Teaching learning based optimization of squeeze casting process for quality castings

Abstract

The hybrid squeeze casting method combines the immense features of forging and casting processes. Ultimate tensile strength (UTS), surface roughness (SR), and yield strength (YS) are the important casting quality characteristics influenced mainly by input variables. Determining the set of optimal input variables (pressurized duration, squeeze pressure, pouring and die temperature) for conflicting requirement in casting quality (maximize: YS and UTS and minimize: SR) is considered to behave complex and non-linear. Multiple objective functions with conflicting requirements are modified suitably to single fitness function with different set of weight fractions for maximization to solve the optimization problem. Four scenarios are selected with different sets of weight fractions by assigning equal weights for all outputs, followed by assigning maximum weight fraction for each individual output function, after keeping the rest at fixed low equal weights. Teaching learning based optimization (TLBO) is selected to optimize the squeeze casting input-output parameters. TLBO results are found to be comparable with evolutionary algorithms (i.e. GA, PSO and MOPSO-CD). TLBO has outperformed the evolutionary algorithms with regard to computation time.