Optimisation of welding parameters for weld penetration in FCAW

of Achievements in Materials and Manufacturing Engineering VOLUME 16 ISSUE 1-2 May-June 2006 Optimisation of welding parameters for weld penetration in FCAW N.B. Mostafa* , M.N. Khajavi Mechanical Engineering Department, Shahid Rajaee University, Tehran, Iran * Corresponding author: E-mail address: [email protected] Received 15.11.2005; accepted in revised form 15.04.2006 Analysis and modelling ABSTRACT Purpose: This paper describes prediction of weld penetration as influenced by FCAW process parameters of welding current , arc voltage , nozzle-to-plate distance, electrode-to - work angle and welding speed . Optimization of these parameters to maximize weld penetration is also investigated Design/methodology/approach: It deals with the statistical technique of central composite rotatable design to develop a mathematical model for predicting weld penetration as a function of welding process parameters. The constrained optimization method is then applied to this model to optimize process parameters for maximizing weld penetration Findings: The result obtained from the developed model indicates that the model predicts the weld penetration adequately. The optimization result also shows that weld penetration attains its maximum value when welding current, arc voltage , nozzle-to-plate distance and electrode-to-work angle are maximum and welding speed is minimum Research limitations/implications: The statistical technique of developing a model for prediction of weld penetration is valid only within the specified limits of welding process parameters and hence maximization of penetration is also valid within these limits. This technique can be modified to include other parameters such as plate thickness affecting penetration. Practical implications: The statistical method for modeling and the optimization method used have found increasing applications in a variety of fields for investigation because through these methods research becomes economical , fast and versatile. Originality/value: The methods described in this paper for weld penetration prediction and optimization can eliminate the need for performing experiments on the basis of the conventional trial and error method which is time consuming and economically not justifiable. Keywords: Manufacturing and processing; Welding; Weld penetration modelling; Optimisation; Sequential quadratic programming 1. Introduction 1. Introduction Inadequate weld bead dimensions such as shallow depth of penetration may contribute to failure of a welded structure since penetration determines the stress carrying capacity of a welded joint [1].To avoid such occurrences the input or welding process variables which influence the weld bead penetration must therefore be properly selected and optimized to obtain an acceptable weld bead penetration and hence a high quality joint [2]. To predict the effect of welding process variables on weld 132 Research paper bead geometry and hence quality researchers have employed different techniques [ 2-17]. This investigation is based on a previous work [3] carried out by one of the authors in which the central composite rotatable design technique had been used to develop a mathematical model correlating the five welding process variables of welding current (I), welding speed (S), arc voltage (V), nozzle-to-plate distance (N), and electrode to work angle (�) to the weld bead penetration in flux cored arc welding (FCAW) of a grade of high strength low alloy (HSLA) steels. This is a useful technique which has been used by some researches [2-4,7,9,13-17].This technique © Copyright by International OCSCO World Press. All rights reserved. 2006 Analysis and modelling was able to predict the linear, quadratic and two factor interactive effects [18-21] of process parameters on weld penetration. Using this model or equation the Sequential Quadratic Programming (SQP) which is an optimization method for constrained optimization problems [22] was used to maximize weld bead penetration thus achieving a high quality joint. where y, Response variable or yield, e.g. penetration. 