Probabilistic forecasting for sizing in the capacity firming framework

TL;DR

This paper develops a probabilistic approach to size a photovoltaic and battery system within capacity firming regulations, using stochastic modeling and optimization to improve accuracy and compliance.

Contribution

It introduces a stochastic modeling framework with Gaussian copula scenarios for PV sizing under capacity firming constraints, advancing beyond deterministic methods.

Findings

Stochastic approach improves sizing accuracy over deterministic methods.

Mixed-integer quadratic programming effectively models the problem.

Case study demonstrates practical applicability with real PV data.

Abstract

This paper proposes a strategy to size a grid-connected photovoltaic plant coupled with a battery energy storage device within the \textit{capacity firming} specifications of the French Energy Regulatory Commission. In this context, the sizing problem is challenging due to the two-phase engagement control with a day-ahead nomination and an intraday control to minimize deviations from the planning. The two-phase engagement control is modeled with deterministic and stochastic approaches. The optimization problems are formulated as mixed-integer quadratic problems, using a Gaussian copula methodology to generate PV scenarios, to approximate the mixed-integer non-linear problem of the capacity firming. Then, a grid search is conducted to approximate the optimal sizing for a given selling price using both the deterministic and stochastic approaches. The case study is composed of PV production monitored on-site at the Li\`ege University (ULi\`ege), Belgium.