Estimation of Storage Parameters from Weather Data and Energy Systems Models

Thermo-mechanical energy storage systems, such as Carnot Batteries and compressed air energy storage, currently operate at a low technology readiness level, prompting ongoing research. While the engineering community focuses on efficiency improvements, other essential storage parameters and requirements are often estimated or neglected. Costs, typically expressed as the levelized cost of storage (LCOS), are considered another critical parameter. In many cost assessments, a daily charging and discharging cycle is assumed throughout the year, leading to an expectation of 365 cycles annually, often with assumed durations of 8 to 10 hours. This paper questions the realism of these assumptions and explores alternative approaches for estimating relevant parameters, as well as identifying key characteristics of power systems that influence these figures (e.g., differences between wind-rich and solar-rich countries). We compare two approaches for assessing energy storage needs: (i) The first utilizes weather data from the Deutscher Wetterdienst (DWD), which offers 10-minute resolution data over the past 30 years for various stations across Germany. This dataset allows for a straightforward analysis of actual periods of sunshine or wind exceeding predefined thresholds, as well as durations lacking these resources. Although in this method demand-side factors must be estimated, it provides a pragmatic means to identify local storage requirements without necessitating complex models. (ii) In contrast, the second approach employs energy systems models (ESMs) for fully decarbonized European countries, simulating positive and negative residual loads over time for various scenarios aimed at future carbon dioxide reduction, typically with hourly resolution. This method enables an in-depth analysis of period lengths and time-resolved positive and negative residual loads—where negative loads indicate surplus energy and positive loads represent energy demand—as well as the frequency of specific negative loads followed by positive loads, extracting essential technical parameters for storage solutions. However, utilizing ESMs requires considerable expertise and time, leading to questions about how effectively a purely data-driven analysis of weather data can replicate insights typically derived from ESMs. Both approaches reveal notable similarities and differences: The proposed design of daily charging and discharging periods of 8±2 hours appears unrealistic for wind-rich countries, where the number of alternating residual load periods from ESMs typically ranges between 150 and 175. Conversely, daily cycles are prevalent in sun-rich countries. The ratio of positive to negative residual loads varies between 0.9 and 1.88, often resulting in mismatches between directly following negative and positive residual loads, making complete discharges infrequent. Although the first approach may be more useful for designing local autonomous energy systems, an averaging approach could lead to more ESM-like results. When weather data for Cuxhaven (Germany) is analyzed and also converted to a consistent hourly resolution, the number of alternating periods is approximately 125 for solar energy and 226 for wind energy. Correlation analysis shows similarities between weather data and different country groups. This approach facilitates the analysis of storage duration needs across various time scales, leading to more realistic estimations of the levelized cost of storage (LCOS).