Sage Journals: Discover world-class research

Abstract

The prime objective of this work is to find the best combination of parameters in laser welding of stainless-steel 304 H plates. CO₂ laser welding was used to join 5 mm thick stainless-steel 304 H plates. Laser power, welding speed, and focal position were used as input parameters, and experiments were conducted on the basis of the Taguchi L27 matrix. The quality of the weld was analysed by measuring depth of penetration, bead width and hardness. The best parameter combination was found using artificial neural network and genetic algorithm. The Levenberg–Marquardt algorithm predicted output responses more accurately. A confirmatory test was carried out for the optimized parameters identified by the genetic algorithm. The percentage error between the experimental and predicted genetic algorithm values was below 6%. In comparison to the base metal, the optimized weld hardness was increased by 8%. The electron backscatter diffraction analysis of the optimized weld revealed the presence of a higher percentage of high-angle grain boundary than low-angle grain boundary.

Keywords

artificial neural network genetic algorithm scanning electron microscope energy dispersion spectroscopy electron backscatter diffraction

Introduction

For longer days, the scientific community is addressing power generation with improved efficiency that does not affect the environment. Increasing the steam working pressure and temperature¹ is one of the best ways to improve the efficiency of the power plant. Researchers are now working on advanced ultra-super-critical (AUSC) boilers that operate at a pressure of 35 MPa and temperature of 760 °C.² New materials are being developed by the research community to withstand high-temperature operating conditions. Nickel-based super alloys and austenitic stainless steels (ASS) are preferred in the manufacturing of boiler components owing to their high thermal strength, corrosion resistance, and superior creep resistance.³ Among the different ASSs, stainless-steel 304 H (SS 304 H) is one of the widely used materials in the fabrication of boiler components. SS 304 H is used in the superheater and reheater sections of power plants.⁴ Besides boiler applications, SS 304 H steels are extensively used in automotive and aerospace applications. Welding is one of the major processes that is widely used in the manufacture of boiler components. To satisfy the requirements of the AUSC boiler specifications, advanced welding methods are constantly being developed. Arc welding techniques such as tungsten inert gas (TIG), metal inert gas (MIG), and submerged arc welding processes have been extensively investigated in the joining of ASS.⁵ Laser welding of SS 304 H is still in the initial stages of research.

In the arc welding of the SS 304 H, many authors have investigated the influence of various welding parameters on the hardness of the weld. Agarwal et al.⁶ welded SS 304 using a TIG welding process operated in pulsed mode. The fusion zone (FZ) area did not show much change with an increase in the mean current and welding speed. The authors found that the heat-affected zone (HAZ) size was comparatively smaller at a higher welding speed. This was mainly due to the increased mean current. An undercut presence was noticed in the welded sample. The authors reported a trend of increasing undercut size with an increase in welding speed. The authors noticed superior hardness values at higher welding speed and mean current. In another work related to MIG welding of SS 304 H, Rizvi et al.⁷ identified that gas flow rate is the most influential factor on the mechanical properties, followed by arc voltage and wire feed speed. Few authors explored the joining of SS 304 with filler materials and analysed the property variation with different filler materials. Bing-gang Zhang et al.⁸ studied the electron beam welding of SS 304 with copper as the filler material. Hardness was found to be low with an increase in weld copper content. The hardness was different at the upper and lower ends of the weld. At the weld top, the copper inhomogeneity existed due to higher copper content. A similar pattern of decreasing microhardness with increasing copper content was observed by Hao et al.⁹ while laser welding SS 304 – titanium (Ti) alloy with copper as filler material. Increased copper content promoted the formation of a Cu-rich phase in the depleted layer. In many industrial applications, the demand for dissimilar joints has increased significantly. ASS is welded with nickel-based alloys, especially to meet the requirements of AUSC boilers. The dissimilar joining of ASS with nickel-based super alloys will reduce the overall cost involved in the fabrication of boiler components. Nickel (Ni)-based super alloy are preferred in the high-temperature region, while ASS is preferred in the low-temperature region of power plants. The joining of SS 304 H with Ni-based alloys is very difficult owing to their different physical and mechanical properties. Few researchers have investigated the dissimilar joining of SS 304 with Ni-based alloys, and martensitic steels to get the best of both materials. Thakare et al.¹⁰ welded Super 304 L with martensitic P91 using TIG welding process. P91 was used as filler material. The authors were able to achieve defect-free full penetration welds. The average hardness of the weld was around 444 HV. Hardness varied across the different zones of the weld. P91 side weld had 220 HV, whereas the SS304 L side had a hardness of 210 HV. Hardness was higher in the weld centre and lower in the HAZ. Pavan et al.¹¹ joined super 304 H with Inconel 617 using the TIG welding process and noticed that the microhardness was least on the super 304 H side. The presence of Niobium (Nb)-rich precipitate as well as Manganese Sulphide (MnS) inclusions affected the weld hardness.

