Erudite Journal of Engineering Technology and Management Sciences
Vol. 1, No. 1, September, 2021, pp. 24-29
Journal homepage: http://www.ejetms.com
Prediction and Optimization of Welding Bead Shape Parameters in Stainless Steel Cladding
Sambaturu Gopi1
1
Department of Mechanical Engineering, Aditya Engineering College, Madanapalli, Chittoor, Andhra Pradesh, India
Received: 23 June 2021 ABSTRACT
Accepted: 18 August 2021 In this paper, the process parameters for the optimum weld bead in the flux cored
arc welding (FCAW) of super duplex stainles steel is posed on mild steel (IS-
2062). Conducting the experiment and take the input and out process parameters
Keywords: for the further process. The structure of the welding bead has many consistency
optimisation, bead shape parameters, Flux characteristics, such as the front height, width, the back height, width and the
core arc welding, cladding penetration depth of the welding bead. In order to consider these quality
attributes together in the proper selection parameters, the modificatied artifical
neural networks approach is used to evalute the optimization of each welding
process parameter of weild bead geometry and also to define with optimum
weilding bead geometry.
1. INTRODUCTION relatively thick layer of welding metal to provide a corrosion-
resistant surface. In global industry, the development of clad
Now a day’s the development of corrosion resistant materials is growing as a way of accomplishing an optimal
materials accelerated because of technological developments balance of strength, specific properties of surface and
demanding high performance materials like stainless steels. economy.
Stainless steel has excellent corrosion resistance because of Rao, P.S., Gupta, O.P., Murty, S.S.N. et al [1] addressed the
thin adherent self-healing passive chromium oxide layer. The methods used by Taguchi to design the experiments. Process
stainless steel are comparatively expensive than the mild steel parameters, i.e., wire feed rate, plate thickness, pulse
because of high alloy content. The mild steel possess the frequency, pulse current magnitude, and travel speed are
required strength but lacks the required corrosion resistance. selected to create models through multiple regression analysis.
Stainless steels are sensitive to the microstructure and The models produced were tested for the suitability. The
composition this restricts the processing of the stainless steel results of the confirmatory experiments of the designs can
by hot working processes like rolling, forging and extrusion. predict bead geometry with considerable accuracy.
However, cladding offers a cost effective viable solution. Ganjigatti, J.P., Pratihar, D.C. And RoyChoudhury, A. [2]
Cladding is the method of depositing a proportionately thick attempted to establish by using a regression analysis based on
layer of material over a normally mild steel substrate. The the data acquired in the full-factorial configuration of the
performance of the clad layer depends on the percentage experiments to create the input and output relationship of the
dilution, which in turn depends on the bead shape parameter MIG welding process. The outcomes of these regression
like penetration depth, bead width and bead height usually approaches are evaluated and some concluding observations
determine the productivity of the process. In order to monitor are developed.
the process, a relationship must be formed between the process Xu, Wen-jing and Wu, C.S. De-gang, Zou. [3] concluded
parameters that influence the process and the responses to the on the basis of a hydrostatic model for the liquid metal surface
regulated process. The relationship established by using the is used to examine the onset of bead undercut flaws in high
mathematical model and neural network models. Neural speed welding and the effect of different welding parameters
network is found to be suitable because of its quick learning on the bead undercut tendency, based on the actual welding
and the model can be easily updated. pool geometry and the dimensions calculated from the
numerical model.
Xu, G., Wu, C. [4] Defined on the geometry of the welding
2. LITERATURE REVIEW pool and its dimension in the globular transfer mode during
gas metal arc welding (GMAW) were analyzed numerically
Increasing the efficiency of an welding process while using the thermal conduction system, which considered the
preserving or even bettering the welding quality was the goal effect of the deformation of the welding pool surface on the
of researchers in the area of welding process creation. The heat transfer in a quasi-steady state. The experiments showed
concept weld cladding typically refers to the application of a
24
that the results expected were in good agreement with the 3.2. System Specification
results measured.
