Stochastic Optimization Course for Metropolitan Electricity Authority (MEA)
#Metadata
Instructors
- Parin Chaipunya, KMUTT, Bangkok (parin.cha[at]kmutt.ac.th)
- Michel De Lara, École Nationale des Ponts et Chaussées, Paris (michel.delara[at]enpc.fr)
Course objectives
- Capacity building. Prepare the practitioners for the current and new trends on energy management that involves more and more unpredictable elements — demand, availability of renewable resources, etc.
Course organization
- A lecture in the morning of each day.
- In the afternoon, the attendees should prepare the problems from the MEA side to discuss and hopefully to frame them using stochastic optimization.
#Program
Week 1 (Preparation)
Lecturer: Parin Chaipunya
Monday 17 November 2025
Morning (08h30–11h30)
Introduction to optimization. Linear and quadratic optimization. Convex optimization.
Afternoon (12h30–15h30)
Modeling session
Thursday 20 November 2025
Morning (08h30–11h30)
Mathematics of uncertainty. Probability and random variables. A simple stochastic optimization.
Afternoon (12h30–15h30)
Modeling session
Week 2 (Stochastic optimization. Part I)
Lecturers: Michel De Lara, Parin Chaipunya
Monday 24 November 2025
Morning (08h30–11h30)
One-stage stochastic optimization. First examples.
Overview
This lecture covers the question of how to commit to produce energy for an uncertain demand.
Afternoon (12h30–15h30)
Modeling session
Thursday 27 November 2025
Morning (08h30–11h30)
Two-stage stochastic optimization, scenario decomposition, Progressive Hedging.
Overview
How do you commit yourself to produce energy for an uncertain demand and when you can correct your first decision by a second decision mobilizing costly energy after the demand has been revealed. How do you handle demand and production scenarios ?
Afternoon (12h30–15h30)
Modeling session
Week 3 (Stochastic optimization. Part II)
Lecturers: Michel De Lara, Parin Chaipunya
Monday 15 December 2025
Morning (08h30–11h30)
Introduction to multi-stage stochastic optimization, Bellman equation
Overview
Answers the question of how to manage a storage (dam, battery) when you make sequential decisions (charge/discharge sequence) on a given timespan, under unpredictable demand and renewable sources production.
Afternoon (12h30–15h30)
Modeling session
Thursday 18 December 2025
Morning (08h30–11h30)
More advanced methods. Decomposition of large-scale stochastic optimization problems. Risk measures and optimization.
Overview
How to both maximize profit and minimize risk. Example of smart grid management with multiple storages located on a network. How to avoid high costs with low probability by using suitable risk measures.
Afternoon (12h30–15h30)
Modeling session