DC 4 – Influence of multiple layers of uncertainty dynamics in large scale stochastic energy planning

Project Title: Influence of multiple layers of uncertainty dynamics in large scale stochastic energy planning
Doctoral Candidate: N.N.
Host Institution: Electricité de France
Supervisors: Wim van Ackooij, Welington de Oliveira

Objectives: In energy planning, one typically has to handle several temporal dimensions at once. This is already the case classically, when dealing with seasonal storage, relying in turn on operational management. Frequently simplifications allow one to focus on some of the aspects of the problem, while neglecting or greatly simplifying others and thus come to computational results. However, in order to properly appreciate the flexibility offered by some means of generation, e.g., hydro, it can be argued that the added value is only correctly evaluated in the measure at which inflexibility is represented in the system. The latter would result from a) operational constraint, b) un certainty. Incorporating either offers significant challenges – mathematical, algorithmic and computational. The open-source platform SMS++ offers many possibilities for setting up extremely detailed versions of energy management tools and also provides a set of possible algorithms – the whole mostly focussing on part a) – representing more realistic models. The second part b) is as of yet still somewhat simplified. First of all, it is clear that uncertainty may have different dynamics between shorter and longer term time spans. These dynamics may influence in a different way the system. As a matter of fact, the short-term dynamics could be thought of as embedded in the larger ones and would need to be consistent (think of short-term weather-related dynamics influencing demand, wind power generation,… within a larger scale yearly dynamic responsible for seasonal inflows). The DC will suggest new formulations of stochastic programs accounting for such multiple layers of uncertainty, investigate theoretical properties of the thus formulated models, suggest algorithms for resolution and ideally experiment within the SMS++ framework.

Expected Results: New insights as to how multiple layers of uncertainty may influence the need for / value of flexibility

Planned secondment: The DC will be at PSL (W. De Olivera) during the time the courses (s)he has chosen are offered, which is expected to mostly happen during the first year. The DC will also spend a part of the initial period at UNIPI (A. Frangioni) for courses and to get acquainted with the SMS++ framework, and at EDF premises to gain a good understanding of the energy applications. The DC will then divide the time among the three institutions in accordance with the needs of the research as they arise, in compliance with the rule of a maximum of 12 months of secondment.

Degree awarding institution: Mines Paris — Université Paris Sciences et Lettres (PSL)