Systematic comparison of high-resolution electricity system modeling approaches focusing on investment, dispatch and generation adequacy

The transparency and availability of energy system models and their input data are of particular importance due to the increasing systems complexity and policy relevance. In recent years, a large number of model-based scenario analyses have been carried out. These analyses are based on various model approaches and lead to a range of results, which due to different data structures and mathematical approaches seem hardly comparable.

In this investigation, we attempt to develop an understanding of different electricity system models and their mathematical approaches through systematic model comparison to facilitate comparison of model results. Detailed and harmonized scenario input parameters are the basis of this model experiment including the electricity system models IDILES-JMM1, PowerACE2 and ELTRAMOD3.

The focus is on a selection of output parameters, such as investment and dispatch decisions in flexible power plants, storage dispatch, carbon emissions and generation adequacy in hours with critical supply situations with a focus on Germany and a considered time horizon until 2030.

The results indicate that all models see a need for investments in flexible power plants of at least 2.6 GW until 2030. ... mehrAdditionally, the generation adequacy indicators signal despite the expansion in continuously reliable capacities that the residual load cannot be covered in every hour by domestic generation. There is a maximum need for additional flexibility of 12.4 GW in the German electricity market in 2030 (by assuming that pumped storage plants (PSP) are fully available). The supplementary flexibility needs to be provided in form of further generation capacities (e.g., flexible power plants, storage discharge), electricity import capacities or demand flexibility. Furthermore, gas-fired power plants will play an important role in terms of their share in electricity generation. The full load hours of gas-fired power plants increase, while the utilization of lignite-fired power plants decrease substantially. A driver of this development is the increase in CO2 prices that lead to rising contribution margins particularly in 2030 for power plants with low CO2 emissions (e.g., gas power plants, reservoir or PSP).

These differences in the results can not only be distinguished to their mathematical approaches (i.e., linear optimization vs. agent-based simulation), but also to their level of detail and the foresight perspective (e.g., rolling planning algorithm with 24- and 36-hours loops vs. perfect foresight in a closed loop for one year). Overall, the deviations between the model results are relatively small. This is an important finding as it confirms the validity of the models and concludes that results of electricity system analyses are comparable under a few premises, even between differing methodical approaches.