Distribution network optimization: predicting computation times to design scenario analysis for network operators

Abstract

The increased use of renewable energies promotes decarbonization and raises the load on power distribution networks, forcing responsible distribution network operators to re-evaluate and re-design their networks. Infrastructure planners employ a rolling-horizon planning procedure with frequent recalculations to face informational uncertainty, which require solving multiple scenarios. Keeping complexity manageable is particularly challenging as distribution network areas may span multiple cities and counties. In this study, we focus on infrastructural decomposition, where the distribution network is decomposed into multiple parts and planning problems, which are then optimized separately. However, infrastructure planners lack the knowledge of how they should design a scenario analysis for a subnetwork to account for informational uncertainties subject to limited planning time and computing resources. Based on empirical requirements from literature and discussions with experts, we present a novel mixed integer linear optimization model that allows to use exact solution approaches for realistic large-scale distribution networks. Our approach considers the primary and secondary distribution network in an integrated way and designs a flexible topology for high reliability power distribution. We perform extensive computational experiments and a sensitivity analysis to determine correlations between the values of model parameters and computation times required to solve the resulting model instances to optimality. The results of the sensitivity analysis indicate that the combination of the number of buses, lines and the considered action scope have a considerable influence on the solving time. In contrast, a higher number of available transformers led to a better solvability of the model. From these computational insights, we derive implications for infrastructure planners who wish to perform scenario analysis for planning their power distribution networks.