Storage demand in highly renewable energy scenarios for Europe : the influence of methodology and data assumptions in model-based assessments

Abstract

Future low-carbon energy systems are likely to rely on power generation from variable, renewable energies (VRE) sources, thus fostering an increased demand for flexibility options which can balance mismatches of power demand and supply. Electrical energy storage (EES) is a promising option to tackle this matter and its required capacity is typically studied with model-based assessments. However, such analyses barley account for uncertainties in data assumptions or the chosen methodology, and, in consequence, lack an understanding of the robustness of the derived EES capacity. Therefore, this thesis aims to shed light on the main drivers for EES demand in highly renewable European energy systems in a comprehensive approach, considering parametric and methodological uncertainty. Using and enhancing the linear optimization model REMix, this study analyzes the required storage capacity and its utilization for northern, western, and central Europe in energy scenarios with high shares of renewable power generation (> 80%). The robustness of the storage capacity was tested against a large set of parameter variations (e.g. cost parameters or the meteorological year as input for VRE power) and methodological assumptions. The latter include different levels of technological detail (e.g. modeling approaches for thermal power plants), variations in the spatial and temporal resolution, as well as more general assumptions (e.g. restricted curtailments). In the reference scenario an overall EES capacity of approximately 200 GW and 30 TWh for Europe was derived. These results are particularly sensitive to investment costs variations of EES and VRE technologies (I) and to assumptions regarding the transmission grid infrastructure (II). (I) Reduced costs for storage and higher investment costs for VRE technologies increase the need for EES to 270 GW/55 TWh and to 235 GW/38 TWh, respectively. (II) Reducing transmission grid congestions can lower the ESS demand considerably, however, the analysis also showed that - even in the scenario which favor transmission the most (i.e. low investment costs for grid expansion) - around 120 GW of storage converter power and 13 TWh of storage unit capacity is still required for temporal balancing. In this sense, grid expansion and storage are not complete substitutes, but complementing flexibility options, both essential for future energy systems with high shares of VRE power generation. Moreover, the model-endogenously derived EES capacity mix in all scenarios is technology-diverse, underlining the necessity for a balanced storage portfolio. These findings are supported by the high dependency of the spatial capacity distribution of storage with the regionally predominant VRE technology and its temporal power generation characteristics. In this regard, significant correlations between the electricity generation from offshore and onshore wind systems with hydrogen storage charging are observed. Onshore wind power production also correlates with adiabatic compressed air storage, whereas the generation of photovoltaic systems is predominantly balanced by stationary lithium-ion batteries. To analyze the impact of the technological detail on storage demand, a comparison of two approaches for modeling thermal power plants was carried out: a detailed, mixed-integer unit-commitment approach and a simplified economic dispatch method. The results indicate that for larger observation areas (e.g. Europe) with high VRE shares, in-depth modeling is not necessarily required, however, analyses for smaller model regions in combination with lower VRE penetration levels can greatly benefit from detailed power plant modeling.