Keywords- Full Factorial Technique, Mathematical Modelling, Submerged Arc Welding, Welding Process Parameters, Weld Bead Width.

Transcription

1 Prediction of the Effect of Submerged Arc Welding Process Parameters on Weld Bead Width for MS 1018 Steel Shahnwaz Alam, Mohd.Ibrahim Khan Abstract- The present investigation uses arc voltages, current, welding speed, wires feed rate and nozzle-to-plate distance as process parameters. Two level full factorial a technique has been used for experiment design. Multiple regression analysis has been used to develop a mathematical model to predict weld width using a single wire Submerged arc welding depositing bead-on-plate welds on 12mm plate. Adequacy and significance of the model has been checked by using analysis of variance, F-test and t-test respectively. Main and interaction effects of process variables on weld-width have been graphically presented. Experiments have been performed to compare the results obtained with the corresponding predicted values. The models developed have been found to be adequate with a confidence level 95%.Weld width has been found to increase with increase in voltage, current and wire feed rate and decreases with increase in welding speed and nozzle-to-plate distance. Cross validation test fulfills the validity of the model developed. Keywords- Full Factorial Technique, Mathematical Modelling, Submerged Arc Welding, Welding Process Parameters, Weld Bead Width. I. INTRODUCTION Submerged arc welding is widely employed as one of the major fabrication processes in industries to-day due to its inherent advantages of deep penetration, smooth bead, superior joint quality, good welding speed and excellent weld appearance (without spatter) and high utilisation of electrode feed wire. The important process parameters include: welding current arc voltage, welding speed nozzle to plate distance and wire feed rate. The present work investigates the effect of these parameters on weld bead width. SAW is one of the oldest automatic welding processes developed during 1930s [1],[2],[3] and contributes to approximately 10% of the total welding needs over the world and is commercially used for welding of low carbon steels, high strength low alloy steels, nickel base alloys and stainless steels [4]. Apart from joining, this process can also be used for cladding applications to increase corrosion and wear resistance on the surfaces. Welds produced are sound, uniform, ductile, and have good impact resistance [5]. Researchers are often faced with the problems of relating the process variables and bead geometry to the weld bead quality because there are some unknown, nonlinear process parameters [6]. Many efforts have been made to develop a mathematical model to study these relationships and the best way to solve this problem is by using experimental models. A multiple regression technique has been utilized to establish the models for various welding processes [7], [8]. Theoretical predictions of the effect of current, electrode polarity, electrode diameter and electrode extension on the melting rate, bead height, bead width and weld penetration have been investigated [9].Multiple regression model for predicting volume of submerged arc deposit has also been proposed per unit time[10]. Fractional factorial technique has been proposed to be used to predict geometric dimensions of the weld bead in SAW [11].Mathematical model to predict weld bead geometry for the flux cored welding process has been proposed [12]. Fig. Experimental Set- up Fig.2 Weld bead Width (W) 97

