Vehicle routing with endogenous learning: Application to offshore plug and abandonment campaign planning

Abstract

When a particular service is performed many times, the duration of the service might reduce due to the effect of learning from similar tasks that have been performed before. In this article, we present an approach to account for such learning effects that arise in the context of vehicle routing operations. Our approach enables the consideration of endogenous learning, where the service times are dependent on the experience that is to be gained in the same routing horizon. We apply our approach to the problem of planning an offshore plug and abandonment campaign, where different vessels are being used to perform plugging operations on offshore oil and gas wells. We extend existing instances for this problem with observed learning data and investigate the effects of learning and cooperation. Results show that the inclusion of an endogenous learning effect leads to different and significantly better solutions compared to those that are found when the learning effect is neglected.