Vasudevan, M. & Kuppuswamy, M. & Bhaduri, A. [5] Hardware Requirement
Designed a computational model using a genetic algorithm • Automatic wire feed rate
(GA) to determine the optimum / near-optimum GTAW • Inverter base welding machine
process parameters for achieving the intended weld-bead • Manipulator for table movement
profile through automatic 316LN stainless steel • Shielding gas supply
welding. Utilising experimental data produced on the effect
of process variables on weld-bead geometry, the regression Software Requirement
models compare the weld-bead shape parameters with the • MAT LAB & C Language are used for the artificial
process parameters formed to determine the objective function neural network (ANN)
of the GA. • Optimizations using Microsoft-solver
Dey, V., Pratihar, D.K., Datta, G.L. et al. [6] Bead-on-plate • Artificial neural network (ANN)
welding was explained on Al-1100 Aluminium plates using an
electron beam-welding unit. The constrained optimization
4. FLUX CORE ARC WELDING
problem was resolved by using a GA with a penalty function
approach. The GA was able to identify optimal weld-bead
geometry and to prescribe the appropriate process parameters. FCAW Welding
Benyounis, K. & Olabi, Abdul Ghani. [7] Discussed on the FCAW is known as flux core arc welding and it was semi -
application of DOE, the evolutionary algorithm and the automated or fully automated welding process where a
conceptual network are commonly used to establish a continuous and consumable wire electrode and shielding gas
mathematical relationship between the input parameter of the are fed by a welding gun.
welding parameter and the output variable of the welding joint
in order to implement this process in the field of welding. This
analysis is categorized according to the welding performance Table 1. Composition of stainless steel 316
function. i.e., the geometry of the bead and the mechanical S.No Materials presents % of material
characteristics of the welds. present
P.E.Murray [8] Defined on the relationship in welding 1 Carbon 0.03%
parameter and the process variable developed by the 2 Chromium 20%
regression and dimensional analysis of experimental results. 3 Nickel 10%
These relationships were used to define the range of stable
4 Molybdenum 0.75%
welding parameters and to find the welding parameter by
regulating the length and welding bead geometry to ensure arc 5 Magnesium 2%
stability, appropriate welding bead size and adequate joint 6 Silicon 0.50%
penetration. 7 Phosphorus 0.03%
Kannan, T., Yoganandh, J. [9] Described on the 8 Sulphure 0.03%
experimental study carried out to introduce a mathematical 9 Copper 0.75%
model to predict the geometry of clad beads and their
10 Ferrous Remaining
relationship of shape to the austenitic stainless steel claddings
provided by the process of GMAW. The experiment was
performed on the basis of a four-factor, five-level central
rotatable composite system with complete study model.
3. PROBLEM IDENTIFICATION
Corrosion is a issue that causes the steel framework to
collapse. It cannot be absolutely removed. It can be minimized
to some degree by creating a protective layer of corrosion
resistance over base metal. Generally, this is a concern in
factories like chemical-based industries, fertilizer, food
Figure 1. shows the welding profile and work pieces of weld
processing, petrol, steam and nuclear power plants.
bead
3.1 Object of the present research
Cladding
1. Minimum dilution with desired microstructure in order to The cladding typically involves in the application of a
preserve the alloy content and to have the desired relatively thick layer (approx. 3 mm) of welding metal to
composition resulting the corrosion resistance of the provide corrosion-resistant surface. In manufacturing society,
deposited layer matches that of the weld metal. the use of clad materials is increasingly being made as a way
2. Dilution of the base metal minimized to a very low value of accomplishing an optimal strength, special properties of
without sacrificing the joint integrity. surface and economy. Among all the different welding
3. Weld bead free from defects such as porosity, crack, processes used, the GMAW was a cost-effective option for
discontinues more depth penetration. small and medium cladding areas owing better efficiency, all
positioning capabilities and easy mechanization.
25
5. DESIGN OF EXPERIMENT Xi = necessary coded value of the X parameter.
X = parameter value of Xmin to Xmax,
5.1. Introduction Xmin, Xmax = lower and upper limit of parameter, Design
of Matrix with parameter:
DOE is a mathematical method used to analyse several
variables at the same time. Sir, sir. R. A. Fisher of England
Table 2. Limits of Parameters
brought DOE in the early 1920s. The primary objective was to
find the optimum water, rain, sunlight, fertiliser and soil Factor -2 -1 0 1 2
conditions required to grow the best crop. Using the DOE Level
method, Fisher was able to lay out all the combination of WFR 200 225 250 275 300
factors used in the experimental analysis. Inch/Min
WS 100 120 140 160 180
mm/Min
5.2. Purpose of Experimentation
NTPD 14 16 18 20 22
mm
DOE have many possible applications to enhance processes WGA 90 100 110 120 130
and items, including: Deg
• Compare the alternatives.
15 17.5 20 22.5 25
• Define the significant inputs (factors) influencing the
Lit/Min
output response.
• Elimination of variability.
