Swap based process schedule optimization using discrete-event simulation

Large construction projects usually involve many tasks, which are connected through dependencies and usage of common resources and materials. Determining the optimal order of task execution is in most of the cases impossible to do by hand. Therefore different methods for automatic optimization of large process schedules using a discrete-event simulation system were investigated. This paper introduces a new heuristic method for the resource constrained project scheduling problem, called swap-based optimization. Compared to creating an optimal schedule from scratch, the swap approach facilitates obtaining metrics about the performance of the result, before having worked through the entire construction process. Swaps are introduced into the simulation model by assigning priorities to the tasks. After running a simulation a list of possible swaps is created. Applying one of them and restarting the simulation will introduce a change into the sequence of the tasks within the schedule, generating a different schedule than the one before. Different tree search algorithms, traversing the space of possible swaps throughout a construction process, were analyzed. The suitability of the method is proven by an extensive case study.