A genetic algorithm for solving a multi-floor layout design model of a cellular manufacturing system with alternative process routings and flexible configuration

This paper presents a novel integer linear programming model for designing multi-floor layout of cellular manufacturing systems (CMS). Three major and interrelated decisions are involved in the design of a CMS; namely cell formation (CF), group layout (GL), and group scheduling (GS). A novel aspect of this model is concurrently making the CF and GL decisions to achieve an optimal design solution in a multi-floor factory. Other compromising aspects are: multi-floor layout to form cells in different floors is considered, multi-rows layout of equal area facilities in each cell is allowed, cells in flexible shapes are configured, and material handling cost based on the distance between the locations assigned to machines are calculated. Such an integrated CMS model with an extensive coverage of important manufacturing features has not been proposed before and this model incorporates several design features including alternative process routings, operation sequence, processing time, production volume of parts, duplicate machines, machine capacity, new machine purchasing, lot splitting, material flow between machines, intra-cell layout, inter-cell layout, multi-floor layout and flexible configuration. The objective is to minimize the total costs of intra-cell, inter-cell, and inter-floor material handling, new machines purchasing and machine processing. Two numerical examples are solved by the Lingo software to verify the performance of the proposed model and illustrate the model features. Sensitive analysis is also implemented on some model parameters. An improved genetic algorithm (GA) is proposed to derive near-optimal solutions for the integrated model because of its NP hardness. It is then tested using several problems with different sizes and settings to verify the computational efficiency of the developed algorithm in comparison to a classic simulated annealing algorithm and the Lingo software. The obtained results show the efficiency of proposed GA in terms of objective function value and computational time.