Pinch and the welding volt are constant parameters.
• Minimization, Maximization or Targeting of Output
Welding volt= 25
(Response).
Pinch= 0
• Enhancement of process or product Robustness.
The parameters used in the design matrix are
• Balance of trade-offs. GFR-Gas Flow Rate
WFR-Wire Feed Rate
5.3. Input Parameters WS-Welding Speed
NTPD-Nozzle to plate Distance
WGA-Weld Gun Angle
The following input parameters chosen for the analysis
HI-Heat Input
• The rate of wire feed
In this design matrix there are
• Welding velocity Experimental Run : 16
• Nozzle to the distance of the plate Centre Point : 6
• The angle of the weld gun Star Point : 10
• Rate of gas flow Total : 32
5.4. Output Responses
Table 3. Design Matrix
The following were the output responses chosen for the
Design matrix for five factor Central composite rotatable
analysis
design, k=5
• Penetration depth
• Bead width S.No GFR W WS NT WGA HI
• Bead height FR PD
1 0 0 0 0 0 1.74
5.5. Limits of Parameters 2 -1 1 -1 1 -1 1.508
3 0 0 0 2 0 1.74
The design matrix chosen for the experiment is five and 4 0 0 0 0 0 1.889
five-level core composite rotatable designs. The appropriate 5 0 0 0 -2 0 2.274
design matrix with a total of 32 experimental tests was 6 0 0 0 0 0 1.74
established for five parameters of welding: WFR,WS NTPD, 7 0 0 2 0 0 1.353
GFR and WGA. The upper (+2) and lower (-2) grades of all 8 1 1 -1 -1 -1 1.508
five factors as shown in table were calculated by the test run 9 0 -2 0 0 0 1.466
prior to actual welding to ensure that the deposited are exempt 10 1 -1 1 -1 -1 1.109
from various visual defects like discontinuity, cracking, 11 0 0 0 0 0 1.74
porosity etc. of the appropriate welding bead. The intermediate 12 1 -1 -1 -1 1 1.479
grades of-1, 0, +1 of all factors were determined by using 13 -1 -1 1 -1 1 1.109
formula. 14 0 0 0 0 0 1.74
15 0 0 0 0 2 1.74
2[2X − (Xmax + Xmin) 16 1 1 -1 1 1 1.508
𝑋1 = 17 2 0 0 0 0 1.74
(Xmax − Xmin)
18 0 0 -2 0 0 2.436
Where, 19 -1 1 1 1 1 1.131
20 -1 1 -1 -1 1 1.508
26
21 -2 0 0 0 0 1.74 The over fit penalty is almost like the tolerance valve that
22 1 -1 1 1 1 1.109 will increase the accuracy of the results. The number is tours
23 0 0 0 0 0 1.74 is 16. Maximum number of operations was taken in the process
24 1 -1 -1 1 -1 1.479 is 50000. And finally, the converge criterion which is nothing
25 -1 -1 -1 1 1 1.479 but the percentage of error and the data give to the converge
26 1 1 1 -1 1 1.131 criterion is 0.00001. in this method there is no cross validation
27 -1 -1 -1 -1 -1 1.479 and the method used for the process is NEWTON GAUSS
28 -1 -1 1 1 -1 1.109 These is about the input data.
29 0 0 0 0 -2 1.74
30 1 1 1 1 -1 1.131 Formula to find the over fit penalty is
31 0 2 0 0 0 1.926
32 -1 1 1 -1 -1 1.131 To find the over fit penalty = {weight x Input parameter}+Bias
6. RESULTS AND DISCUSSION
After the welding process was over then the next step is
prediction. The software used to find the prédiction value SAS
JMP 8 and the module is Artificial neural networks. let see the
procedure for these processes.
Figure 4. Scatter diagram of Actual penetration depth Vs
Penentration depth predicted
Figure 2. Shows the Input and output paramètres in Neural
network
From the above figure the WFR, WS, NTPD, WGA, GFR
are all act as a input process parameter and it is selecting as a
X-factor. The penetration depth, bead width, bead height are
all act as a output parameter and it is selecting as a Y-factor.
While the prediction process is going on that time the one
hidden layer has ten nodes (H1-H10).
Figure 5. Bead width Vs Bead width predicted
INPUT DATA
There are 5 main data given for this process they are 10
hidden nodes, over fit penalty, number of tours, maximum
iterations, converge criterion. While the process is taking place
on that time the hidden layer contains 10 hidden nodes, the 10
hidden nodes was helping the process to match the actual valve
to the predicated valve.
