. Balasubramanian, G. Buvanashekaran, K. Sankaranarayanasamy, Modeling rials, Opt. Laser Eng. 51 (2013) 832–839.
of laser beam welding of stainless steel sheet butt joint using neural networks, [39] B. Acharjee, A.S. Kuar, S. Mitra, D. Misra, Empirical modeling and multi-re-
CIRP J. Manuf. Sci. Technol. 3 (2010) 80–84. sponse optimization of laser transmission welding of polycarbonate to ABS,
[11] S. Nemecek, T. Muzik, M. Misek, Differences between laser and arc welding of Laser Manuf. Mater. Proces. 2 (2015) 103–123.
HSS steels, Phys. Proc. 39 (2012) 67–74. [40] J.H. Lee, H.C. Kim, C.Y. Jo, S.K. Kim, J.H. Shim, S. Liu, R. Trivedi, Microstructure
[12] J. Um, I.A. Stroud, Total energy estimation model for remote laser welding evolution in directionally solidified Fe–18Cr stainless steels, Mater. Sci. Eng. A
process, Proc. CIRP 7 (2013) 658–663. 413 (2005) 306–311.
[13] C. Chaminade, E. Fogarassy, D. Boisselier, Diode laser soldering using a lead- [41] J.Y. Li, S. Sugiyama, J. Yanagimoto, Microstructural evolution and flow stress of
free filler material for electronic packaging structures, Appl. Surf. Sci. 252 semi-solid type 304 stainless steel, J. Matter Proc. Technol. 11 (2005) 396–406.
(2006) 4406–4410. [42] H.W. Kerr, W. Kurz, Solidification of peritectic alloys, Int. Mater. Rev. 41 (1996)
[14] Y. Zhang, G. Chen, H. Wei, J. Zhang, A novel “sandwich” method for observation 129–164.
of the keyhole in deep penetration laser welding, Opt. Laser Eng. 46 (2008) [43] O. Grong, Metallurgical Modeling of Welding, The Institute of Materials,
133–139. Cambridge, Great Britain, 1994.
[15] T. Sibillano, A. Ancona, D. Rizzi, S.S. Rodil, J.R. Nieto, A.R. Konuk, R. Aarts, A. [44] J.C. Lippold, Solidification behaviour and cracking susceptibility of pulsed-la-
J. Huis, Study on the correlation between plasma electron temperature and ser welds in austenitic stainless steel, Weld. J. 7 (1994) 129s–139s.
penetration depth in laser welding processes, Phys. Proc. 5 (2010) 429–436. [45] W. Kurz, D.J. Fisher, Dendrite growth at the limit of stability: tip radius and
[16] H.A. Abdullah, R.A. Siddiqui, Concurrent laser welding and annealing ex- spacing, Acta Metall. 29 (1981) 11–20.
ploiting robotically manipulated optical fibers, Opt. Lasers Eng. 38 (2002) [46] K. Pal, S.K. Pal, Effect of pulse parameters on weld quality in pulses gas metal
473–484. arc welding: a review, J. Mater. Eng. Perform. 20 (2011) 918–931.
[17] J.R. Berretta, W.D. Rossi, M.D.M.D. Neves, I.A.D. Almeida, N.D.V. Junior, Pulsed [47] P. Ludwik, Elemente der Technologichen Mechanik, Springer-Verlag OHG,
Nd:YAG laser welding of AISI 304 to AISI 420 stainless steels, Opt. Laser Eng. Berlin, 1909.
45 (2007) 960–966. [48] J.W. Simmons, Strain hardening and plastic flow properties of nitrogen-al-
[18] A. Bauchspiess, S.C.A. Alfaro, L.A. Dobrzanski, Predictive sensor guided robotic loyed Fe-17Cr-(8-10) Mn-5Ni austenitic stainless steel, Acta Mater. 45 (1997)
manipulators in automated welding cells, J. Mater. Process. Technol. 109 2467–2475.
(2001) 13–19. [49] E.M. Anawa, A.G. Olabi, Using Taguchi method to optimize welding pool of
[19] S.A. Tsirkas, P. Papanikos, T. Kermanidis, Numerical simulation of the laser dissimilar laser-welded components, Opt. Laser Technol. 40 (2008) 379–388.
welding process in butt-joint specimens, J. Mater. Process. Technol. 134 (2003) [50] V.A. Ventrella, J.R. Berretta, W.D. Rossi, Pulses Nd:YAG laser seam welding of
59–69. AISI 316L stainless steel thin foils, J. Mater. Process. Technol. 210 (2010)
[20] T. Sibillano, D. Rizzi, A. Ancona, S.S. Rodil, J.R. Nieto, H. Chmlickova, 1838–1843.
H. Sebestova, Spectroscopic monitoring of penetration depth in CO2 Nd:YAG [51] C.Y. Ho, T.K. Chu, Electrical Resistivity and Thermal Conductivity of Nine Se-
and fiber laser welding processes, J. Mater. Process. Technol. 212 (2012) lected AISI Stainless Steel, American Iron and Steel Institute, Washington, DC,
910–916. 1977.