Optimizing Industrial Scheduling under Energy Volatility

Details

Supervisors

Creemers, Stefan

Faculty

Louvain School of Management

Degree label

Master [60] en sciences de gestion (horaire décalé)

Abstract

enIn a single day on the Belgian electricity market, prices can swing from negative around midday to above 300 €/MWh by evening. For an industrial site using several megawatts, the bill turns into something to watch every day. Most production scheduling tools were built before this volatility existed; they look at duration and resources, while the energy cost arrives later as a result of choices that ignored it. This thesis adds an energy layer to the classical Resource-Constrained Project Scheduling Problem (RCPSP). Hourly prices, machine power profiles, and a peak penalty matching the Belgian capacity tariffs are now part of the objective. Four solvers (Constraint Programming, GRASP, and two variants of LNS) are tested on PSPLIB benchmark instances of up to 90 tasks, with a 30-second budget per run. The best solver reduces the electricity bill by 60 to 70% compared to a schedule that minimises duration only. Applied to a Belgian cement plant, the same idea projects to a conservative 15% saving on the annual bill, with a payback of a few months. The trade-off between duration and bill is shown on a Pareto curve. The managerial half of the thesis builds a screening index (EASI) for the sectors most likely to benefit, and explicit ROI numbers for the deployment case. The author is a civil engineer working as energy manager in Belgian industry and completing a master in management in the evening. The main message: price-aware scheduling is ready for industrial use today. A large share of demand response can be captured simply by using the prices published on the market.