This title appears in the Scientific Report :
2024
Optimizing Computational Runtimes in Minutely Resolved Renewable Energy Systems Models
Optimizing Computational Runtimes in Minutely Resolved Renewable Energy Systems Models
The shift towards the modeling of energy systems with a high degree of temporal resolution is imminent, as future energy systems will utilize a large proportion of renewable energy sources, which are subject to intermittency. Hitherto, energy systems have been generally modeled using hourly data bec...
Saved in:
Personal Name(s): | Omoyele, Olalekan (Corresponding author) |
---|---|
Hoffmann, Maximilian / Weinand, Jann / Stolten, Detlef | |
Contributing Institute: |
Technoökonomische Systemanalyse; IEK-3 |
Imprint: |
2024
|
Conference: | Young Energy Ecomomists and Engineers Seminar, Leuven (Belgium), 2024-05-30 - 2024-05-31 |
Document Type: |
Abstract |
Research Program: |
Societally Feasible Transformation Pathways Effective System Transformation Pathways |
Publikationsportal JuSER |
The shift towards the modeling of energy systems with a high degree of temporal resolution is imminent, as future energy systems will utilize a large proportion of renewable energy sources, which are subject to intermittency. Hitherto, energy systems have been generally modeled using hourly data because of the lack of data at sub-hourly resolutions and the associated exponential increase in computational runtimes. However, the accuracy of these hourly models in terms of the total annualized costs, capacities of technologies, and general feasibility, is unreliable. This study analyzes the impact of different time resolutions on energy system modeling, examining various scenarios and weather conditions. The focus is on evaluating the accuracy and computational runtime in the capacity expansion and economic dispatch of a self-sufficient building and a micro-grid, respectively. The total annualized cost of the systems is underestimated by up to 1.7% between hourly and minute resolutions. Furthermore, higher temporal resolutions lead to significant changes in the system components’ capacities and operations, especially the highly dynamic ones such as inverters and storage. Likewise, there is an exponential increase up to a factor of 500 in computational runtime for minute resolutions relative to hourly ones.Additionally, to address the complexities inherent in the computational runtime, temporal aggregation is employed using two methods, namely period clustering and segmentation. The results for minute-resolved models show up to 99.5% accuracy, with computational runtimes equivalent to hourly resolutions without aggregation. We conclude that sub-hourly modeling produces more accurate energy system modeling and the deterrent in computational runtime can be alleviated through the proper use of aggregation methods in period clustering and segmentation. Future research may refine hourly data to sub-hourly levels to enhance energy system modeling accuracy and precision. |