Are complex energy system models more accurate? An intra-model comparison of power system optimization models

•Identification of complexity drivers in power system optimization models.•Identification of Pareto-optimal models in terms of complexity and accuracy.•A more complex model formulation does not guarantee more accurate results.•Combining multiple time-coupling variables is a major complexity driver.•Complexity can be more easily reduced in investment models than in dispatch models. Today, optimization models are by far the most popular choice when analyzing energy systems. Impressive advances in computer and data sciences have allowed for a multitude of complex energy system optimization models. In our manuscript, we assess the hypothesis of a positive correlation between complexity of a model and accuracy of its results. For this, we propose a framework based on alternative model formulations and apply it in a case study with 160 different, more or less complex, implementations of power system optimization models for dispatch and investment. Our results indicate that a certain degree of complexity is indeed necessary to obtain sufficiently accurate results. However, a careful balancing is also necessary for acceptable use of computational resources. In terms of partial load efficiency of conversion processes, our findings indicate that (two) different formulations should be used for dispatch and for investment decisions. Furthermore, a model including simple grid constraints comes with only minor increases in complexity compared to a model assuming a copper plate. Moreover, the combination of multiple technical constraints for conversion processes leads to a significant gain in accuracy while the combination of multiple time-coupling constraints is a major complexity driver.