The selection of optimized parameters is critical for all manufacturing processes. The selection of optimized parameters allows not only obtain better properties but also minimizes the time and costs associated with the process. Advanced optimization techniques such as grey relational analysis (GRA), technique for order of preference by similarity to ideal solution, ViseKriterijuska Optimizacija I Komoromisno Resenje technique, artificial neural network (ANN) coupled with genetic algorithm (GA), particle swarm optimization are used by the research community to identify the optimized parameters of manufacturing processes. In particular, GA and ANN are widely used to optimize process parameters, since they are simple to understand and easy to apply. Srinivasan et al.¹² compared the different learning algorithms of ANN in TIG welding of 15CDV6 aerospace material and found that the Levenberg–Marquardt (LM) is the best among the different learning algorithms. The authors used a single hidden layer and varied the number of neurons in the hidden layer from 5 to 20. Similarly, Kannan et al.¹³ successfully utilized ANN coupled with GA for identifying the optimized parameters in laser welding of NiTinol shape memory alloys. The authors compared different learning algorithms and found that the LM learning algorithm with a hidden layer eight neurons resulted in better predictive accuracy. Poonguzhali et al.¹⁴ compared the prediction capabilities of quick propagation (QP), incremental back propagation (IBP), and batch back-propagation (BBP) learning algorithms of ANN in Spin Arc welding of Al 5083 alloy. The authors noted that the error percentage was minimal in the BBP learning algorithm with two hidden layers and 10 neurons in each hidden layer.

According to the aforementioned literary works, welding plays an important role in the manufacture of boiler components, and the SS 304 H is one of the most commonly used materials for boiler components. TIG and MIG welding are frequently used to join SS 304 H, while advanced radiation-based processes such as laser welding are rarely used. Therefore, an attempt is made in this work to join SS 304 H using the CO₂ laser welding process. ANN with GA has been explored by many researchers in different manufacturing processes to find the right combination of input parameters. Most of the work related to ANN is carried out using a single hidden layer and very few are carried out with more than one hidden layer. Since the number of hidden layers plays a vital role in prediction, in this work an attempt is made to compare the prediction capability of different learning algorithms with one, two hidden layers. The metallurgical characterization of the optimized weld is also discussed with help of microstructures taken using scanning electron microscope (SEM), energy dispersion spectroscopy (EDS), and electron backscatter diffraction (EBSD).

Materials and methods

TRUMPF TLC 1005 CO₂ laser welding setup with a 4-kW power capacity was used to perform the bead-on-plate welding on SS 304 H plates. The laser beam wavelength was around 10.6 μm. 5-mm thick plates were used for this study and the chemical composition of the same is presented in Table 1.

Table 1.

Base metal chemical composition in weight (wt.) %.

