Off-the-Peg or Made-to-Measure? Timetabling and Scheduling with SA and TS

Modern heuristic search techniques such as simulated annealing (SA) and tabu search (TS) are particularly suited to solving problems with a mix of hard and soft constraints or hierarchies of objectives such as those commonly encountered in real-life timetabling and scheduling problems. However, it is well-known that such methods are sensitive to the way in which the problem is modelled within a local search framework and to the generic parameters used within the algorithm. This not only raises questions concerning the robustness of a particular implementation when faced with changes in data characteristics or problem specification but also casts doubt as to the extent to which features from a solution to one family of timetabling problems may be successfully incorporated into a solution to another. This paper examines these issues from a personal point of view and uses case-studies of scheduling, timetabling and staff-rostering problems arising in the education and hospital sectors to show that it is possible to design robust solutions based on SA and TS and that lessons learned when tackling one family of problems are frequently useful for another.