M. Siva Kumar

A new algorithm for minimizing surplus parts in selective assembly by using genetic algorithm

Precision assemblies are produced from low precision subcomponents by partitioning and assembling them randomly from their corresponding groups. Surplus part is one of the important issues, which reduces the implementation of selective assembly in real situations. A new algorithm is introduced in this present paper to reduce surplus parts almost to zero and it is achieved in two stages by using a genetic algorithm. For demonstrating the proposed algorithm, a gearbox shaft assembly is considered as an example problem in which the shaft and pulley are manufactured in wider tolerance and partitioned in three to nine bins. The surplus parts are divided into three bins equally and a best combination of groups is obtained for both cases. It is observed that nearly 995 assemblies are produced out of one thousand subcomponents with the manufacturing cost savings of 19.5% for T max and 992 assemblies are produced with 13.5% saving in manufacturing cost for 0.9T max.

A simple heuristic for linear sequencing of machines in layout design

This paper presents a simple heuristic to determine a common linear machine sequence for multiple products with different operation sequences and a limited number of duplicate machine types available for the job. The heuristic is based on minimisation of the total flow distance travelled by a product on the linear machine sequence. It is assumed that the flows of products are allowed only in the forward direction, either in-sequence or by-pass. It is also assumed that backtrack movements are not allowed. The effectiveness of the proposed heuristic is demonstrated through the solutions of two typical layout design problems taken from the literature. Subsequently, a number of additional problems are solved and their results are compared with the results applying existing methods. The results indicate that the proposed method can be an effective tool in solving layout design problems.

Construction of closed-form equations and graphical representation for optimal tolerance allocation

The Lagrange multiplier method (LM) is currently used to allocate tolerance for optimum manufacturing cost. This is a tedious iterative process and sometimes it allocates a components tolerance outside its process tolerance limits. The present work develops a graphical representation which can help the process engineer to visualize the minimum and maximum values for assembly tolerances. The graphical representation developed can also help the process engineer to determine the exact total manufacturing cost of the assembly and help to fix the tolerance, which would not fall outside the limits prescribed. A simple C program is developed to construct the closed-form equations (CFE), and a single EXCEL graphical representation is derived for assembly tolerance, allocated tolerance, and total manufacturing cost. The developed algorithm has been demonstrated on a two- to five-component linear assembly, to help the process/manufacturing engineers visualization before determining the tolerance specification on component dimensions. The test results show a maximum percentage deviation of 0.09% of assembly tolerance and 0.33% of total manufacturing cost between the LM and the newly developed CFE method.

A Tabu Search for Multi-Objective Single Row Facility Layout Problem

This paper addresses the problem of multi-objective facility layout planning. The aim is to solve the single row facility layout problems (SRFLP) and find the linear machine sequence which minimizes the following: The total investment cost of machines; the total material handling cost; the total number of machines in the final sequence; and the total flow distance of the products in units. The tabu search algorithm (TSA) which has now become a very useful tool in solving a variety of combinatorial optimization problems is made use of here. TSA is developed to determine the product sequence based on which a common linear machine sequence is found out for multi-products with different machine sequences. We assume that, limited number of duplicate machine types available for job. The results are compared with other approaches and it shows the effectiveness of the TSA approach as a practical decision support tool to solve problems in SRFLP.

Optimum tolerance synthesis of simple assemblies with nominal dimension selection using genetic algorithm

Optimum tolerance allocation plays a vital role in minimization of the direct manufacturing cost, and it is sensitive to tolerances related to variations in manufacturing processes. However, optimal adjustment of both nominal dimensions and selection of tolerances may further reduce assembly manufacturing cost and wastage of materials during processing. Most studies in existing literature focus on optimum tolerance allocation for the assemblies without considering nominal dimension selection. The method proposed in this work uses genetic algorithm techniques to allocate tolerances to assembly components, thereby minimizing costs. The component alternate nominal dimensions are predicted based on critical dimensions and its tolerances. The effectiveness of the developed algorithms demonstrated using randomly generated problems as well as sample problems taken from the literature. Test results are compared with those obtained using the Lagrange multiplier method. It is shown that by adjusting the nominal dimensions, the proposed method yields considerable savings in manufacturing costs.

Tolerance allocation of complex assembly with nominal dimension selection using Artificial Bee Colony algorithm

Optimum tolerance allocation plays a vital role in minimizing the direct manufacturing cost of mechanical assembly. It is very sensitive due to the variations in manufacturing processes of the components. Most of the earlier studies are aiming at optimum tolerance allocation for assemblies without considering the selection of nominal dimensions of components and considering them as discrete values. It is proposed to minimize the manufacturing cost of an assembly with tolerance allocation and alternate nominal dimension selections by considering them in closer decimal intervals. The evolutionary algorithms such as Genetic and Artificial Bee Colony algorithms are developed and proposed to achieve the above objectives. The performance of the algorithms has been enhanced with the seed solution obtained using Lagrange Multiplier method. The complex assembly problems proposed by various authors with the required parameters have been considered for investigating the proposed method. The critical dimensions of the assemblies are fixed and the nominal dimension has been varied with its tolerances. The resultant manufacturing cost by various methods is presented and compared with corresponding nominal dimensions and tolerances. Based on the percentage of improvement of manufacturing cost, it is observed that the Artificial Bee Colony algorithm outperforms.