Material	C	Cr	Mn	Ni	P	S	Si
SS 304H	0.063	18.87	1.54	8.29	0.032	0.01	0.37

Nitrogen Gas (N₂) was used as a shielding gas to prevent atmospheric contamination during the welding process. Laser power (A), welding speed (B) and focal position (C) were used as input parameters, and 3 levels were selected for each input parameters. The input parameters along with their ranges are shown in Table 2. The schematic of the CO₂ laser welding process and the machine used in this work is shown in Figure 1. Experiments were carried out on the basis of the L27 Taguchi matrix. L27 Taguchi array along with measured output parameters for each experimental run is shown in Table 3. Bead-on-plate welded samples is shown in Figure 2.

Figure 1.

(a) Schematic diagram of laser welding, (b) laser head, and (c) laser welding machine.

Figure 2.

Bead on plate welded samples.

Table 2.

Laser welding input parameters and their ranges.

S. No	Parameter	Unit	Level 1	Level 2	Level 3
1	Power (A)	W	1600	2000	2400
2	Welding speed (B)	mm/min.	600	800	1000
3	Focal position (C)	mm	−3	0	+3

Table 3.

Input and output responses.

Exp. No	Input parameters			Output parameters
Exp. No	A (W)	B (mm/min.)	C (mm)	DOP (mm)	BW (mm)	FZ hardness (HV)	HAZ hardness (HV)
1	1600	600	−3	2.9751	0.8416	217	198
2	1600	600	0	2.9556	0.9883	189.67	190.5
3	1600	600	3	2.7695	1.0569	192	195.335
4	1600	800	−3	2.6034	0.8221	226	203.17
5	1600	800	0	2.76	0.734	202.67	196.33
6	1600	800	3	2.584	0.7633	237	200.17
7	1600	1000	−3	2.5153	0.6752	184	180.005
8	1600	1000	0	2.4756	0.7045	201	193.835
9	1600	1000	3	2.4563	0.7046	182.67	188.83
10	2000	600	−3	4.4237	1.1156	186.67	179.165
11	2000	600	0	4.3355	0.9981	186	181.165
12	2000	600	3	4.0023	1.1451	213.67	193.335
13	2000	800	−3	3.7286	0.8025	185	182.5
14	2000	800	0	3.9244	0.9589	196.3	193.835
15	2000	800	3	3.7679	0.8807	192.67	190.17
16	2000	1000	−3	3.4351	0.7729	191	195.835
17	2000	1000	0	3.3666	0.7632	163	184.67
18	2000	1000	3	3.3374	0.8612	179	194
19	2400	600	−3	5.2358	1.6343	193	181
20	2400	600	0	5.2064	2.1432	178	178
21	2400	600	3	5.4508	1.8202	180.67	178.835
22	2400	800	−3	5.2157	0.8611	190.33	183.35
23	2400	800	0	5.3817	1.742	193.34	185.835
24	2400	800	3	4.6287	1.5267	192	188.835
25	2400	1000	−3	4.7463	1.6539	197.67	188.5
26	2400	1000	0	4.4625	0.7536	208	192.665
27	2400	1000	3	4.2576	0.8024	196.3333	184.17

DOP: depth of penetration; BW: bead width; HAZ: heat-affected zone.

In order to analyse the bead geometry and microstructure, the welded samples were cut using electric discharge machine. The cut samples were mounted using cold mounting powder, and liquid. The mounted samples were polished using SiC emery sheets of various grit sizes, viz. 220, 600, 800, 1200, 1500, and 2000. Alumina powder, along with velvet cloth, was used to get a mirror finish on the mounted samples. To reveal the microstructure, etchant having a chemical composition with 3 parts of distilled water, 2 parts of HCL, and 1 part of HNO₃ were applied to the weld with an etching time of 10–15 s. The OLM vision measuring system was used to take the weld macrographs at a magnification of 5X. Microstructure of the weld was analysed using a thermos-scientific Apreo S Make Hi-Resolution SEM, EDS and EBSD. Vickers hardness machine was used to test the hardness of the sample in various areas of the sample. The load applied was 5 kg with 10 s of dwell time.