2. Experiments details 2. Experimental details Table 1 Central composite rotatable design matrix for five factors, k=5 Independent welding factors such as welding current (I), welding speed (s), etc. b0 , b1 ,..., b5 , b12 , b13 ,..., b45 , b11 ,..., b55 , Regression coefficients. Factors 2.1. Materials 2. 1. Materials A grade of HSLA steel of 12 mm thickness had been used as the plate material. Flux cored wire of 1.6 mm diameter (AWS classification E70T5) with 100% CO2 shielding gas had also been employed on a semi automatic flux cored arc welding (FCAW) machine. 2.2. Experimental design 2. 2. Experimental design The required design matrix (table 1) with a total of 32 experimental runs had been developed for the five welding factors selected. The upper (+2) and lower (-2) levels of all the five variables as shown in table 2 had been established by trial runs prior to the actual welding to ensure deposition of an acceptable weld bead. The intermediate levels of -1, 0, +1 of all the variables had been calculated by interpolation. 2. 3. Welding procedure and penetration 2.3. Welding procedure and penetration measurement measurement Bead-on-plate welds were deposited randomly according to the levels of parameters in the design matrix. Specimens were cut off from these plates and depth of penetrations was measured using a profile projector at a magnification of 10. 3. Selection Selection and development 3. and development of the of the model model A model relating the depth of penetration to the levels of one or more welding factors is an indispensable aid in the interpretation of results from an experimental design. Wide experience had shown that quadratic relationships are usually adequate and so most surface designs in the literature are for fitting a quadratic equation [18,20]. Based on these considerations the following second degree polynomial relating the response or penetration to all the five welding variables was assumed. y � b0 � b1 x1 � b2 x2 � b3 x3 � b4 x4 � b5 x5 � b11 x12 � b22 x22 � b33 x32 � b44 x42 � b55 x52 � b12 x1 x2 � b13 x1 x3 � b14 x1 x4 � b15 x1 x5� b23 x2 x3 � ... x1 ,..., x5 , (1) S.No. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. x0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 x1 -1 1 -1 1 -1 1 -1 1 -1 1 -1 1 -1 1 -1 1 -2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 x2 -1 -1 1 1 -1 -1 1 1 -1 -1 1 1 -1 -1 1 1 0 0 -2 2 0 0 0 0 0 0 0 0 0 0 0 0 x3 -1 -1 -1 -1 -1 -1 -1 -1 1 1 1 1 1 1 1 1 0 0 0 0 -2 2 0 0 0 0 0 0 0 0 0 0 x4 -1 -1 -1 -1 -1 -1 -1 -1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 -2 2 0 0 0 0 0 0 0 0 x5 � x1 x2 x3 x4 1 -1 -1 1 -1 1 1 -1 -1 1 1 -1 1 -1 -1 1 0 0 0 0 0 0 0 0 -2 2 0 0 0 0 0 0 x0 , A dummy variable x1 , Welding current (I) x2 , Welding speed (S) x3 , Arc voltage (V) x4 , Nozzle-to-plate distance (N) x5 , Electrode-to-work angle (�) READING DIRECT: www.journalamme.org 133 Journal of Achievements in Materials and Manufacturing Engineering Table 2 Factors and their levels N o 1 2 3 4 5 Factors Welding Current Welding speed Arc voltage Nozzle to plate distance Electrode to work angle Unit Nota tion Volume 16 Issue 1-2 May-June 2006 P = 3.531 + 0.333 * I - 0.116 * S + 0.2 * � + 0.112 * I * S - 0.175 * S * V + 0.162 * V * N + 0.066 * V – 0.058 * N (2) Level -2 -1 27 5 0 +1 +2 300 325 350 Amp I 250 Cm/min S 35 40 45 50 55 Volt V 27 29 31 33 35 mm N 15 17 20 22 25 degree � 90 97 105 112 120 The model so developed could be utilized in predicting the value of depth of penetration for each given set of welding variables by inserting the coded values of the variables in the above equation. In addition to the adequacy test performed, the validity of the results was also tested with the help of scatter diagram Fig. 1. As is evident from this figure there is a fairly good correlation between the observed and predicted value of the bead penetration. Using a computer program all the regression coefficients of the model were computed for the response or depth of penetration. Table 3 Significant coefficients for penetration Coefficient Penetration b0 b1 b2 b3 b4 b5 b12 b13 b14 b15 b23 b24 b25 b34 b35 b45 b11 b22 b33 b44 b55 3.531 0.333 -0.116 0.066* -0.058* 0.200 0.112 -------0.175 ----0.162 --------------- � Insignificant coefficients The developed model was tested for