Figure 6. Bead height Vs Bead height predicted
The scatter diagram was drawn between the actually valve
Figure 3. Shows Input Data for prédiction and the predicted valve, and the scatter diagram was drawn for
the output process parameter are penetration depth, bead width,
bead height. In the scatter diagram there was a line pass, the
27
line angle is 450 and the points are valve. In the scatter
diagram both the valve actually valve and the predicted valve
are lies 900 (ie., perpendicular to each other). And center of
the diagram some points are away from the center line that is
the (zero) 0 valve in the design matrix.
Figure 10. Shows the fit results and fit history
Figure 7. Penetration depth Residual 7. CONCLUSION AND FUTURE SCOPE
The investigation with the aid of the neural network has
made it possible to predict the penetration of the welding bead.
Based on this research, it can be inferred with the built model
which can be used to accurately predict not only the
penetration of the welding bead, but also the width of the bead
and the height of the bead within the defined limits of the
process parameters. It has the required data for the
construction of Artificial neural network and validates model
will be used for the prediction of the responses mentioned
above. The prediction has done well but the optimized result
using the optimization software is to get much better in result
accuracy considered as the future work.
REFERENCES
Figure 8. Bead width Residual
[1] Rao, P.S., Gupta, O.P., Murty, S.S.N. et al. Effect of
process parameters and mathematical model for the
prediction of bead geometry in pulsed GMA welding. Int
J Adv Manuf Technol 45, 496 (2009).
https://doi.org/10.1007/s00170-009-1991-1
[2] Ganjigatti, J.P., Pratihar, D.K. & RoyChoudhury, A.
Modeling of the MIG welding process using statistical
approaches. Int J Adv Manuf Technol 35, 1166–1190
(2008). https://doi.org/10.1007/s00170-006-0798-6.
[3] Xu, Wen-jing & Wu, C.S. & Zou, De-gang. (2008).
Predicting of bead undercut defects in high-speed gas
metal arc welding (GMAW). Frontiers of Materials
Science in China. 2. 402-408. 10.1007/s11706-008-
Figure 9. Bead height Residual 0065-x.
[4] Xu, G., Wu, C. Numerical analysis of weld pool
geometry in globular-transfer gas metal arc
The above residual diagram was drawn to find the error, welding. Front. Mater. Sci. China 1, 24–29 (2007)
here also the parameter let us consider is penetration depth, https://doi.org/10.1007/s11706-007-0005-1
bead width, bead height. The center line in the diagram is nil [5] Vasudevan, M. & Kuppuswamy, M. & Bhaduri, A..
error (ie., zero error) based on the line the error will be decide (2010). Optimising process parameters for gas tungsten
to the points. And the center of the diagram some points are arc welding of an austenitic stainless steel using genetic
away from the center line that is the (zero) 0 valve in the design algorithm. Transactions of the Indian Institute of Metals.
matrix. 63. 1-10. 10.1007/s12666-010-0001-5.
These are obtained valve which come through the prediction [6] Dey, V., Pratihar, D.K., Datta, G.L. et al. Optimization
software for the penetration depth, bead width, bead height. and prediction of weldment profile in bead-on-plate
welding of Al-1100 plates using electron beam. Int J Adv
Manuf Technol 48, 513–528 (2010).
https://doi.org/10.1007/s00170-009-2307-1
[7] Benyounis, K. & Olabi, Abdul Ghani. (2008).
Optimization of Different Welding Process Using
28
Statistical and Numerical Approaches – A Reference
Guide. Advances in Engineering Software. 39. 483-496.
10.1016/j.advengsoft.2007.03.012.
[8] Murray, P.E.. (2002). Selecting parameters for GMAW
using dimensional analysis. Welding Journal (Miami,
Fla). 81. 125/S-131/S.
[9] Kannan, T., Yoganandh, J. Effect of process parameters
on clad bead geometry and its shape relationships of
stainless steel claddings deposited by GMAW. Int J Adv
Manuf Technol 47, 1083–1095 (2010).
https://doi.org/10.1007/s00170-009-2226-1
CC-BY
This is an Open Access article that uses a funding
model which does not charge readers or their
institutions for access and distributed under the
terms of the Creative Commons Attribution
License (http://creativecommons.org/licenses/by/4.0)
and the Budapest Open Access Initiative
(http://www.budapestopenaccessinitiative.org/read)
which permit unrestricted use, distribution, and
reproduction in any medium, provided original
work is properly credited.
29