Results and discussion

Optimization

ANN and GA were used to identify the optimized parameter combination that would result in lower bead width (BW), higher depth of penetration (DOP), and higher hardness in both the weld zone and HAZ.

Artificial neural networks

Neural power 2.5 was used in this work for carrying out the modelling. ANN consists of two steps, namely training and testing. Around 80% of the data is usually used for training, while the remaining 20% is used for testing. The first 22 welding trial data sets were used for training the neural network, and the remaining 5 welding trial data sets were used for testing. The ANN architecture with different learning algorithms (LM, QP, IBP, BBP) and different numbers of neurons in the hidden layers was executed to identify the best neural network structure. The Tanh function was used for both the hidden and output layers. The Tanh function relation is shown in equation 1.

P (k) = \frac{1 - \exp (- b k)}{1 + \exp (- b k)}

(1)

The number of iterations was set to 1000 for all the learning algorithm. The precision of the neuronal network structure was assessed using the root mean square error (RMSE) value. The RMSE value is calculated using equation 2.

RMSE = \sqrt{\frac{1}{m} \sum_{i = 1}^{n} {(X - Y)}^{2}}

(2)

where m = number of testing data, X = predicted value, Y = experimental value.

The RMSE values for different ANN learning algorithms are presented in Tables 4 and 5. From the above Tables 4 and 5, it is understood that the LM learning algorithm with 2 hidden layers and 10 neurons in each hidden layer predicts the output parameter with less error. The best neural network architecture is shown in Figure 3. The predicted and experimental values of the best learning algorithm are presented in Supplemental Tables 6 and 7. The testing data obtained from the best algorithm are presented in Supplemental Tables 8 and 9. Analysis of variance available in the neural power 2.5 was used to identify the most influential parameter. The most influential factor in the overall objective was found to be laser power, closely followed by welding speed and focal point position. Pie chart showing the significance of input parameters is shown in Figure 4.

Figure 3.

Neural network architecture with two hidden layer and 10 neurons.

Figure 4.

Percentage contribution of Individual factor on the weld quality.

Table 4.

Average RMSE value for one hidden layer.

No of neurons in hidden layer	Average RMSE value
No of neurons in hidden layer	IBP	BBP	QP	LM
5	0.8846	0.82596	0.88921	0.89064
6	0.88902	0.83717	0.88639	0.062109
7	0.88771	0.83494	0.89053	0.56362
8	0.88568	0.83177	0.89064	0.1048
9	0.88521	0.83527	0.89064	0.47581
10	0.8905	0.83798	0.89064	0.75637
15	0.89057	0.83456	0.89039	0.68546
20	0.88979	0.83156	0.89054	0.17799

LM: Levenberg–Marquardt; QP: quick propagation; IBP: incremental back propagation; BBP: batch back propagation; RMSE: root mean square error.

Table 5.

Average RMSE value for two hidden layer.

No of neurons in hidden layer	Average RMSE value
No of neurons in hidden layer	IBP	BBP	QP	LM
5	0.88327	0.8453	0.89062	0.89064
6	0.89043	0.83616	0.89051	0.080596
7	0.89063	0.82707	0.89046	0.21209
8	0.89051	0.82942	0.89064	0.33202
9	0.89064	0.82205	0.89064	0.098363
10	0.89054	0.8287	0.89064	0.021557
15	0.89064	0.83486	0.88945	0.35849
20	0.89064	0.82339	0.87802	0.022866

LM: Levenberg–Marquardt; QP: quick propagation; IBP: incremental back propagation; BBP: batch back propagation; RMSE: root mean square error.

