Operations Research 2025
Abstract Submission

2242. Maintenance Scheduling with Conditional Generalized Temporal Constraints and Disjunctive No-Wait Constraints

Invited abstract in session TC-3: Maintenance Scheduling, stream Project Management and Scheduling.

Thursday, 11:45-13:15
Room: H5

Authors (first author is the speaker)

1. Jasmin Montalbano
Information Process Engineering, FZI Research Center for Information Technology
2. Jonas Saupe
Information Process Engineering, FZI Research Center for Technology
3. Stefan Nickel
Institute for Operations Research (IOR), Karlsruhe Institute of Technology (KIT)

Abstract

We consider a planning problem arising in the context of waterway infrastructure maintenance. Such maintenance tasks require the temporary drainage of locks and weirs, necessitating the deployment of various resources such as closure gates, watercraft and personnel. The planning objective is to find a maintenance schedule that minimizes the maximum resource capacity required at any time. This problem is known as the resource investment problem (RIP). In practice, additional constraints must be observed. First, tasks correspond to a type of maintenance work that must be carried out at a specific frequency at each building. These requirements are modeled using generalized temporal constraints where time lags vary depending on the start periods. Second, due to operational reasons, predefined groups of tasks must be executed without interruption. We refer to this problem as RIP with conditional generalized temporal constraints and disjunctive no-wait constraints (RIP-cmax-dnw). To the best of our knowledge, RIP-cmax-dnw has not been studied previously.
We formulate RIP-cmax-dnw as a mixed-integer programming (MIP) model. However, obtaining an exact solution takes prohibitively long, rendering it impractical for applications, such as scenario analyses involving changes in maintenance frequencies. Complicating factors are the large number of resources and the presence of disjunctive no-wait constraints. To address this, we propose a problem relaxation by aggregating resources based on similar capacity requirements. Furthermore, we eliminate disjunctive no-wait constraints by heuristically fixing task sequences. Experimental results show a significant improvement in solving performance, making our approach suitable for scenario analyses in real-world applications.

Keywords

Status: accepted

Back to the list of papers