Siting renewable power generation assets with combinatorial optimisation

This paper studies the problem of siting renewable power generation assets using large amounts climatological data while accounting for their spatiotemporal complementarity. The is cast as a combinatorial optimisation selecting pre-specified number sites so to minimise simultaneous low electricity production events that they experience relative reference level. It shown resulting model closely related submodular and can be interpreted generalising well-known maximum coverage problem. Both deterministic randomised algorithms are discussed, including greedy, local search relaxation-based heuristics well combinations these algorithms. usefulness methods illustrated by realistic case study inspired onshore wind plants in Europe, instances featuring over ten thousand candidate locations years hourly-sampled meteorological data. proposed solution benchmarked against state-of-the-art mixed-integer programming solver several found consistently produce better solutions at fraction computational cost. physical nature provided also investigated, all deployment patterns unable supply constant share demand times. Finally, cross-validation analysis shows that, except an edge case, successfully reliably identify perform on previously unseen from historical spanning small weather years.