How temporal and technological simplifications affect the performance of capacity expansion models

Bottom-up capacity expansion models often rely on simplified representations of the real energy system to cope with its complexity in manageable computing times. This allows to limit the size of the mathematical problem but entails a certain accuracy loss. In this paper, a novel indicator, the complexity reduction efficiency coefficient, is introduced, which relates accuracy losses to savings in deterministic computing time. This indicator is applied to measure the efficiency of different complexity reduction methods and their combinations in an electricity market model based on mixed-integer linear programming. This includes simplifications on the level of detail of thermal power plants and demand response, the level of aggregation of thermal power plants, and the temporal resolution. Results show strong correlations between simplifying one model dimension and the levels of detail and granularity of the other dimensions. Moreover, the combination of small simplifications in multiple dimensions is more efficient than large simplifications in one dimension. In addition, reducing the size of the mathematical problem can result in higher computing times due to an increasing similarity between possible optimum solutions. The complexity reduction efficiency coefficient can support the modeling community in identifying optimum ways of reducing complexity without compromising the accuracy of results. • Technological and temporal simplifications applied to a bottom-up capacity expansion model. • Novel indicator to relate losses in accuracy to savings in deterministic computing time. • The impact of simplifying a certain model dimension depends on the setup of the other dimensions. • Small simplifications in multiple dimensions are more efficient than large simplifications in one dimension. • Reducing the size of the mathematical problem may increase its tightness.