Corrective receding horizon scheduling of flexible distributed multi-energy microgrids

Highlights: • Defining optimal configuration between centralized and distributed MEM configurations. • Flexibility of MEM increases with consideration of additional energy vectors. • Annual operational costs are independent of the model approximations (e.g. MEM unit efficiency modelling). • Significant errors occur if approximations are used for MEM daily operation analyses. - Abstract: The goal of the paper is to provide a comprehensive operational flexibility evaluation of different Multi-energy Microgrid (MEM) options. This is done by incorporating Mixed Integer Liner Programming (MILP) model for annual simulations and expanding it with Receding Horizon Model Predictive Control (RH-MPC) algorithm for short term daily operational analyses. The model optimizes flows of various energy vectors: heat, fossil fuels (natural gas), cooling and electricity, coordinating different microgrid elements with the goal of serving final consumer needs and actively participating in energy markets. The second novelty of the work is in the approach to multi-energy operational flexibility assessment, capturing different technologies, MEM configurations and different modelling concepts. When MEM is connected to the upstream power system its flexibility manifests as capability to alleviate variability and uncertainty in local production of RES and demand. On the other hand, when operating isolated from the rest of the system, the main flexibility indicator is minimum waste of energy while ensuring the satisfaction of all demand needs (electrical and heating/cooling). Following on this, multiple MEM configurations have been analyzed, showing different levels of available flexibility and capability to follow scheduled day-ahead exchange with the rest of the system, but also different amounts of wasted/curtailed energy in off-grid mode. Additionally, detailed analyses are performed concerning algorithm approximations which are often introduced in MEM modelling, such as efficiency of generation units. While these approximations have smaller impact on annual operational flexibility assessment (the difference is around 2–5% in terms of total cost), the result clearly show their significant impact on daily operational flexibility estimates.