Day-Ahead Bidding Strategies for Wind Farm Operators under a One-Price Balancing Scheme

TL;DR

This paper develops a stochastic optimization approach for wind farm operators to optimize day-ahead bidding strategies under a one-price balancing scheme, accounting for market impact and risk, with real data demonstrating improved profits.

Contribution

It introduces a risk-constrained stochastic optimization model for bidding strategies that explicitly considers market impact and provides an analytical solution.

Findings

Risk-constrained strategies improve expected profit.

All-or-nothing bidding negatively impacts balancing prices.

Real data shows strategies can enhance profits despite market impact.

Abstract

We study day-ahead bidding strategies for wind farm operators under a one-price balancing scheme, prevalent in European electricity markets. In this setting, the profit-maximising strategy becomes an all-or-nothing strategy, aiming to take advantage of open positions in the balancing market. However, balancing prices are difficult, if not impossible, to forecast in the day-ahead stage and large open positions can affect the balancing price by changing the direction of the system imbalance. This paper addresses day-ahead bidding as a decision-making problem under uncertainty, with the objective of maximising the expected profit while reducing the imbalance risk related to the strategy. To this end, we develop a stochastic optimisation problem with explicit constraints on the positions in the balancing market, providing risk certificates, and derive an analytical solution to this problem. Moreover, we show how the price-impact of the trading strategy on the balancing market can be included in the ex-post evaluation. Using real data from the Belgian electricity market and an offshore wind farm in the North Sea, we demonstrate that the all-or-nothing strategy negatively impacts the balancing price, resulting in long-term losses for the wind farm. Our risk-constrained strategy, however, can still significantly enhance operational profit compared to traditional point-forecast bidding.