adequacy by applying the analysis of variance (ANOVA) and analysis of regression techniques. Similarly to test the significance of regression coefficients the Student’s test was employed to find the insignificant terms for 95% confidence level. Once the insignificant terms were identified they were dropped from the model without sacrificing the adequacy of the model. The significant coefficients are shown in table 3. Using the significant regression coefficients the final model for penetration could be constructed as given below: 134 Research paper Fig. 1. Scatter diagram showing observed vs. predicted penetration 4. Definition of optimization 4. Definition problemof optimization problem An optimization or a mathematical programming problem can be stated as follows: Find � x1 � �x � � 2� X �� � �� � �� xn �� which minimizes f(x) Subject to constraints Gi (x) = 0, i = 1, …, me Gi (x) <= 0, I = me+1, …, m (3) where x is the vector of length n design parameters, f(x) is the objective function, which returns a scalar value, and the vector N.B. Mostafa , M.N. Khajavi Analysis and modelling function G(x) returns a vector of length m containing the values of the equality and inequality constraints evaluated at x . An efficient and accurate solution to this problem depends not only on the size of the problem in terms of the number of constraints and design variables but also on characteristics of the objective function and constraints. When both the objective function and the constraints are linear functions of the design variable, the problem is known as a Linear Programming (LP) problem. Quadratic Programming (QP) concerns the minimization or maximization of a quadratic objective function that is linearly constrained. For both the LP and QP problems, reliable solution procedures are readily available. More difficult to solve is the Nonlinear Programming (NP) problem in which the objective function and constraints can be nonlinear functions of the design variables. A solution of the NP problem generally requires an iterative procedure to establish a direction of search at each major iteration. This is usually achieved by the solution of an LP, a QP, or an unconstrained subproblem. multipliers (�i , i = 1,…,m ) are necessary to balance the deviations in magnitude of the objective function and constraint gradients. Because only active constraints are included in this canceling operation, constraints that are not active must not be included in this operation and so are given Lagrange multipliers equal to zero. This is stated implicitly in the last two equations of Eq. 3. The solution of the KT equations forms the basis to many nonlinear programming algorithms. These algorithms attempt to compute the Lagrange multipliers directly. Constrained quasi-Newton methods guarantee superlinear convergence by accumulating second order information regarding the KT equations using a quasi-Newton updating procedure. These methods are commonly referred to as Sequential Quadratic Programming (SQP) methods, since a QP subproblem is solved at each major iteration (also known as Iterative Quadratic Programming, Recursive Quadratic Programming, and Constrained Variable Metric methods). 4.1. Constrained optimisation 4. 1. Constrained optimization 4. 2. Sequential Quadratic Programming 4.2. Sequential Quadratic (SQP)Programming (SQP) In constrained optimization, the general aim is to transform the problem into an easier subproblem that can then be solved and used as the basis of an iterative process. A characteristic of a large class of early methods is the translation of the constrained problem to a basic unconstrained problem by using a penalty function for constraints that are near or beyond the constraint boundary. In this way the constrained problem is solved using a sequence of parameterized unconstrained optimizations, which in the limit (of the sequence) converge to the constrained problem. These methods are now considered relatively inefficient and have been replaced by methods that have focused on the solution of the Kuhn-Tucker (KT) equations. The KT equations are necessary conditions for optimality for a constrained optimization problem. If the problem is a so-called convex programming