A continuation approach to the optimization of hydropower operations

Jorn H. Baayen, Julia Rauw, Teresa Piovesan

Published 2018-05-02Version 1

The instantaneous power generation from a hydroelectric turbine is proportional to the product of head difference and turbine flow. The equation relating power to hydraulic variables is therefore nonlinear. Hence, optimization problems subject to this relation, such as release schedule optimization, are nonconvex and may admit multiple local isolated minima. This renders such problems problematic for use in operational model predictive control. This paper shows that release schedule optimization problems subject to the nonlinear turbine generation equation may be set up using a continuation approach to be both zero-convex and path stable. In this way such optimization problems become suitable for decision support systems based on model predictive control. An example problem is studied, and it is shown that significant productivity gains may be realized using the presented methodology.