Quantifying resilient urban energy systems: Statistical analysis of climate adaptation, renewable integration, and socioeconomic dynamics

Abstract

Multi-energy systems (MES) become increasingly important to accommodate renewable energy (RE) since fossil fuels diminish and RE is more prevalent. Power systems are challenged by high penetrations of RE, including frequent wind curtailments and uncertain RE resources threatening their security. A 2-step scheme for MES is proposed in the study to generate reliable operational decisions by incorporating integrated heat-electricity demand response (DR). In the initial phase, dynamic programming was applied to determine the optimal solution for previous linkage sessions. Subsequently, machine learning techniques were trained using historical data and the resulting optimal decisions to make optimal real-time decisions without prior knowledge of future power costs or vehicle usage. An accurately trained deep neural network is capable of significantly reducing charging costs, usually approaching the retrospectively calculated optimum charging price. In the suggested scheme, the operating prices are minimized under a base scenario ensuring that no constraint is violated regardless of the scenario. RE curtailment can be reduced significantly with power-to-gas (P2G) devices that convert excess energy from wind and photovoltaic sources into natural gas. In addition, this paper proposes integrated heat-electric DR as a means of reducing operating prices and enhancing security against uncertainties by employing couplings between different systems. The outcomes show that robust optimization enhances the security of systems, and the unified electric-heat DR can boost the use of RE while improving the economic benefit. It should be noted that this paper investigated under the impact of the climate changes and also socioeconomic dynamics that can change consuming energy in urban systems.