Energy storage design and integration in power systems by system-value optimization

Abstract

Energy storage can play a crucial role in decarbonising power systems by balancing power and energy in time. Wider power system benefits that arise from these balancing technologies include lower grid expansion, renewable curtailment, and average electricity costs. However, with the proliferation of new energy storage technologies, it becomes increasingly difficult to identify which technologies are economically viable and how to design and integrate them effectively.

Using large-scale energy system models in Europe, the dissertation shows that solely relying on Levelized Cost of Storage (LCOS) metrics for technology assessments can mislead and that traditional system-value methods raise important questions about how to assess multiple energy storage technologies. Further, the work introduces a new complementary system-value assessment method called the market-potential method, which provides a systematic deployment analysis for assessing multiple storage technologies under competition. However, integrating energy storage in system models can lead to the unintended storage cycling effect, which occurs in approximately two-thirds of models and significantly distorts results. The thesis finds that traditional approaches to deal with the issue, such as multi-stage optimization or mixed integer linear programming approaches, are either ineffective or computationally inefficient. A new approach is suggested that only requires appropriate model parameterization with variable costs while keeping the model convex to reduce the risk of misleading results.

In addition, to enable energy storage assessments and energy system research around the world, the thesis extended the geographical scope of an existing European opensource model to global coverage. The new build energy system model ‘PyPSA-Earth’ is thereby demonstrated and validated in Africa. Using PyPSA-Earth, the thesis assesses for the first time the system value of 20 energy storage technologies across multiple scenarios in a representative future power system in Africa. The results offer insights into approaches for assessing multiple energy storage technologies under competition in large-scale energy system models. In particular, the dissertation addresses extreme cost uncertainty through a comprehensive scenario tree and finds that, apart from lithium and hydrogen, only seven energy storage are optimizationrelevant technologies. The work also discovers that a heterogeneous storage design can increase power system benefits and that some energy storage are more important than others. Finally, in contrast to traditional methods that only consider single energy storage, the thesis finds that optimizing multiple energy storage options tends to significantly reduce total system costs by up to 29%.

The presented research findings have the potential to inform decision-making processes for the sizing, integration, and deployment of energy storage systems in decarbonized power systems, contributing to a paradigm shift in scientific methodology and advancing efforts towards a sustainable future.