TurboSPORES. Computational improvements to generate spatially-explicit alternatives for renewable capacity deployment

Abstract

Operations research is used to support large-scale capacity expansion planning in the context of the energy transition, most typically in the form of cost-minimising linear programming models. However, cost-optimality is often not the goal, or only one of many competing goals, in real-world planning. In recent work, we developed the SPORES method to generate spatially explicit near-optimal solutions. Previous near-optimal solution methods were unable to generate spatially diverse model results, while our method focuses explicitly on siting generation capacity on the sub-national scale, which is also where many competing objectives matter most, such as mitigating local landscape impacts. However, applying our method at the spatial resolution required by real-life decision-making structures means modelling hundreds of sub-national locations, leading to a trade-off with respect to computational tractability. Here we develop and empirically test several improved mathematical formulations that maximise the diversity of our near-optimal results, including formulations adapted from methods proposed in the literature. We compare and critically discuss the benefits and limitations of each, thereby providing generally applicable insights around how to best use limited computational capacity to generate the most diverse set of near-optimal alternatives without redundant computation.