Genetic algorithm

The flow chart depicting the working methodology of the ANN-GA technique is given in Figure 5. GA operates on the basis of the natural selection process. It comes under the wider category of evolutionary algorithms. GA is one of the unique optimization techniques that are straightforward and easy to use. GA is more effective than the Taguchi optimization. The error percentage in GA is very less compared with the traditional methods. GA has the capability to solve static as well as dynamic objective function. Even if the objective function has some noise, GA can solve them. GA has been successfully implemented in many manufacturing processes to find the best parametric combination.¹⁵ GA is generally used to identify the optimized parameters based on bio-inspired operators such as mutation, crossover and selection. GA works with a set of individuals called a population. Each individual is a solution to the given problem. Individuals are characterized by a set of parameters called genes. Each gene is combined into a string to form a chromosome. The fitness function quantifies the optimality of a solution. The fitness function gives a score to each individual. Individuals will be selected for reproduction based on their fitness score. The steps involved in the GA are mentioned in Figure 5.

Figure 5.

ANN-GA working methodology. ANN: artificial neural network; GA: genetic algorithm.

The GA parameters used in this work are as follows, population size: 100, crossover rate: 0.9, and mutation rate: 0.01.^13,15 Selection was done based on roulette wheel method and single point crossover type was used. Tournament selection, Roulette wheel selection, rank selection, steady state selection is some of the methods available to implement GA. Roulette wheel is one of the selection methods which helps in selecting the individuals for the next generation.

Confirmation test

The GA-optimized values and the confirmation test results are presented in Supplemental Table 10.

Metallurgical studies

Macrostructure of weld

To understand the metallurgical aspects of the weld, two welds have been considered. The first one is the weld obtained through GA optimized parameter combination, and the second one is the weld obtained from trail number 20. For convenience, the optimized weld will be represented by the letter O and the weld obtained from trail 20 will be represented by the letter W. Weld W was chosen because it had full penetration and, at the same time, it had the lowest hardness value in comparison with other full penetration welds.

The macrostructure of welds O and W is illustrated in Supplemental Figure 6 (a) and (b). Both welds O and W had keyhole shapes and excessive penetration. Out of the 27 welding trails which were conducted initially, only welding trials from 19 to 23 had full penetration, whereas other welding trails had incomplete penetration. It is understood from the macrostructure of all the welded samples that full penetration was achieved only when the laser power reached a value of 2400 W. Among the 3 input parameters, laser power had significant control on the weld bead geometry. There were three areas in the weld, viz., FZ, HAZ, and unaffected base metal. The width of the HAZ was significantly smaller due to less heat input supplied during the laser welding process.

Microstructures of the welded sample

The base metal microstructure was obtained through an optical microscope and the same is presented in Supplemental Figure 7. The base metal had a full austenitic matrix and the twinning lines were clearly visible in the base metal.

The microstructure of the weld O obtained through SEM is shown in Supplemental Figure 8 (a) to (c). Epitaxial growth was observed in the weld since no filler material was used during the laser welding process. The weld was austenitic and it was dominated by the presence of columnar dendritic grains. The weld centre was dominated by the presence of equiaxed grains. The grain shape was found to be cellular in the HAZ and grain coarsening phenomenon was observed. In HAZ, the grain boundary was free from carbide precipitation. The microstructure of the weld W is shown in Supplemental Figure 9 (a) to (c).

The weld W also had columnar dendrites and no equiaxed grains were observed in the weld centre. Cellular grains were seen in the HAZ of weld W. EDS analysis was done on different zones of weld O, and Supplemental Figures 10 (a) and (b) and 11(a) to (d) show the elemental variations in the weld O. The percentage of various elements within the weld O are presented in Supplemental Table 11. Weld O showed slight variations in the elemental percentage relative to the base metal. The point scan was taken in the dendritic arm of the FZ. The dendritic arm was high in carbon, and the nickel content was nearly reduced by 50%. The chromium percentage in the dendritic arm was reduced by 18.5%, when compared with the overall FZ. Presence of oxygen was noticed only in the dendritic arm. Within the dendritic arm, Ni, Cr, Fe, and Mn varied, indicating the occurrence of micro segregation. Based on the EDS results, the Ni equivalent for weld centre, near the interface was found to be 58.11, and 62.61, respectively. Similarly, Cr Equivalent was 17.42 and 18.64, respectively. Cr and Ni equivalent value were mapped in the Schaeffler diagram¹⁶ and the results showed that weld will be fully austenitic. The EDS analysis of weld W is presented in Supplemental Figures 12 (a) and (b) and 13 (a) to (c).

