As the global energy landscape undergoes a transformative shift towards sustainability, the integration of different energy carriers into multi-energy systems becomes pivotal to increase system flexibility and effectively integrate renewable energy sources. This is essential to meet energy demands that have traditionally relied on fossil fuels.

However, multi-energy systems should be wisely optimised to maximise their potential, strategically buying and selling electricity on markets and ideally managing storage while meeting local energy demand. Commonly used optimisation algorithms are based on linear programming; more accurate approaches use quadratic, conic or non-linear programming to capture the complexity of the objective functions, but all require the input variables to be deterministic.

Nevertheless, these algorithms typically run a day ahead of real-time to define optimal market strategies, so only forecasts of input variables are available, but these are subject to uncertainty. Although neglected, these uncertainties can have a significant impact. A different than expected electricity price may result in the scheduled dispatch of energy resources not being the cheapest possible.

A deviation of the constrained local energy demand from the forecasted value requires an adjustment in the intraday market, which could be reflected in additional costs, making the adopted solution suboptimal. Therefore, this paper proposes a stochastic approach for robust optimisation of electricity market bidding strategies, specifically addressing uncertainties in electricity prices and local heat demand, which affect the problem objective functions and constraints, respectively.

The methodology involves running an appropriate number of deterministic optimisations, solving each unit commitment problem deterministically, and transposing it into a set of bids submitted to the electricity market. Subsequently, market clearing is simulated in each scenario, determining the accepted and rejected bids, as well as the actual cost of each deterministically optimal solution.

Finally, the robustness of each solution is assessed using various statistical metrics, and the best bidding strategy is defined, taking into account both the expected actual cost and its uncertainty. The proposed methodology is applied to the district heating network in East Milan, where different conversion technologies such as power-to-heat, gas-to-heat, cogeneration, thermal storage and solar collectors are installed.

The results highlight the importance of robust optimisation in ensuring the resilience of energy systems, especially in the context of dynamic markets and evolving energy landscapes.