Evaluate fitted-distribution reformulations for the joint chance-constrained EV aggregator bidding model

Determine how satisfactory the bidding outcomes are when the joint chance-constrained optimization model for an electric vehicle aggregator’s FCR-D up/down bidding (enforcing Energinet’s P90 and LER requirements) is reformulated analytically under a fitted parametric probability distribution for available upwards, downwards, and energy flexibility, instead of using empirical sample-based approximations of the chance constraints.

Background

The paper develops a joint chance-constrained optimization model for an EV aggregator bidding into Nordic FCR-D up and down markets, incorporating Energinet’s P90 and LER requirements. To avoid assuming a specific probability distribution for the EV portfolio’s available flexibility, the authors employ sample-based techniques (ALSO-X and CVaR) to reformulate and solve the chance constraints.

The authors note that chance-constrained programs may admit analytical reformulations when the underlying probability distribution possesses certain properties. They explicitly leave for future work an investigation of how satisfactory the bidding results would be if one fits a parametric distribution and uses analytical reformulations rather than empirical sample-based approaches.