Adequacy aware long-term energy-system optimization models considering stochastic peak demand

An important aspect of long-term power system planning models is their adequacy awareness, i.e., their ability to ensure generation adequacy in the final solution, which in turn strongly depends on the level of temporal detail included in the model structure. To maintain computational tractability, the temporal detail included in long-term planning models is, however, often limited, which decreases their adequacy awareness resulting in inadequate capacity mixes. To compensate for this, several adequacy-improving measures can be taken. The aim of this paper is to investigate the performance of three of these measures, namely (i) adding a traditional static planning reserve margin (PRM) constraint, (ii) adding a dynamic PRM constraint and (iii) a novel approach that increases the level of temporal detail with which critical peak periods are modeled. To this end, we compare each of these three adequacy methods in a long-term planning exercise based on (i) total system costs, (ii) adequacy of the resulting capacity mix and (iii) technology choices. Our results suggest that the novel approach more effectively increases the adequacy of the obtained capacity mix, resulting in lower total system costs and less technology biases. Also the dynamic PRM approach performs well, although occasional inadequacies can occur in individual milestone years due to exogenously imposed capacity credits.