2 II. EXPERIMENTAL WORK any systematic error. In 32 trials, beads were laid on 32 A. Identification of important parameters- Review of plates. Two specimens of 16mm width were cut literature and a large number of trial runs indicate that transverse to the welding direction from each welded the dominant factors which are having greater influence on the responses are open circuit voltage plates. These specimens were cleaned, ground, polished (OCV), welding current (I), wire feed rate (F), welding and etched with 10% nital (90% alcohol + 10% of nitric speed (S) and nozzle- to- plate distance (C). acid). All specimens of the first set were macro etched to Determining the working limits of the parameters reveal the bead profile and some specimens of second set Extensive trial runs have been carried out to find out the feasible working limits of submerged arc welding were micro etched to reveal the microstructures. Weld parameters on mild steel. Different combinations of open bead profiles were traced by using an optical profile circuit voltage (OCV), welding current (I) wire feed rate projector/image at 20X magnification and the magnified (F), welding speed (S) and nozzle- to- plate view was drawn on the transparencies for further distance (C) were used to observe their effect on the desired response. The weld deposits were visually analysis.measurments were made for bead width (W) inspected to identify the working limits of the welding The observed values of the responses are given in parameters. Table.3. B. Developing the experimental design matrix- The D. Material used-for making specimen of size 295x145 feasible limits of the parameters were selected in step2 such a way that the welds obtained were free from mm 2 with a steel plate,12 mm thick was used with the surface defects. A two level full factorial design of (2 5 = composition of (C-0.102%,S-0.179%,Mn-0.466%,S- 32) thirty two experimental runs, which is a standard %,Cr-0.036%,Ni %). statistical tool to investigate the effects of five E Flux used- The study was carried out by using the independent direct welding parameters. This technique available agglomerated flux, automelt GR-4 with a reduces the experimentation costs and provides the required information about the main and interaction composition of (C-0.08%, Mn.00%, Sn-0.25%). effects. The commonly employed method of varying one Table Important Process Control Variables with Notations parameter at a time, though popular, does not give any and Range. information about interaction among parameters. S.no Paramete Units Notati High Low (- Table presents the working range of factors considered. rs ons () 1) For the convenience of the recording and processing the 1 Voltage volts V experimental data the upper and lower levels of the 2 Current amp I factors are coded as and respectively. C. Conducting the experiments as per designed matrix- 3 Wire mm/mi F The experiments were conducted at the welding research feed rate n division of Research design and standard 4 Welding mm/mi S speed n organization (RDSO) Lucknow India. The experiments have been performed on a submerged arc welding machine of constant potential transformer- rectifier type with a maximum welding current capacity of Amp. Experimental set-up used is shown in Fig.1. A 3.2mm copper coated mild steel feed wire electrode was used with composition of, (C-0.10%, Mn-0.45%, Si- 0.02%). Bead on- plate weld were deposited according to the conditions directed by the design matrix using 2- level 2 5 full factorial technique as given in Table.2. The signs under the columns 1, 2, 3 and 4 were arranged in standard full factorial order, while the other columns were obtained by following a standard procedure for designing such a matrix. The plates were cleaned mechanically and chemically so as to remove oxide layer or any other sources of impurities, before welding. The experiments were performed in a random order to avoid 5 Nozzle to plate distance mm C F.Selection of the Useful Limits of the Welding Parameters: - Trial runs were carried out by varying one of the process parameters whilst keeping the rest of them at constant values. The working range was decided upon by inspecting the bead for smooth appearance and the absences of any visible defects. The upper limit of a factor was coded as and lower limit as. The selected process parameters with their limits, units and notation are given in Table.1 above. G. Developing the design matrix:- The designed matrix is given below the level (or signs) for the parameter C (nozzle to plate distance) are derived by the relation 98

3 Wel d. no V volt s I amp Table-2 Design Matrix F mm/mi n S mm/m C* mm 1 C 2 S 3 C 4 S 5 F Treatme nt combinat ion+ve 6 FSC 7 F 8 FSC 9 I 10 ISC 11 I 12 ISC 13 IFC 14 IFS 15 IFC 16 VIFSC 17 V 18 VSC 19 V 20 VSC 21 VFC 22 VFS 23 VFC 24 VFS 25 VIC 26 VIS 27 VIC 28 VIS 29 VIF 30 VIFSC 31 VIF 32 VIFSC The numbers in the first column indicate the number of trial run Table- 3 The Observed Values of The Responses. Weld. V I F S C Bead no vol Am mm/mi mm/mi m widt t p n n m h (W) mm III. DEVELOPMENT OF MATHEMATICAL MODELS Mathematical models can now be developed for the SAW process to predict weld bead width and to establish the interrelationship between welding process parameters to weld bead width. The experimental data were used to develop linear models, and analysis of the models was 99

4 carried out through ANOVA Minitab, SPSS software 25 was used for this purpose. The general response 26 function representing any of the weld-bead dimensions can 27 be Expressed as Where, Y= Weld bead response, V= Open Arc voltage, 30 I= Arc Current, F=Wire feed rate, S= Welding speed, 31 and C= Nozzle-to-plate distance. 32 The effect caused by change in five main factors and their first order interaction can be expressed as the relationship Trial IF IS IC FS FC SC selected being a first order linear equation no Where Y represents any of the weld bead dimensions, b0 3 is constant and b1,b2,b3,b4,b5,b11, 4 b12,b13,b14,b15,b23,b24,b25,b34,b35,b45 are coefficients 5 of the model. Signs for calculating effects of parameters and their interactions. The signs for the 6 7 interaction were obtained by multiplying the 8 corresponding signs of the involved factors (Table 4) Table-4. The signs for main and interaction effects 9 10 X0 V I F S C VI VF VS VC IF IS IC FS 11 FC SC

