![/system/images/000/000/219/Logo_Journ_es_Optimisation__2012_Outl_default.png](/system/images/000/000/219/Logo_Journ_es_Optimisation__2012_Outl_default.png)
Including an Industrial Optimization Day
HEC Montréal, May 7 - 9, 2012
JOPT2012
HEC Montréal, 7 — 9 May 2012
![](/assets/cal-add-6d138cf3e30399796b32f524ba2175862f0503314e61a73e25c8269466c9e3d9.png)
WB5 Optimisation de la production d’électricité 2 / Electricity Generation Planning 2
May 9, 2012 11:00 AM – 12:15 PM
Location: Raymond Chabot Grant Thornton
Chaired by Michel Gendreau
2 Presentations
-
11:00 AM - 11:25 AM
Multistage Stochastic Optimization Methods Applied to Solve the Medium-Term Operation Planning Problem
The purpose of this work is to present a comparative study about the performance of different multistage stochastic optimization methods applied to the Medium Term Operation Planning problem: Nested Decomposition, a common approach for solving these kinds of problems, and the Progressive Hedging method, particularly promising to solve multistage stochastic.
-
11:25 AM - 11:50 AM
Midterm Hydro Generation Scheduling under Inflow Uncertainty Using the Progressive Hedging Algorithm
We propose a new stochastic optimization model to solve Hydro-Québec's midterm generation scheduling problem (MGSP). The aim is to establish weekly generation targets for controllable hydro plants to maximize reservoir energy storage at the end of a 18-24 months planning horizon. Reservoir inflow variability is modeled using a finite scenario tree. Variablehead hydro plants generation functions are modeled as concave piecewise linear functions of reservoir storage and turbined outflow. The MGSP is formulated as a huge multistage stochastic linear program. A Lagrangean relaxation is applied on non-anticipativity constraints of the stochastic program. A scenario decomposition approach is used to solve efficiently the stochastic program. We apply the well-know progressive hedging algorithm. This optimization model is tested on Hydro-Québec large-scale hydro-dominated power system.