Impact of Spatial and Technology Aggregation on Optimal Energy System Design

TL;DR

This paper introduces a novel two-step aggregation scheme for simplifying energy system models with renewable sources, balancing accuracy and computational efficiency in system design.

Contribution

A new aggregation method considering all model parameters and spatial contiguity, optimizing the trade-off between model accuracy and computational complexity.

Findings

System costs vary with aggregation levels, stabilizing at higher resolutions.

Optimal aggregation balances cost deviation (<5%) and computational savings (~93%).

Aggregation reduces runtime significantly while maintaining acceptable accuracy.

Abstract

Designing an optimal energy system with large shares of renewable energy sources is computationally challenging. Considering greater spatial horizon and level of detail, during the design, exacerbates this challenge. This paper investigates spatial and technology aggregation of energy system model, as a complexity-reduction technique. To that end, a novel two-step aggregation scheme based on model parameters such as Variable Renewable Energy Sources (VRES) time series and capacities, transmission capacities and distances, etc, is introduced. First, model regions are aggregated to obtain reduced region set. The aggregation is based on a holistic approach that considers all model parameters and spatial contiguity of regions. Next, technology aggregation is performed on each VRES, present in each newly-defined region. Each VRES is aggregated based on the temporal profiles to obtain a representative set. The impact of these aggregations on accuracy and computational complexity of a cost-optimal energy system design is analyzed for a European energy system scenario.The aggregations are performed to obtain different combinations of number of regions and VRES types, and the results are benchmarked against initial spatial resolution of 96 regions and 68 VRES types in each region. The results show that the system costs deviate significantly when lower number of regions and/or VRES types are considered. As the spatial resolution is increased in terms of both number of regions and VRES types, the system cost fluctuates at first and stabilizes at some point, approaching the benchmark value. Optimal combination can be determined based on an acceptable cost deviation and the point of stabilization. For instance, if <5% deviation is acceptable, 33 regions and 38 VRES types in each region is optimal. With this setting, the system cost is under-estimated by 4.42% but the run time is reduced by 92.95%.