Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
The core of this activity is to provide distribution system operators with tools for the optimal and grid-aware operation of utility-scale distributed battery energy storage systems (BESSs) in order to optimize the integration of stochastic distributed generation. The ultimate goal is the optimal control of active distribution networks with high penetration of stochastic (i.e., non-controllable) renewable-based generation. In particular, unscheduled fluctuations of the power exchanged with the upper grid level are minimised via the proposed control and scheduling framework, where we compute a dispatch plan in day-ahead using advanced forecasts of the aggregated prosumption and track it during the real-time operation using a grid-aware optimal power flow (OPF)-based control of the controllable BESS accounting for both grid constraints and BESS operational constraints. We experimentally validated the proposed control and scheduling strategy to dispatch the operation of a medium voltage active distribution network interfacing stochastic heterogeneous prosumers by using a grid-connected BESS as a controllable element coupled with a distributed monitoring infrastructure. In particular, the framework consists of two algorithmic layers. In the first one (day-ahead scheduling), an aggregated dispatch plan is determined, which is based on the day-ahead forecast of the prosumption and accounts for the operational constraints of grid and BESS state-of-energy. An adaptive data-driven scheme based on multi-variate Gaussian distribution is used to forecast the power consumption and photovoltaic generation and used as an input at the day-ahead stage. Then, the dispatch plan for the next 24 hours is computed using a scenariobased iterative AC OPF (Codistflow) algorithm, which accounts for forecasts of RESs and load profiles with 95% confidence interval with 1h time resolution. The second layer consists of real-time operation, where a grid-aware model predictive control determines the active and reactive power set-points of the BESS so that their aggregated contribution tracks the dispatch plan while obeying to BESS’s operational constraints as well as the grid’s ones. The grid constraints are modelled using the Augmented Relaxed OPF developed at the EPFL-DESL. The proposed control framework is validated by dispatching the operation of a 12kV/20MVA MV distribution network in Aigle, Switzerland (i.e. the REeL demonstrator) using a 1.5 MW/2.5 MWh BESS, which is controlled in real-time given the online grid state estimation enabled by the deployed distributed PMU-based sensing infrastructure.