problem, that is, f(x) and Gi(x), i=1,…,m are convex functions, then the KT equations are both necessary and sufficient for a global solution point. A function f(x) is said to be convex if for any pair of points X1 and X2 and all �, 0 � � � 1 SQP methods represent the state of the art in nonlinear programming methods. Schittkowski [23], for example, has implemented and tested a version that outperforms every other tested method in terms of efficiency, accuracy, and percentage of successful solutions, over a large number of test problems. Based on the work of Biggs [24], Han [25], and Powell ([26,27]), the method allows you to closely mimic Newton’s method for constrained optimization just as is done for unconstrained optimization. At each major iteration, an approximation is made of the Hessian of the Lagrangian function using a quasi-Newton updating method. This is then used to generate a QP subproblem whose solution is used to form a search direction for a line search procedure. An overview of SQP is found in Fletcher [28], Gill et al. [29], Powell [30], and Schittkowski [31]. The general method, however, is stated here. Given the definition of optimization problem (Eq. 3) the principal idea is the formulation of a QP subproblem based on a quadratic approximation of the Lagrangian function. f [�X 2 � (1 � � )X1 ] � �f (X 2 ) � (1 � � )f (X1 ) that is, if the segment joining the two points lies entirely above or on the graph of f(X). Referring to the definition of optimization problem (Eq. 3) the Kuhn-Tucker equations can be stated as �f ( X � ) � � � .�G (X ) � 0 ��i .G i (X * ) ��i � 0 m i �1 �0 � i i � i � 1,..., m i � m e � 1,..., m (4) The first equation describes a canceling of the gradients between the objective function and the active constraints at the solution point. For the gradients to be canceled, Lagrange Optimisation of welding parameters for weld penetration in FCAW L ( X, � ) � f ( X ) � � � .g (X) m i �1 i i (5) Here Eq. 3 is simplified by assuming that bound constraints have been expressed as inequality constraints. You obtain the QP subproblem by linearizing the nonlinear constraints. 4.3. Quadratic Problem 4. 3. Quadratic Problem (QP)(QP) Subproblem Subproblem The quadratic problem is as follows: 1 T d H k d � �f ( X k ) T d 2 �g i ( X k ) T d � g i ( X k ) � 0 i � 1...., m e Minimize 135 Journal of Achievements in Materials and Manufacturing Engineering �g i (X k ) T d � g i (X k ) � 0 i � m e � 1,..., m (6) This subproblem can be solved using any QP algorithm. The solution is used to form a new iterate of the following form X k �1 � X k � � k d k Volume 16 Issue 1-2 May-June 2006 definite matrix. When q Tk s k is not positive, q k is modified on an element-by-element basis so that q Tk s k � 0 . The general aim of this modification is to distort the elements of q k , which contribute to a positive definite update, as little as possible. Therefore, in the initial phase of the modification, the most � The step length parameter is determined by an appropriate line search procedure so that a sufficient decrease in a merit function is obtained. The matrix Hk is a positive definite approximation of the Hessian matrix of the Lagrangian function (Eq. 5). Hk can be updated by any of the quasi-Newton methods, although the BFGS method appears to be the most popular. A nonlinearly constrained problem can often be solved in less iteration than an unconstrained problem using SQP. One of the reasons for this is that, because of limits on the feasible area, the optimizer can make informed decisions regarding directions of search and step length. negative element of q k * s k is repeatedly halved. This procedure is continued until q Tk s k is greater than or equal to 1e-5. If, after T this procedure, q k s k is still not positive, modify q k by adding a vector v multiplied by a constant scalar w, that is, q k � q k � wv where v i � �g i ( x k �1 ).g i ( x k �1 ) � �g i ( x k ).g i ( x k ) if (q k ) i .w � 0 and (q k ) i .(s k ) i � 0 4.3. SQP implementation 4. 