Turnaround Scheduling in Chemical Manufacturing

In chemical manufacturing and petroleum refining, large-scale maintenance activities must be conducted on a regular basis. Entire production units are shut down for disassembly and comprehensive inspection and renewal. Such a process is called Shutdown and Turnaround (or turnaround for short). It is an essential process but causes high out-of-service cost. Therefore a good schedule for the turnaround is of high priority to the manufacturer. In a turnaround a huge number of interdependent operations must be executed by maintenance groups of different specializations (different resources). Scheduling these is already a complex task if working shifts must be respected. However, in this particular problem another issue increases the complexity drastically: the duration of an operation is flexible in the sense that it is determined by the number of resources that is assigned to it. Technical reasons restrict the choice to a range between a maximum and minimum duration. Clearly, due to communication delays, the duration of an operation does not decrease with the same rate as the number of assigned resources increases but we can assume a non-increasing discrete function. Each resource unit causes a certain cost for each time unit it is used. It can be assumed that the number of available resources is unbounded because a presumably arbitrary number of external resources can be acquired. However, for some resource types resource units cannot be obtained for arbitrary small time periods but must be kept and paid for a certain minimal period. That means that the resource requirement of each resource type (or a subset) must be leveled over at least such a time period – in a simplification this is the entire turnaround period. A feasible schedule consist of a feasible temporal assignment of resource units to operations and a feasible sequencing for all of them. Now, the goal is to find feasible schedules of minimum length for different budgets. In fact, we have a generalized version of the Discrete Time-Cost Tradeoff Problem. In case that the out-of-service cost for taking the production unit off-stream are given per time unit, the objective changes to finding a schedule that minimizes the total cost, i.e. the sum of out-of-service cost and the cost for acquiring resources over the turnaround period.