5 Table-5. Coefficients of the Models Serial no Coefficients b j Width (W) 1 b b b b b b b b b b b b b b error total Model name SEE R 100R² (adj) weld width From the above tests it can be concluded that the model is adequate. D. T- Test for Significance of Regression Coefficients- The determination of co-efficient (R 2 ) indicates the goodness of the fit for the model. In the case of reinforcement height, the value of the determination coefficient (R2 = 0.998) indicates the high level of significance. Scatter diagram between actual and predicted values of reinforcement height has been presented in Fig.3. It can be observed that there is a good correlation between the actual and predicted values which further supports the validity of developed model. A. Predicted response: The final mathematical model Bead width was expressed as a linear function of the input process Parameters (in coded form) are as follows 15 b b A. Evaluation of the co-efficient of the model- The values of the coefficients of the linear equation were calculated by the regression method. The Minitab and SPSS software package were used to calculate the values of these coefficients for different responses. All the coefficients were tested for their significance at 95% confidence level applying student s t-test. B. Response:-Bead width was expressed as a linear function of the input process Parameters (in coded form) are as follows C. Checking adequacy of the model- The analysis of variance (ANOVA) technique has been used to check the adequacy of the developed model. Analysis of variance tests for models, t-test, p- value, standard error of estimate, coefficient of multiple correlations, has been estimated and the results are given in Table 6. Table.6 Result Table Source DF SS MS F P regression Residual Table.7 Experimental, Predicted value and Error s.no WW WW E

6 would expect 31cases or 97% of cases to have standardized residual within about -2 to 2 and 1 case or 3% that are outside of the limits). Therefore our sample appears to conform to what we would expect for a fairly accurate model. Table.8.Results Case Number Standard Residual Penetration Predicted Value Residual G. Cross validation of models- For the purpose Sten s formula is used which as shown below: Adjusted R 2 =1-[(N/N-k) (N-2/N-k-2) (N/N)] (1- R 2 ) Adjusted R 2 =1-[(32/32-5) (32-2/32-5-2) (32/32)] ( ) =0.998 This above value is very similar to the observed value of R 2 (0.999) indicating that cross- validity of this models is very good. Fig.3 Plot of Actual Vs. Predicted Response of Weld Width E.Normally Distributed Errors: - A residual is the difference between the observed and model predicted value of the dependent variable. Fig 4 below shows the histogram plot of the residual to check the assumption of normality of error terms. The shape of the histogram approximately follows the shape of the normal curve. Thus histogram is acceptably close to normal curve, this indicates that normality assumption is not violated. Fig.4 Normality distributed errors F. Case Wise Diagnostics- SPSS produces a summary table of the residual statistics and these should be examined for extreme cases. Table 7 shows the cases that have a standardized residual less than -2 or greater than 2 in case wise diagnostic statistics in ordinary sample we IV. RESULTS AND DISCUSSION: The predicted influences of the welding parameters on the weld bead width within the range of parameters used are shown in Fig.4.and the interaction effects between variables are shown in Fig.5. A. Direct effect or main effect of process parameters on weld width: Fig. 4 shows the effect of input parameters on bead width in submerged arc welding process.. The bead width increases with voltage, current and wire feed rate and decreases with welding speed and nozzle to plate distance. B.Interactive effect of open circuit voltage, current, wire feed rate, welding speed and nozzle to plate distance: Fig. 5 shows the combined effect of open circuit voltage, current, wire feed rate, welding speed and nozzle to plate distance on the weld width. Depicts the effect of voltage on weld width. It is evident that at high voltage value as the current increases the weld width also increases but at low voltage value as the current increases the weld width decreases at a constant rate. It is clear that at high voltage value the weld width increases with increase in wire feed rate the weld width decreases at low voltage value with increases in wire feed rate. The bead width decreases with increase in welding speed at high voltage value and bead width is less at low voltage value.this is due to the fact that at high welding speed, the heat input per unit length decreases and therefore the bead width also decreases At high voltage value the bead width decreases with increase in nozzle to plate distance, however at low voltage value bead width also decreases at very fast rate with increase in nozzle 102