3. SQP implementation v i � 0 otherwise. The SQP implementation consists of three main stages: •Updating of the Hessian matrix of the Lagrangian function •Quadratic programming problem solution •Line search and merit function calculation These stages are explained briefly (i � 1,..., m) T and increase w systematically until q k s k becomes positive. Quadratic Programming problem solution Updating the Hessian matrix At each major iteration a positive definite quasi-Newton approximation of the Hessian of the Lagrangian function, H, is calculated using the BFGS method, where �i ,( i =1,…,m) is an estimate of the Lagrange multipliers. Hk � Hk � q k q Tk q Tk s k � H Tk H k Ai d � bi s k � X k �1 � X k �� � �� f ( X k ) � �� �� n i �1 n i �1 i i * �g i (X k �1 ) � � * �g i ( X k ) � �� T point. A positive definite Hessian is maintained providing q k s k is positive at each update and that H is initialized with a positive Research paper 1 T d Hd � c T d 2 i � 1,..., m e i � m e � 1,..., m The method used in the Matlab Optimization Toolbox is an active set strategy (also known as a projection method) similar to that of Gill et al., described in [32] and [33]. It has been modified for both Linear Programming (LP) and Quadratic Programming (QP) problems. The solution procedure involves two phases. The first phase involves the calculation of a feasible point (if one exists). The second phase involves the generation of an iterative sequence of feasible points that converge to the solution. In this method an Powell [27] recommends keeping the Hessian positive definite even though it might be positive indefinite at the solution 136 Minimize q (d ) � Ai d � bi s Tk H k s k where q k � �f ( x k �1 ) � At each major iteration of the SQP method, a QP problem of the following form is solved, where Ai refers to the ith row of the m-by-n matrix A. active set, A k is maintained that is an estimate of the active constraints (i.e., those that are on the constraint boundaries) at the solution point. Virtually all QP algorithms are active set methods. This point is emphasized because there exist many different methods that are very similar in structure but that are described in widely different terms. N.B. Mostafa , M.N. Khajavi Analysis and modelling Line search and merit function calculation The solution to the QP subproblem produces a vector dk , which is used to form a new iterate of the following form: X k �1 � X k � � k d k The step length parameter is determined in order to produce a sufficient decrease in a merit function. The merit function used by Han [25] and Powell [27] is used in Matlab Optimization toolbox. 5. Weld optimization 5. Weldpenetration penetration optimization problem formulation problem formulation The objective function for weld penetration which must be maximized was derived in section 3 of the paper as Eq 2. The constraints of the welding parameters are given in table 2. Here we restate these equations. Maximize: P = 3.531 + 0.333 * I - 0.116 * S + 0.2 * � + 0.112 * I * S - 0.175 * S * V + 0.162 * V * N + 0.066 *V - 0.058 * N (2) Subject to constraint: 250 � I � 350 27 � V � 35 15 � N � 25 90 � � � 120 35 � S � 55 Matlab Optimization toolbox and the function fmincon was used for this optimization problem. The fmincon function uses SQP method for optimization. Two m-file was written namely objfun and confun. In the objfun the objective function was defined and in the confun the constraints were set. Optimization result shows in order to attain the maximum weld penetration the first four factors must be at their maximum value and the last one must be at its minimum value. So the maximum weld penetration will be attained when the following values are chosen for the five factors: I = 350 amp V = 35 volt N = 25 mm � = 120 degree S = 35 cm/min Using these values the weld penetration (P) becomes 5.745 mm. 6. Results discussions 6. Results andand discussions In addition to the adequacy test performed, the validity of the results of mathematical modeling for prediction of depth of Optimisation of welding parameters for weld penetration in FCAW penetration was also tested with the help of a scatter diagram (Fig. 1). As is evident from this figure there is a fairly good correlation between the observed and predicted value of weld bead penetration. The optimization result shows penetration will be maximum when welding current , arc voltage, nozzle-to-plate