Optimizing Task Allocation on Fire Fighting

In this work we consider the problem of allocating tasks to fire brigade roles when facing major incidents like conflagrations or natural disasters. Due to missing experience with major incidents, a balanced workload of the different roles cannot be guaranteed, as task allocation is usually done manually by operations managers based on their empirical knowledge. Concerning this, we will introduce a new concept of role specific penalty functions that assign a stress level to every role depending on its working time. As the resulting problem is NP-complete, we will use an optimization strategy that combines a greedy approach with a modified version of the simulated annealing algorithm to approximatively solve this optimization problem. In our experimental study we will see that the assignment of tasks to roles computed by this algorithm on an empirical data set leads to a smaller total processing time and to a more balanced workload of roles compared to action plan recommendations given by six operations managers from the fire brigade.