Stochastic optimization of a hydroelectric production system for the aluminium industry

Rio Tinto operates aluminium plants in Saguenay that are powered by their hydroelectric system. An efficient management of water available in the system is primordial to ensure energy supply to the aluminium smelters. This quantity is uncertain since the exact inflows in the reservoirs are unknown when decisions are taken. Stochastic optimization is used to make decisions under uncertainty. Mid-term optimization models determine reservoir volumes while short-term models dispatch the available water as efficiently as possible between the power plants and turbines in the system. The goal of this project is to develop operational optimization models for Rio Tinto by coupling the mid and short-term models and by developing a method to define a calendar of unit outages. The optimization models must consider energy demand, hydrologic forecast uncertainty as well as uncertainty linked to the outage of smelters. Their use leads to using water efficiently and operational costs reduction.

Faculty Supervisor:

Amaury Tilmant

Student:

Sara Séguin

Partner:

Rio Tinto Alcan

Discipline:

Engineering - civil

Sector:

Energy

University:

Université Laval