Hochflexibles Workforce Management

It can be observed that companies tend to use a very demand driven personnel scheduling instead of using fixed shifts. In this context the term highly flexible workforce management (WFM) is used. With instruments such as the planning of subdaily workplace rotations, the combination of working time model generation and personnel scheduling or the combination of personnel scheduling and vehicle routing the demand for personnel can be covered very well. Such problems are novel and found little attention by researchers up to now.In this work classical OR-algorithms, metaheuristics and multi-agent systems (MAS) are evaluated on real world problems from logistics, retail and British Telecom. It can be shown, that classical OR-algorithms are not appropriate for these problems of highly flexible WFM, because of impractical CPU-times. On the other hand selected metaheuristics are very suitable. MAS should not be favoured, because selected metaheuristics performed always better. It must point out that a hybrid algorithm (a metaheuristic with a problem-specific repair) is responsible for the success of metaheuristics. MAS lack of a central planning instance that makes major changes for which agents are not able to do. Numerous algorithms of this work where originally developed for continuous problems. The adaption to combinatorial problems is described too. The appropriate adaption of parameters is also addressed.