Košice 15.- 16.11.2007 University of Brescia, Mechanical and Industrial Engineering Department IMPLEMENTING ADVANCED SCHEDULING TECHNIQUES IN INDUSTRIAL ENVIRONMENTS: EVIDENCES AND OUTCOMES Andrea PASOTTI, Lucio ZAVANELLA Abstract: Production planning and scheduling is a crucial part of operations management. In the Italian industrial scenario some relevant cases have already been intercepted. As an example, it was found that in some circumstances additional details should be included into optimization models, since it was revealed that neglecting a number of aspects could bring to an inapplicable solution. The present paper is intended to show a significant sample of manufacturing cases, useful to infer important insights for production planning and scheduling applications. Each case is structured by contextualization, problem explanation, solving techniques and, where data are available, benefit estimation for advanced tools employment. The conclusions are summarized at the end of the paper. Key words: Metal working; scheduling; advanced techniques Introduction Scheduling problems fall into the well known NP-hard class [14] and have attracted many researchers, since they are easy to formulate, though difficult to solve. Earlier research has proposed numerous solving techniques, many of them being heuristic procedures to be exploited in order to approximate the optimal solution [16]. However, a proliferation of research on scheduling has produced a limited improvement in production planning practice, despite of the requests for more applicative models [15]. Among the major problems, we would like to mention some remarkable ones: minimization of makespan Cmax, minimization of maximum lateness Lmax, minimization of total Earliness ΣEj (or Tardiness ΣTj), minimization of total setup time. There are many more of them and several variations of these problems have gained plenty of attention (e.g. introducing weights for jobs). Above all, single machine problems belong to the simplest category and one of the most studied case is defined as the permutation scheduling problem, where an optimal sequence of jobs to be processed on the machine has to be found, minimizing an objective function specified beforehand. A well known link connects this issue to the Travelling Salesman Problem (TSP): this approach is commonly used to work out the puzzle. Many formulations of scheduling problems have been proposed since the fifties by means of linear programming (LP), mixed integer linear programming (MILP) or dynamic programming (DP). Moreover, a great amount of scheduling software is available on the market, but its relevance is confined in a limited set of (sometimes) easy applications or, else, it requires a long time for customisation or adaptation to a specific reality. Nevertheless, the enormity of existing problems still leaves space for some individual cases. These are the starting points of the present contribution, the main intent of which is to present three significant industrial cases belonging to the manufacturing field. Important insights inferred from the following cases are discussed in the conclusion. The paper is organized as follows: the three cases are presented consecutively, for each of them, the starting point is the context presentation, the problem explanation, the solving techniques and their potential benefit estimation; finally, our conclusions and remarks are synthetically offered. Case 1 The present case refers to the industrial environment of a factory with three departments: forming, painting line and assembly shops. Two intermediate buffers between departments are installed for the storage of work in process (WIP). Our study involves the central department of the factory, the schedule of which significantly influences the two adjacent departments. Historically, the firm adopted a push planning policy to manage the system. Starting from an annual forecast of sales, every week (i.e., the rolling period) a MPS (Master Production Schedule) is issued by an MRP (Material Requirement Planning) software. Such a plan contains a list specifying quantities, colours, release dates and due dates per item. Every production order, which we consider as an indivisible job (no pre-emption is admitted), has to be painted with the appropriate colour between its release date and due date. If not, the production plan is disregarded and an idle time cost and/or an inefficiency cost is determined. The powder painting line is made up by a closed moving chain for conveyance of differently shaped items manually hanged on hooks. Each of them may carry a known family of items, which receives the