OPTIMIZATION OF PROCESS VARIABLES. AFFECTING THE QUALITY OF A141% Si ALLOY CASTINGS PRODUCED BY V-PROCESS

Abstract

Vacuum sealed moulding process, popularly called the V-process was developed in Japan in 1971. Since then it has gained considerable importance due to its capability to produce dimensionally accurate and smooth castings. It can be employed in making casting of all sizes and shapes in most metals. The V-process can be profitably employed in making aluminium alloys castings requiring surface finish and dimensional tolerances approaching that of the die-casting process but where production requirements do not justify the die-casting tooling cost. Critical review of the published literature indicates that the V-process is commercially viable and number of plants have already been installed worldwide under licence agreement with Sintokogio, Japan. However, the available literature lacks in systematic investigation of the effect of different process variables on the quality of castings produced by the V-process. Moreover, very little effort has been made to obtain an optimal set of process variables that may yield the optimum quality of the castings. The objective of the present work is to optimize various process variables that affect the quality of the castings produced by the V-process. In the present work, following process variables have been selected : (i) Sand grain size and its distribution (ii) Percent zircon sand mixed with silica sand of AFS 50. (iii)Vibrating frequency (iv) Vibrating time (v) Degree of vacuum imposed (vi) Metal pouring temperature The thickness and type of plastic film (polyethylene film), composition of alloy, pouring time of the molten metal and distribution pattern of the sand (3-screen) were kept fixed during the entire investigation. ii The aluminium-silicon alloys generally have excellent casting properties and very good resistance to corrosion; the A1-11% Si alloy is best in this latter respect. This alloy is suitable for exposed marine castings, motor car and lorry fittings, and engine parts and generally wherever corrosion resistance and ease of casting are essential. Hence for the present work A1-11% Si alloy has been selected. In order to evaluate the effect of selected process variables on the characteristics of sand mould and castings, the Response Surface Methodology (RSM) has been used to correlate the independent process variables with the desired quality characteristics by a mathematical model. The second order response surface is found suitable for the present work. The following scheme is used for the development of various response surface. (1) Selection of V-process variables and their operating range. (ii) Selection of response characteristics. (111)Experimental design (iv) Coding of variables as required by the selected scheme of experiments. (v) Experimentation and collection of desired data. (vi) Analysis of data and estimation of coefficient of assumed regression equations. (vii)Testing the adequacy of the model. Central composite rotatable experimental designs have been adopted to plan the experiments. These designs provide the necessary information with minimum experimental trials. The data obtained through the experiments have been used to estimate the coefficients of the assumed regression equation. The following sand mould and casting quality characteristics have been analyzed : 1. Sand mould characteristics : (i) Bulk density (ii) Packing factor iii (iii)Void Ratio (iv) Mould Hardness (v) Coefficient of relative compressibility of sand. The sand mould characteristics have been measured in a specially designed sand testing rig that simulates the actual conditions of V-process. 2. Casting quality characteristics : (i) Dimensional accuracy (ii) Surface finish (iii)Casting soundness (iv) Mechanical properties The importance of dimensional accuracy is realized by the fact that if castings can be produced to the designed dimensions, it will result in reduced finishing cost as well as save time and energy. A regression model is developed to predict the pattern dimensions so that the desired casting dimensions can be obtained. This model is also validated by actually producing the castings and then comparing with the estimated casting dimensions. The deviation of castings dimension from pattern dimension for the Set II and Set III are found to be ± 2.2% and ± 1.4% respectively. Casting surface finish has assumed great importance to foundrymen due to more emphasis on using as cast components as well as on appearance. In order to analyse the effect of selected process variables on surface roughness of the castings, two regression equations, one each for Set II and Set III have been developed. It is found that the sand grain size as well as their packing characteristics on the mould surfaces, the surface tension of the metal poured and degree of vacuum imposed are the main factors responsible for casting surface roughness variations. In Set III, the higher the percent zircon added better the surface finish of the castings. iv Casting soundness is quantified by measuring volume porosity. A regression model that relates volume porosity with the selected process variables has been developed and analysed. The as cast microstructure is evaluated by measuring dendrite arm spacing. The dendrite arm spacing (DAS) is correlated with the pouring temperature and the local solidification time. It is observed that the dendrite arm spacing increases with the increase in the pouring temperature and local solidification time. Various mechanical properties viz. tensile strength, percent elongation, impact strength and hardness have also been measured. The mechanical properties are correlated with the DAS. All the mechanical properties improve with decrease in the dendrite arm spacings. Finally, since the various quality characteristics respond differently to various selected process variables, a set of optimal values of the process variables is to be obtained. Since more than one quality characteristic are to be optimized simultaneously then the problem may be formulated as multiobjective optimization problem. In multiobjective problem it is rarely possible to find a feasible solution where all the objectives achieve their optimal values. The present problem is viewed as the minimization of deviations of quality characteristics from their specified goals. In most of the real life problems the nature of these goals is fuzzy . Hence in the present work a multiobjective goal programming model using the concept of fuzzy set theory has been developed. Price's controlled Random Search Technique for global optimization is used in the present work to obtain an optimal set of process variables that may yield the optimum quality characteristics of the Al-1l% Si alloy castings produced by the V-process. The proposed multiobjective model is used to determine optimal variables for different set of users' priorities or weightage for goals. The proposed model is validated by actually producing castings using the predicted optimal variables and then comparing the predicted characteristics with the actual ones. It is found that there is no significant difference between the predicted and the actual characteristics. This model is valid within the selected range of process variables. The chapter wise break up of the present thesis is as following : The first chapter deals with the general introduction, V-process technique, statement of the problem and objectives of the problem. The second chapter presents the review of the published literature regarding the V-process. In the third chapter the details of the experimental set-up is given. The fourth chapter deals with the response surface methodology and experimental design technique and its analysis. A regression model for the prediction of pattern dimensions has also been given in this chapter. The fifth chapter deals with the selection of process variables and experimentation. The different casting qualities and sand properties have been measured and are reported in this chapter. The sixth and seventh chapter present the analysis of results and discussion regarding sand and casting, quality characteristics respectively. In the eighth chapter optimization models using the concepts of fuzzy sets logic have been developed and are solved by Random search technique. Recommendations regarding the optimal values of variables have also been given in this chapter. Finally in the ninth chapter conclusions of the results and discussions presented in chapter 6-8 have been summarized. Also, at the end of this chapter, some suggestions for further work on related topics have been enumerated.