The OGRE project stands as an umbrella project that welcomes companion PGMO projects related to centralized versus decentralized management of energy systems. The OGRE project is a PGMO-IROE Project funded by Programme Gaspard Monge pour l'Optimisation et la recherche opérationnelle (PGMO), Électricité de France (EDF) et Fondation Mathématique Jacques Hadamard (FMJH) for the years 2016 and 2017.
The transformation of energy systems is accelerating. Local initiatives are blossoming, with the drop in renewable energy costs and the impulse of decentralized actors (individuals, collectivities). Managing an energy system with myriads of decentralized sources (wind, sun) and actors is becoming more and more challenging.
The PGMO has launched several projects which, directly or indirectly, touch the subject of the role of optimization and game theory, in the new energy landscape. The 2016 PGMO call for projects has attracted new proposals that touch related issues: bi-level optimization, smart-grids, etc. When relevant, the sponsor EDF encourages projects to work together, to give more impact to the PGMO outputs.
This is why we propose OGRE -- Optimization, Games and Renewable Energy -- as an umbrella project that welcomes companion projects (ongoing and new), related to centralized versus decentralized management of energy systems. Each companion project preserves its financial and scientific autonomy.
The OGRE project will coordinate the research programs of the companion projects in two ways:
Summaries of the following companion projects can be found in §A.
Participants: Laetitia Andrieu (EDF R& D), Sebastien Lepaul (EDF R& D), Olivier Beaude (EDF R& D), Nadia Oudjane (EDF R& D), Yezekael Hayel (Université d'Avignon), Didier Aussel (Université de Perpignan), Luce Brotcorne (Inria), Pierre Carpentier (Ensta), Jean-Philippe Chancelier (ENPC), Vincent Leclère (ENPC), Michel De Lara (ENPC)
Roundtable followed by discussion on organization. Summary of decisions:
The following speakers have agreed to give a talk under an OGRE stream.
OGRE invited session program
Location: EDF Lab, Saclay How to get there
Location: Room Opale 1AB01, EDF Lab, Saclay How to get there
Host OGRE session inside the SESO 2017 International Thematic Week ``Smart Energy and Stochastic Optimization''
Location: ENSTA ParisTech, Palaiseau, France
How to get there
see SESO 2017 program
The OGRE report is made of three main parts.
We expect that the OGRE researchers will build, in common, a panoply of models and problems that are destined to be common instances for the methods developed by each companion project.
The world's energy landscape is changing fast. Three key drivers are remolding power systems: renewable energies penetration (intermittent and highly variable), expansion of markets and of new players, deployment of telecommunication technology and smart meters. These changes put to the front stochastic and decentralized optimization as an adapted formalism. Even if methods, algorithms and softwares have been developed for a long time, they are less common in practical applications than their deterministic counterparts. The LORI project - Logiciels pour l'Optimisation des Réseaux Intelligents - aims at making academics and companies closer in working together on the development of stochastic and decentralized optimization methods, algorithms and softwares (dedicated toolboxes, including modelers and solvers).
The operation of power production and trading is challenged by renewable energies penetration, which stresses physical and market risk factors. Specific approaches and tools are required. The PALON project aims at initiating collaboration between researchers, in Paris (PA) and London (LON), who represent a unique combination of mathematical, computational and financial expertise. They have identified common interest and complementary skills in stochastic modeling and optimization of renewable energy systems within modern electricity markets. Thanks to two workshops, the PALON project will clarify which mathematical models and computational tools are to be developed for future long term research.
The classical energy management optimization framework considers that a central planner decides how to efficiently dispatch energy to consumers, in a centralized fashion. Due to the recent growth of alternative and renewable energy sources, new relevant actors have appeared in the context of smart grids. Such new actors (for generation, storage, demand response) challenge the classical generation model in which a central planner is the sole decision maker. As a consequence, we propose to explore new models and methodologies in order to achieve optimal operation of emerging decentralized energy systems.
The project aims at studying how a risk-averse public decision maker can design a market in a way such that the competitive equilibrium minimizes the utility of the decision maker. More precisely, a recent paper by M.Ferris, A.Philpott and R.Wets has shown that a competitive equilibrium of risk-averse agents in a certain market is equivalent to a global risk-averse minimization problem with specific risk measure. We propose to take the reverse road. We consider a risk averse global problem with given risk-measure and decompose the problem in subproblems through Lagrangian price-decomposition. We want to deduce from the decomposition the adequate market design.
The aim of the project is to solve a joint pricing and design problem of energy services in a competitive environment for residential demand side management. More precisely, the objective is twofold: generate revenue for an energy provider and encourage the customers to individually and voluntarily reduce their consumption at peak periods. A bilevel approach is considered to take explicitly into account the strategic behaviour of consumers into the optimization process. The robustess of the bilevel problem with respect to the demand will be studied. Solution algorithms will be developped to solve a mixed integer bilinear bilinear bilevel and an Equilibrium problem with equilibrium constraints
The aim of this project is to realize a deep analysis of the equilibrium offer/demand in a decentralized context of power systems. Indeed, the emergence on the power system of local actors and renewable energy lead to new framework of decentralized optimization problems. By considering three different aspects of the interactions between local actor/aggregators and main producers/sellers, we intend to evaluate the influence of such new exchanges processes on the steering/flexibility of the demand and the market prices. This analysis will be based on multi-leader-follower game theory.
We aim at jointly studying two problems which are usually independent: transportation and electrical networks. In the context of smart cities, the question addressed is the following: could both problems become inter-dependent? On the one hand, the traffic assignment problem in the transportation network with Electric Vehicles (EV) should consider both (road) congestion effect and energy need for the travel. On the other hand, in the electrical network, electricity prices depend on the demand, that typically itself depends on the aggregate energy need of EVs. It is now clear that energy consumed by driving is directly linked with travel time, and thus with congestion. In turn, both problems are naturally coupled. Both problems can be modeled with bi-level optimization: at the top level is the transportation / electrical network operator -- deciding respectively road toll or electricity pricing - and at the bottom level the EVs -- choosing a route and a charging profile. This collaborative project brings methodologies and expertise of academic researchers for coupling two bi-level problems and also of EDF researcher-engineers for studying the practical interest of coupling driving and charging decisions (smart cities).
With the smart grid revolution, house energy consumption will play a significant role in the energy system. Home users are indeed responsible for a significant portion of the world's energy needs portion, but are totally inelastic with respect to the market (i.e. the energy demand does not follow the price of the energy itself). Thus, the whole energy generation and distribution system performance can be improved by optimizing the house energy management. Those problems are concerned by multiple objectives such as cost and users' comfort, and multiple decision makers such as end-users and energy operators. We propose a home automation system that can monitor appliance scheduling in order to simultaneously optimize the total energy cost and the customer satisfaction.
Nous vivons aujourd'hui une époque cruciale de l'évolution du secteur électrique : l'émergence des énergies renouvelables se combine à l'arrivée de composants dits "intelligents" transformant le réseau électrique en "smart grid". À terme, celle-ci sera constituée d'un ensemble de sous-réseaux ou "microgrids" pouvant se connecter ou se déconnecter en temps réel du réseau principal. L'objectif de chaque "microgrid" sera alors de tendre vers l'autonomie en satisfaisant la demande énergétique de ses consommateurs à l'aide de sites de production qui lui seront propres.
Dans ce projet, nous nous intéressons aux problèmes d'optimisation combinatoire liés au découpage futur de la "smart grid" en "microgrids". En particulier, nous souhaitons étudier l'impact du concept d'autonomie sur les problèmes de partitionnement et d'augmentation de graphes.