distance and electrode-to-work angle are at their maximum possible value and welding speed is at its minimum value . The physical reasons for the above results are discussed below. Increase in welding current (I) increases the depth of penetration (P). It is known that molten metal droplets transferring from the electrode to the plate are strongly overheated. It can be reasonably assumed that this extra heat contributes to more melting of the work piece. As current increases the temperature of the droplets and hence the heat content of the droplets increases which results in more heat being transferred to the base plate. Increase in current reduces the size but increases the momentum of the droplets which on striking the weld pool causes a deeper penetration or indentation. The increase in penetration as current increased could also be attributed to the fact that enhanced arc force and heat input per unit length of the weld bead resulted in higher current density that caused melting a larger volume of the base metal and hence deeper penetration. Increase in welding (S) causes a decrease in depth of penetration (P). This may be attributed to lesser heat input at higher speeds per unit length of the weld bead which caused a smaller weld pool and decreased depth of penetration. Increase in arc welding voltage (V) resulted in an increase in depth of penetration (P), because of increase in heat inputs reported by McGlone and Chadwick [34 ]. Increase in electrode-to-work angle from 90° to 120° (i.e. for normal to backhand) had resulted in increase of depth of penetration. The backhand welding technique has been reported [35] to provide the best penetration, as the arc force keeps the slag from running in front of the liquid pool. Increase in nozzle-to-plate distance (N) also causes an increase in depth of penetration (P) which may be due to higher temperature of the droplets impinging on the weld pool because of more resistance heating of the wire at higher nozzle-to-plate distances. Based on this investigation it can be concluded that the developed model can be used to predict adequately the weld bead penetration within the specified limits of the process parameters . The optimization method can also be used to find optimum welding conditions for maximum weld bead penetration. References References [1] Z. Smati, , “Automatic pulsed MIG welding”, Metal Construction, Jan. 1986, pp. 38R-44R. [2] P. J Konkol. and G. F Koons,. “Optimization of parameters for two-wire AC-AC SAW”, Welding Journal, Dec. 1978, Vol. 27, No.12, pp. 367s-374s. [3] N. B. Mostafa and R. S Parmar. “Mathematical models to predict weld bead dimensions in FCAW’’ Int. Conf. on CAD/CAM, Robotics and Autonomous Factories, Dec. 1619, 1993, Vol. II, pp. 415-431. 137 Journal of Achievements in Materials and Manufacturing Engineering [4] K.Y. Benyounis, et al. “ Effect of welding parameters on heat input and weld bead profile ’’, Journal of Materials Processing Technology , 15 May 2005 , Vol.164-165, pp. 978-985, (AMPT/AMME05 part 2). [5] K.Y. Benyounis, et al.“ Optimizing the laser-welded butt joints of medium carbon steel using RSM ”, Journal of Materials Processing Technology , 15 May 2005 , Vol.164165, pp. 986-989, (AMPT/AMME05 part 2). [6] Xue Yu , et al. “ Fuzzy regression method for prediction and control the bead width in the robotic arc-welding process” , Journal of Materials Processing Technology , 15 May 2005 , Vol.164-165, pp. 1134-1139, (AMPT/AMME05 part 2). [7] Davi Sampaio , et al. “Comparison between generic algorithms and RSM in GMAW optimization”, Journal of Materials Processing Technology , 1 March 2005,Vol.160, Issue 1, pp. 70-76. [8] I.S Kim., et al. “ An investigation into an intelligent system for predicting bead geometry in GMA welding process” , Journal of Materials Processing Technology , 10 Jan.2005 , Vol. 159, Issue 1, pp. 113-118. [9] I.S. Kim, et al. “ A study on relationship between process variables and bead penetration for robotic CO2 arc welding” , Journal of Materials Processing Technology , 10 May 2003 , Vol. 136, Issues 1-3, pp. 139-145. [10] I.S Kim., et al. “ A study on prediction of bead height in robotic arc welding using a neural network ”, Journal of Materials