The percentage of various elements within the weld W are presented in Supplemental Table 12. Weld W had higher carbon content in the weld centre and moving towards the HAZ, a slight reduction in the carbon content was noticed. For weld W, based on the EDS results, the Ni equivalent for weld centre, near the interface was found to be 103 and 91.38, respectively. A greater value of Ni equivalent is primarily due to a higher carbon content. Similarly, Cr Equivalent was found to be 18.48 and 18.75, respectively. The Cr and Ni equivalent values were mapped in the Schaeffler diagram and the results suggested that the weld would be fully austenitic. The higher cooling rate of the laser welding process could have led to an entirely austenitic structure. The average ferritic content of welds O and weld W was analysed with feritscope and it was noted that the ferrite content was 1.625% and 1.42%, respectively. The results obtained from the feritscope and schaeffler graphs were contradictory. The contradictory results might be due to the fact that the schaeffler diagram doesn’t take account of the cooling rate.¹⁷

The EDS analysis of HAZ and the grain boundary of HAZ was carried out and the corresponding elemental percentage is presented in Supplemental Figure 14 (a) to (c). In the HAZ of weld W, the presence of carbides (M₂₃C₆) in the grain boundary was observed. The sensitization phenomenon occurred in the W weld as a result of the formation of carbide-rich precipitates at the grain boundary. Generally, carbides are formed in the following locations: grain boundary, incoherent, coherent twin boundaries, and intragranular. In weld W, carbides formed at the grain boundary. Due to the formation of carbide, the percentage of Cr in the grains was reduced. The presence of M₂₃C₆ moved the grain boundary slightly and revealed the new grain boundary. The carbon content in the HAZ grain boundary was drastically increased and the chromium content was almost equal to 19%.

Macro segregation was observed in both welds O and W. In order to understand the variation of alloying elements, line scan was taken along the FZ and the same is presented in Supplemental Figures 15(a) to (b) and 16(a) and (b). The percentage variation of all the alloying elements in weld W is comparatively greater than in weld O.

Welding parameter effect

The heat input for the optimized weld parameter combination is 144 J/mm. The corresponding input parameters are, laser power: 2400 W, Welding speed: 1000 mm/min, and focal position: −3 mm, respectively. The welding parameters for weld W (worst weld) was laser power: 2400 W, welding speed: 600 mm/min, and focal position: 0 mm, respectively. The heat input supplied for weld W is 240 J/mm. Full penetration was achieved only when the laser power was 2400 W. Both Optimized and worst weld had same laser power but different welding speed and focal position. The higher welding speed in optimized weld helped in minimizing the heat input to the weld. This resulted in higher cooling rate which in turn resulted in the formation of finer grains. Whereas, the welding speed in weld W was lower, which resulted in higher heat input. The focal position comparatively had lesser impact on the weld microstructure as the heat input supplied mainly dependent on laser power and welding speed.

Hardness evaluation

Hardness was measured along the transverse direction of the weld and the average of the five values at five different points for the different areas of welds O and W is shown in Supplemental Figure 17. The base metal had a hardness value of 180 HV. Figure 17(a) and (b), shows that weld O had the highest hardness compared to weld W. In both Weld O and W, the hardness varied in the different regions. All weld region showed higher hardness than the base metal. In both Weld O and W, weld centre had a higher hardness value, which might be attributed to the presence of equiaxed grains. The hardness showed a slight decrement on moving towards the interface from the weld centre, and it was lower in the HAZ. HAZ had grain coarsening effect, which resulted in reduced hardness. A similar trend of hardness variation between different welding zones was observed in Bhargava et al.¹⁸ work.

