(701b) A Multi-Time-Scale Approach for the Integrated Maintenance and Production Scheduling of Multipurpose Process Plants

Abstract

In this work we go beyond the traditional concept of maintenance, which focuses on performing disruptive activities based on short-term needs. In maintenance scheduling it is useful to consider all technical, administrative and supervision actions that intend to retain the productivity. This includes for instance cleaning operations, asset inspection, component lubrication or part-substitutions involving only a partial plant shutdown. 

In the scheduling literature, the interaction of maintenance and production scheduling decision processes has already been widely explored. Dedopoulos at al. (1995) considers enforcing a set of maintenance tasks to be performed within the scheduling horizon through penalties. Hazaras et al. (2012) accounts for different types of maintenance: fixed time maintenance intervals as well as a set of maintenance with flexible start time to be determined by the optimization. Castro et al. (2014) discussed the optimal maintenance scheduling problem for a gas engine power plant where the production is driven by electricity sales assuming seasonal variations in electricity pricing. Biondi et al. (2015) presents an MILP model for short-term scheduling of production and maintenance in multipurpose process plants allowing to effectively “close the loop” between the two decision processes. 

Here we take a step forward assuming that the dynamics of the degradation process leading to a maintenance need is not always fast enough to be captured within the production scheduling horizon. This is supported by the fact that maintenance tasks in real industrial processes are often less frequent. Nonetheless, it is crucial to take them into account while generating a production schedule as their impact on plant operation can be disruptive. In order to accomplish this objective, we introduce a multi-time-scale approach capable of dealing with the short and medium-term decisions needed to satisfy the production requirements, as well as, with the medium-long term ones dealing with maintenance. 

The scheduling horizon is divided into two parts; short term and long term using a discrete time-representation with different granularity (e.g. one hour and one day) to solve an integrated optimization problem. Both the short-term and the long-term models are based on the STN process representation by Shah et al. (1993).

For the short-term model, we considered the model in Biondi et al. (2015) with an enhanced formulation dealing with the asset residual life update. Each asset of the plant is assumed to be able to work in different operation modes impacting the production performance (e.g. speed, throughput…) and degradation speed. Furthermore, each asset is supposed to have a predefined residual lifetime that can be restored through maintenance tasks.

For the medium-long term model, the aggregated STN representation by Maravelias et al. (2004) is extended to account for the residual life update and maintenance requirements. Instead of the proposed single relaxed MILP model for the whole planning horizon, an aggregated model is written for each time slot, allowing not only to get a solution that is closer to feasibility, but also to optimally plan maintenance requirement whenever needed. 

The proposed approach is tested on different STN process representations and is able to deal with small and medium problem sizes. Maintenance tasks are planned on plant assets whenever needed, trying to influence the production as little as possible. Similarly, the production is scheduled on the plant in order to satisfy the production requests and to reduce the total asset degradation and maintenance costs.

Acknowledgements

The Marie Curie FP7-ITN research project "ENERGY-SMARTOPS", Contract No: PITN-GA-2010-264940 is acknowledged for financial support.

References

Biondi, M., Sand, G. , Harjunkoski, I. (2015), “Tighter integration of maintenance and production in short-term scheduling of multipurpose process plants”, Computer Aided Chemical Engineering, Vol 37, 1937-1942

Castro, P. M., Grossmann, I. E., Veldhuizen, P. and Esplin, D. (2014), “Optimal maintenance scheduling of a gas engine power plant using generalized disjunctive programming”. AIChE J., Vol 60, 2083–2097

Dedopoulos, I. T., Shah, N. (1995), “Optimal Short-Term Scheduling of Maintenance and Production for Multipurpose Plants”, Industrial Engineering and Chemistry Research, 34 (1),192–201

Hazaras, M. J., Swartz, C. L. E., Marlin, T.E. (2012), “Flexible maintenance within a continuous-time state-task network framework”, Computers & Chemical Engineering, Vol 46, 167–177

Maravelias, C. T., Grossmann, I. E. (2004), “A hybrid MILP/CP decomposition approach for the continuous time scheduling of multipurpose batch plants”, Computers & Chemical Engineering, Volume 28, 1921–1949

Shah, N. , Pantelides, C. C., Sargent, R.W.H. (1993), “A general algorithm for short-term scheduling of batch operations—II. Computational issues”, Computers & Chemical Engineering, Volume 17, Issue 2, 229–244