Processing Technology , 20 Dec. 2002 , Vol. 130131, pp. 229-234. [11] S.C Juang., et al. “ Process parameter selection for optimizing the weld pool geometry in the tungsten inert gas welding of stainless steel ”, Journal of Materials Processing Technology , 5 March 2002 , Vol. 122, Issue 1, pp. 33-37. [12] G Lothongkum., et al. “ Study on the effects of pulsed TIG welding parameters on delta-ferrite content , shape factor and bead quality in orbital welding of AISI 316L stainless steel plate”, Journal of Materials Processing Technology , 19 March 2001 , Vol. 110, Issue 2, pp.233-238. [13] G. Gunaraj and N. Murugan “ Application of response surface methodology for predicting weld bead quality in SAW of pipes ”, Journal of Materials Processing Technology , 15 April 1999 , Vol. 88, Issues 1-3, pp. 266-275. [14] L.J Yang., et al. “ Linear regression equations for modeling the SAW process ”, Journal of Materials Processing Technology , Oct. 1993 , Vol. 39, Issues 1-2, pp. 33-42. [15] S.C.A Alfaro., et al. “ The use of statistical modeling in welding ”, Journal of Materials Processing Technology , Feb. 1993 , Vol. 38, Issues 1-2, pp. 399-405. [16] L.J Yang., et al. “ An analysis of curvilinear regression equations for modeling the SAW process ”, Journal of Materials Processing Technology , Feb. 1993 , Vol. 37, Issues 1-4, pp. 601-611. [17] N Murugan., et al. “ Effect of submerged arc process variables on dilution and bead geometry in single wire. 138 Research paper Volume 16 Issue 1-2 May-June 2006 surfacing ” Journal of Materials Processing Technology , Feb. 1993 , Vol. 37, Issues 1-4, pp. 767-780. [18] H. E. Hill, and J. W. Prane, “Applied Techniques in Statistics for selected Industries’, John Wiley & Sons, New York, 1984. [19] G. E. R. Box, W. G Hunter, J. S and Hunter. “Statistics for experiments”, John Wiley & Sons, New York, 1978. [20] W. G.Cochran, and G. M.Cox, “Experimental design”, Asia Publishing House, 1963. [21] S Akhanazaroa,. and V. Kafarov, “Experiment optimization in chemistry and chemical engineering”, Mir publishers, Moscow, 1982. [22] Optimization Toolbox User’s Guide, 2004, The Mathworks, Inc. [23] K Schittkowski,., “NLQPL: A FORTRAN-Subroutine Solving Constrained Nonlinear Programming Problems,” Annals of Operations Research, Vol. 5, pp 485-500, 1985. [24] M.C Biggs,., “Constrained Minimization Using Recursive Quadratic Programming,” Towards Global Optimization (L.C.W. Dixon and G.P. Szergo, eds.), North-Holland, pp 341349, 1975. [25] , S.P Han., “A Globally Convergent Method for Nonlinear Programming,” J. Optimization Theory and Applications, Vol. 22, p. 297, 1977. [26] M.J.D Powell,., “The Convergence of Variable Metric Methods for Nonlinearly Constrained Optimization Calculations,” Nonlinear Programming 3, (O.L. Mangasarian, R.R. Meyer and S.M. Robinson, eds.), Academic Press, 1978. [27] M.J.D. Powell, , “A Fast Algorithm for Nonlinearly Constrained Optimization Calculations,” Numerical Analysis, G.A.Watson ed., Lecture Notes in Mathematics, Springer Verlag, Vol. 630, 1978. [28] R. Fletcher, , “Practical Methods of Optimization,” John Wiley and Sons, 1987. [29] P.E. Gill, , W. Murray, and M.H.Wright, Practical Optimization, London, Academic Press, 1981. [30] M.J.D Powell,., “Variable Metric Methods for Constrained Optimization,” Mathematical Programming: The State of the Art, (A. Bachem, M. Grotschel and B. Korte, eds.) Springer Verlag, pp 288-311, 1983. [31] W Hock,. and K. Schittkowski, “A Comparative Performance Evaluation of 27 Nonlinear Programming Codes,” Computing, Vol. 30, p. 335, 1983. [32] P.E. Gill, , W. Murray, and M.H. Wright, Numerical Linear Algebra and Optimization, Vol. 1, Addison Wesley, 1991. [33] , P.E Gill, W. Murray, M.A. Saunders, and M.H. Wright, “Procedures for Optimization Problems with a Mixture of Bounds and General Linear Constraints,” ACM Trans. Math. Software, Vol. 10, pp 282-298, 1984. [34] J. C. McGlone, and , D. B. Chadwick “The Submerged Arc Butt Welding of Bead Geometry from Procedure Parameters” , IIW Report 80/1978/PE, 1978. [35] P. Houldcroft, and R. John, “Welding and Cutting, a Guide to Fusion Welding and Associated Cutting Processes”,Industrial Press Inc., New York, 1989. N.B. Mostafa , M.N. Khajavi