EBSD analysis

The EBSD analysis of weld O was performed to obtain information on the distribution of grain boundary characteristics of different zones. The results related to the EBSD analysis of the different welds are presented in Supplemental Figures 18 and 19.

Based on the EBSD analysis, it was determined that the mean grain size of the base metal, weld centre, and HAZ areas were 9.284, 11.59, and 25.56 µm, respectively. The refinement of grains and columnar grains presence in the weld centre might have contributed to a slight increase in the hardness value. The larger mean grain size in the HAZ resulted in lower hardness value. Figure 19 shows the grain boundary misorientation details for the different zones of weld O. Misorientation angle within the range 2°–5° was considered as a low angle grain boundary (LAGB). Similarly, the misorientation angle between 5° and 15 °, greater than 15°, was considered as a medium angle grain boundary, and High Angle Grain Boundary (HAGB), respectively. Statistical analysis indicated that the HAGB fraction was higher in the three zones of weld O. The HAGB fraction for the weld, HAZ, and base metal was 0.66, 0.75, and 0.78, respectively. Similarly, the LAGB for the weld, HAZ, and base metal was 0.23. 0.19, and 0.185. The HAGB fraction was higher in the base metal, whereas, LAGB fraction was higher in the weld centre. Increased HAGB fraction resulted in higher hardness. Even though the fraction of HAGB is lower in the weld centre, the fine grains and columnar grains presence would have contributed to the increased hardness.

Conclusions

Bead-on-plate welding of 5-mm thick SS 304 H plates was successfully completed. The following remarks were made:

Of the 27 welding trials, only five welding trails resulted in full penetration, whereas the remaining weld trails had incomplete penetration.

The LM learning algorithm with two hidden layers and 10 neurons predicts output responses more accurately. Laser power was found to be the most influencing factor on the weld quality.

The GA optimized input parameters are as follows, welding power: 2400 (W), welding speed: 1000 (mm/min.) and focal point position: −3 (mm), and the corresponding output parameters are DOP: 5.1 (mm), BW: 1.71 (mm), FZ hardness: 201 (HV) and HAZ hardness: 180 (HV). The maximum error percentage between the GA predicted value and the confirmation test was 6%.

The hardness value was higher than the base metal in the optimized weld (O). Hardness was mainly controlled by the grain size.

The optimized weld microstructure was dominated by the presence of dendrites and in the HAZ, the sensitization phenomenon was not observed. The grain boundary character distribution of weld revealed that the HAGB fraction is higher than the LAGB fraction.

Supplemental Material

sj-docx-1-pie-10.1177_09544089221139108 - Supplemental material for Experimental studies and optimization of process parameters in laser welding of stainless steel 304 H

Supplemental material, sj-docx-1-pie-10.1177_09544089221139108 for Experimental studies and optimization of process parameters in laser welding of stainless steel 304 H by Srinath Selvaperumal and Deepan Bharathi Kannan Thangaraju in Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering

Supplemental Material

sj-docx-2-pie-10.1177_09544089221139108 - Supplemental material for Experimental studies and optimization of process parameters in laser welding of stainless steel 304 H

Supplemental material, sj-docx-2-pie-10.1177_09544089221139108 for Experimental studies and optimization of process parameters in laser welding of stainless steel 304 H by Srinath Selvaperumal and Deepan Bharathi Kannan Thangaraju in Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Deepan Bharathi Kannan Thangaraju

Supplemental material

Supplemental material for this article is available online.

References

Prabha

Sundaramoorthy

Suresh

, et al. Studies on stress corrosion cracking of super 304 H austenitic stainless steel. J Mater Eng Perform 2009; 18: 1294–1299. https://doi.org/10.1007/s11665-008-9347-9.

