Fast dynamic programming with application to storage planning

Robin Girard (1) , Vincent Barbesant (2) , Fiona Foucault (1) , Georges Kariniotakis (1)

1 PERSEE - Centre Procédés, Énergies Renouvelables, Systèmes Énergétiques
2 RTE - Réseau de Transport d'Electricité [Paris]

In this paper we propose a new class of algorithms that allows to solve a class of optimization problems including that of finding an optimal storage policy. The proposed algorithm is fast with a quadratic time or even quasi linear time in some cases. The gain in term of complexity with regard to the use of a commercial linear solver is shown empirically. A freely available package in the R statistical software has been implemented and is presented here. Possible extension to more complex framework are presented. This paper ends with an example of application for the determination of the market revenue of a storage device in a local price market as a function of location, energy capacity and power limits.

Mots clés: Dynamic programming; local price market; wind power forecast; optimization; storage; planing; Index Terms-Dynamic programming; Smart-grids

Domaines: Sciences de l'ingénieur [physics]; Energétique

Robin Girard, Vincent Barbesant, Fiona Foucault, Georges Kariniotakis. Fast dynamic programming with application to storage planning. 2014 IEEE PES T&D Conference and Exposition, IEEE, Apr 2014, Chicago, IL, United States. ⟨10.1109/TDC.2014.6863551⟩. ⟨hal-01110689⟩