Knowing in advance the amount of resource needed to carry out the jobs of a project is a major concern for decision makers as it allows for anticipating logistical requirements. Adjusting this resource amount based on initially planned project data is often not sufficient since this data is subject to uncertainty in most real-world applications. On the one hand, mobilizing supplementary resources at the last minute can be very expensive. On the other hand, overestimating the requirements can result in a costly waste of resource. In this context, the question of taking uncertainty into account when anticipating resource requirements arises naturally.

The problem of computing a robust resource profile, called workload, is considered in a setting where independent jobs can be scheduled preemptively on some intervals and are subject to budgeted processing time uncertainty. A solution approach through column-and-constraint generation is proposed as well as a compact MIP formulation. Both methods were tested numerically.