7 to plate distance. This is due to longer the nozzle to plate distance, At high current value the weld width increases with increase in wire feed rate, however at low current values weld width increase with increase in wire feed rate. This is due to the fact that at higher wire feed rate, the current is very high, so the effect of resistance heating will be very small as compared to arc heating. Therefore, the weld width increases. That at the high current value, the weld width decrease rapidly with increases in welding speed, however at high current value the weld width also decreases with increase in welding speed. At the high current value, the weld width decreases very rapidly with increases in nozzle to plate distance. However at low current value the weld width decreases at a lower rate with increase in nozzle to plate distance. Fig.5 Interaction Effect Plot For Weld Width Fig.4 Main effect plot for wild width V. CONCLUSION 1. The two level full factorial designs are found to be an effective tool for quantifying the main and interaction effect of variables on weld width. 2. The developed model can be effectively used to predict the weld width in the submerged arc welding within the range of parameters used. 3. Proposed models are adequate to predict the weld width with a confidence level of 95%. 4. Weld width rapidly increases with voltage, slowly increases with current and wire feed rate and decreases with welding speed and nozzle to plate distance. 5. The F-test indicates that the regression model as a whole is significant. 6. Cross- validation test full-fills the validity of the models developed REFERENCES [1] B.Howard Carry, Modern Welding Technology, (Prentice Hall Inc, Englewood Cliffs, NJ), [2] S.V Nadkarni, Modern Welding Technology, Oxford &IBH Publishing Co. Pvt Ltd, New Delhi, [3] P. T Houldcroft. Submerged Arc Welding. Abington, U.K., [4] E.M.Wilson,. SAW of 1% Titanium 18% Nickle-Co-Mo Mar aging Steel, British Welding Journal, 13, 2, pp 67-74, 1966, 103

8 [5] M.I. Khan, Manufacturing Science PHI, New Delhi, [6] D. Kim, M. Kang, S. Rhee, Determination of Optimal Welding Conditions with a Controlled Random Search Procedure. Welding Journal , [7] J.Raveendra, R.S.Parmar, Mathematical Models to Predict Weld Bead Geometry for Flux Cored Arc Welding. Metal Construction 19/ , [8] L.J. Yang, R.S. Chandel, and M.J. Bibby, The Effects Of Process Variables On The Weld Deposit Area Of Submerged Arc Welds. Welding Journal 72/1 118, [9] R.S. Chandel, H.P. Seow, and F.L. Cheong, Effect of Increasing Deposition Rate on the Bead Geometry Of Submerged Arc Welds, Journal of Materials Processing Technology , [10] S. Datta, M. Sundar, A. Bandyopadhyay, P.K. Pal, S.C. Roy, and G. Nandi, Statistical Modeling For Predicting Bead Volume Of Submerged Arc Butt Welds. Australasian Welding Journal 51/ , [11] V.K.Gupta and R.SParmar, Fractional Factorial Technique to Predict Dimensions of the Weld Bead In Automatic Submerged Arc Welding. J Inst Engr (India), 70, 67-71, [12] J.Ravindra & R. S Pramar, Mathematical Model to Predict Weld Bead Geometry for the Flux Cored Welding Process. Metal Construct, R-35R., [13] S.Alam and M.I.Khan, Prediction of Weld Bead Reinforcement Height for Steel using Submerged Arc Welding Process Parameters, International Journal of Applied Engineering Research ISSN Volume6, Number 15 pp ,Research India Publications, [14] S.Alam,K.Abbasi and M.I.Khan, An experimental study on the effect of increased pressure on MIG welding arc m published in International Journal Of Applied Engineering Research, (IJAER), Dindigul, Volume No.2, No1, ISSN: Pp 22-27, [15] S.Alam, M.I.Khan, Prediction of Weld Bead Penetration for Steel Using Submerged Arc Welding Process Parameters, International Journal of Engineering Science and Technology (IJEST), ISSN: , Vol. 3 No.1 Pp , October ACKNOWLEDGEMENT I wish to expresses my deep sense of gratitude and indebtedness to my supervisor Prof. (Dr.) Mohd. Ibrahim Khan for his valuable guidance, patient reviews, and painstaking efforts in providing valuable suggestion in giving final shape to the text are gratefully acknowledged. AUTHOR BIOGRAPHY SHAHNWAZ ALAM, M.Tech and persuing Ph.D.He is Associate Professor (Jr.), Department of Mechanical Engineering at Integral University, Lucknow.He has published numerous research papers in national and international Journals. Mohd..Ibrahim Khan, Ph.D, is Professor, Department of Mechanical Engineering at Integral University,Lucknow. Earlier,he has been Reader both at MNNIT Allahabad and University of Basrah,Iraq and Professor at University of Garyounis,Benghazi,Libya. He has more than four decades of teaching and research Experience. Dr.Khan authoured four books, and also has published numerous research papers in reputed journals. He is a Fellow of the Institution of Engineers (India) and a Senior Member of the Society of Manufacturing Engineers (USA).He received the title of Man of the Year for 2009 for India from the American Bibliographical Institute, USA.He has also received the Bharat Vidya Ratan Award, Bharat Jyoti Award, Eminent Educationist Award and Jewel of India Award from India. 104