Unified unit commitment formulation and fast multi-service LP model for flexibility evaluation in sustainable power systems

Classical unit commitment (UC) algorithms may be extremely time-consuming when applied to large systems and for long term simulations (for instance, a year) and may not consider all the features required for flexibility assessment, including analysis of different reserve types. In this light, this paper presents a novel flexibility-oriented unified formulation of a large-scale scheduling model considering multiple types of plants (including storage) and reserves, which can seamlessly model binary (BUC), mixed integer linear programming (MILP), and relaxed linear programming (LP) UC. Comparisons are carried out on several case studies for a reduced model of Great Britain, assessing loss of accuracy (as measured according to various metrics specifically introduced) against computational benefits in different renewables scenarios with more or less flexible systems. It is demonstrated how the computational time of the LP model is significantly less than the BUC and MILP approaches while capturing with relatively high precision all the relevant flexibility requirements and allocation of multiple types of reserves to different types of plants. The results indicate that the proposed fast LP model could be suitable for various computationally intensive flexibility studies (e.g., Monte Carlo simulations or planning), with significant reduction in simulation time and only minor errors relative to established MILP models.