Skip to Main content Skip to Navigation

Integrating uncertainties in short-term operational planning

Abstract : The increasing integration of renewable energy sources has a long-lasting impact on the electrical grid, and the liberalisation of the energy sector has significantly changed its regulatory environment. In particular, the distribution network has become an area of interactions of competitive actors, while being managed by a single actor: the distribution system operator (DSO). Among the DSO’ challenges is the short-term operational planning: the selection and activation of levers to ensure the safe exploitation of the grid, taking into account the forecasts of grid users’ activities. Decisions in this context are based on the mathematical model of the Optimal Power Flow (OPF). Sources of uncertainties on these latter forecasts are growing due to the increasing number of actors on the grid. The focus of this thesis is on the integration of uncertainties on power production and consumption in the OPF, using chance-constraints. The resulting probabilistic OPF model is a non-convex non-smooth optimization problem with a Difference-of-Convex (DoC) structure. The class of DoC functions is large enough to include convex, concave, and approximations of arbitrarily precision of every continuous function, while offering strong regularity properties that one can leverage to derive a generic optimisation algorithm. A first contribution of this work is the development of an original bundle algorithm for the class of DoC constrained DoC problems. Chance-constraints are proved to be DoC, and a DoC approximation of chance-constraints is proposed before being applied to the probabilist OPF. A characterization of the first-order information of probabilist functions is presented, based on a variational study of these latter functions. This characterization highlights the variety of choices when it comes to solving chance-constrained programs. Four explicit formulations of probabilist OPFs are then proposed, and their DoC structure is proved. The algorithm’s performance, the impact of parametrisation on its behaviour and the interest of each model are numerically validated on a 33 nodes network. Besides the reasonable computing times, this methodology is particularly relevant as, differently to other works in literature, the electrical viability and validity of a solution are directly accessible.
Document type :
Complete list of metadata
Contributor : Bibliothèque Mines Paris - PSL Connect in order to contact the contributor
Submitted on : Monday, June 13, 2022 - 12:30:11 PM
Last modification on : Saturday, June 25, 2022 - 3:04:27 AM


Files produced by the author(s)


  • HAL Id : tel-03693993, version 2


Paul Javal. Integrating uncertainties in short-term operational planning. Engineering Sciences [physics]. Université Paris sciences et lettres, 2021. English. ⟨NNT : 2021UPSLM062⟩. ⟨tel-03693993v2⟩



Record views


Files downloads