Solving a real world project scheduling problem with a genetic approach

This work is related to a project scheduling problem encountered in the Technical Division of ELCAT S.P.A., a factory located in North-Western Italy which produces vehicle seats and uses the Work Breakdown Structure (WBS) as Project Management Model for all its activities. It is illustrated how the WBS formalism can be mapped (as far as the Technical Division is concerned) as a Job Shop Problem with precedence constraints and Total Weighted Tardiness as objective function. A solution procedure based on two algorithms linked together is proposed. The former is a genetic algorithm for the Job Shop while the latter is an extended neighbourhood search which uses the result of the genetic algorithm as input. A real example is presented and an intensive computational test is outlined and commented upon.