Practical decision tools for parallel batch scheduling of finish-machined pressure die castings

摘要

The paper summarises practical work that was conducted to solve the problem of distributing work on a small machining cell consisting of four non identical machining / drilling centres. Work consists of a number of batches of parts that have been pressure die-cast and need subsequent drilling and threading work. There are two main cases: a small number of batches, less than the number of available machines, and a large number of batches. The first case leads to batch splitting and has been tackled with a mathematical programming approach based on formulating the problem and solving it on Excel?-Solver?. The second case was tackled with genetic algorithms constructed on Matlab?. Comparison shows the limitations of conventional techniques with respect to problem size and combinatorial explosion and the relative simplicity and power of the genetic approach, yet at the expense of possible sub-optimality. In addition, a brief comparison of genetic algorithms to a heuristic proved superiority of the former.