Abstract : Quelques utilisations de la Recherche Opérationnelle pour plus d'énergie renouvelable dans le mix électrique L'arrivée et le développement des énergies renouvelables pour la production d'électricité pose de nouveaux problèmes concernant non seulement la production d'électricité mais aussi son injection dans le réseau mais aussi sa valorisation, sur les marchés et aux yeux des consommateurs finaux. Le groupe Sun'R a développé plusieurs solutions intégrant des méthodes d'optimisation et de recherche opérationnelle pour y répondre. Les applications dans le cadre de la gestion optimisée du stockage, de l'agrivoltaïsme et de la fourniture seront présentées.

Abstract : Artelys a développé un outil web de simulation de réseaux électriques pour Tennet, le gestionnaire du réseau public de transport d’électricité des Pays-Bas et d’une partie de l’Allemagne. Ce logiciel est un outil pédagogique qui propose des « serious games » permettant d’illustrer certaines problématiques liées aux évolutions récentes du système électrique, en particulier l’intégration des moyens de production à base d’énergies renouvelables. Artelys présentera comment les modèles d’optimisation peuvent être utilisés pour expliquer l’impact du design des marché de l’électricité sur les congestions, les écrêtements d’énergies renouvelables et les besoins de réajustement. Une ouverture sera faite sur les techniques d’optimisation avancées mises en œuvre dans le cadre des études Metis pour le compte de la Commission Européenne et l’appel à des moyens de calcul distribués massivement.

Abstract : We consider an uncapacitated multi-echelon lot-sizing problem within a remanufacturing system involving three production echelons: disassembly of used products brought back by customers, refurbishing of the recovered parts and reassembly into like-new finished products. We aim at optimizing the production planning for the corresponding three-echelon system over a multi-period horizon. Production planning involves making decisions about the production level (i.e. which products and how much of them should be made), the timing (i.e. when the products should be made) and the resources to be used. Within a remanufacturing context, production planning includes making decisions on the used products returned by customers, such as how much and when used products should be disassembled, refurbished or reassembled in order to build new or like-new products. The main objective is to meet customers’ demand for the remanufactured products in the most cost-effective way. We consider a stochastic environment, in which the input data of the optimization problem are subject to uncertainty. We propose a multi-stage stochastic integer programming approach relying on scenario trees to represent the uncertain information structure and develop a stochastic dual dynamic integer programming algorithm to solve the problem on large-size scenario trees. More specifically, we investigate a stochastic dynamic programming formulation of the problem based on continuous state variables, which allows us to decompose the problem into a series of single-node subproblems. We then reformulate the obtained sub-problems by using a binary approximation of the continuous state variables, allowing us to use the SDDiP algorithm proposed in [2] to solve the problem. We also study an approximate version of the SDDiP algorithm in which a cutting-plane generation phase based on continuous state variables is carried out to build an approximation of the expected cost-to-go functions. The proposed stochastic dynamic programming formulation provides a starting point for the application of such decomposition methods to production planning problems. Additionally, our numerical results show that the proposed method is capable to obtain near-optimal solutions in practicable computation times for large-size instances, showing its applicability to real-world optimization problems.

Abstract : With the fight against climate change and the need to reduce our dependence on polluting energy sources, the French government has put legislation in place to promote the adoption of distributed renewable energy production. An example of such legislation is the one on collective self-consumption enacted on the 26th of July 2016. A collective self-consumption agreement is one between different consumers and producers who decide to collectively consume the locally produced renewable energy. In the proposed model, we consider a community which has multiple decision makers with conflicting objectives: an aggregator, who decides how the produced energy is distributed among the agreement’s members, and consumers who minimize their energy bills in response. This generates a hierarchical decision problem, modeled as a bilevel optimization program in which the aggregator is the leader and the agreement’s members are the followers. The obtained bilevel problem is solved by using a single-level reduction by expressing the lower level’s optimality with KKT conditions. We explore two kinds of formulations. The first one gives a Mathematical Program with Complementarity Constraints (MPCC), and the second one gives a Non Convex Quadratically Constrained Program (QCP). Solving linear approximations for the obtained problems, we get upper and lower bounds for the optimal value of the original bilevel program. Results provide a practical tool to analyse clients possible behavior in a collective self- consumption agreement.