Kumar

Sehgal

Singh

, et al. Joining of SS304-SS316 through novel microwave hybrid heating technique without filler material. Mater Today Proc 2019; 26: 2502–2505.

Kumar

Balasubramanian

. Hot tensile properties and constant load stress corrosion cracking test data of autogenous weld joints of super 304HCu stainless steel in boiling MgCl2 solution. Data Br 2018; 18: 102–110.

Gokhale

Varma

Singh

, et al. Failure analysis of SS 304 HCu reheater tube of a supercritical power plant. Eng Fail Anal 2022; 137: 106244.

Liu

Jin

, et al. Optimal design for dual laser beam butt welding process parameter using artificial neural networks and genetic algorithm for SUS316L austenitic stainless steel. Opt. Laser Technol 2020; 125: 106027.

Agrawal

Chauhan

Kumar

, et al. GTA Pulsed current welding of thin sheets of SS304 producing superior quality of joint at high welding speed. J Brazilian Soc Mech Sci Eng 2017; 39: 4667–4675.

Rizvi

Tewari

. Optimization of welding parameters by using Taguchi method and study of fracture mode characterization of SS 304 H welded by GMA welding. Jordan J Mech Ind Eng 2018; 12: 17–22.

Zhang

Zhao

, et al. Electron beam welding of 304 stainless steel to QCr0.8 copper alloy with copper filler wire, trans. Nonferrous Met Soc. China (English Ed.) 2014; 24: 4059–4066.

Hao

Dong

Xia

, et al. Microstructure and mechanical properties of laser welded TC4 titanium alloy/304 stainless steel joint with (CoCrFeNi)100-xCux high-entropy alloyinterlayer. J Alloys Compd 2019; 803: 649–657.

10.

Thakare

Pandey

Mahapatra

, et al. An assessment for mechanical and microstructure behavior of dissimilar material welded joint between nuclear grade martensitic P91 and austenitic SS304 L steel. J Manuf Process 2019; 48: 249–259.

11.

Pavan

AHV

Vikrant

KSN

Ravibharath

, et al. Development and evaluation of SUS 304H - IN 617 welds for advanced ultra supercritical boiler applications. Mater Sci Eng A 2015; 642: 32–41.

12.

Srinivasan

Khan

Kannan

TDB

, et al. Application of genetic algorithm optimization technique in TIG welding of 15CDV6 aerospace steel. Silicon 2019; 11: 459–469.

13.

Kannan

TDB

Ramesh

Sathiya

. Application of artificial neural network modelling for optimization of Yb: YAG Laser welding of nitinol. Trans Indian Inst Met 2017; 70: 1763–1771.

14.

Poonguzhali

Deepan Bharathi Kannan

Umar

, et al. Application of ANN modelling and GA optimization for improved creep and corrosion properties of spin-arc welded AA5083-H111 alloy. Russ J Non-Ferrous Met 2020; 61: 188–198.

15.

Sampreet

Mahidhar

Karthic Narayanan

, et al. Optimization of process parameters in laser welding of Hastelloy C-276 using artificial neural network and genetic algorithm. Surf Rev Lett 2021; 28(02): 1–5. https://doi.org/10.1142/S0218625X20500420.

16.

Schaeffler

. Constitution diagram for stainless steel metal weld. Met Prog 1949; 56: 680–680B.

17.

Unnikrishnana

Satish Idurya

KSN

Ismaila

Bhadauriaa

Shekhawatb

Khatirkara

Sapatea

. Effect of heat input on the microstructure, residual stresses and corrosion resistance of 304L austenitic stainless steel weldments. Mater Charact 2014; 93: 10–23.

18.

Bhargava

Paul

Mundra

Premsingh

Mishra

Nagpure

Kumar

Kukreja

. Study on weld bead surface profile and angular distortion in 6 mm thick butt weld joints of SS304 using fiber laser. Opt Lasers Eng 2014; 53: 152–157.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB

29.97 MB