While renewable electricity generation is considered key to a sustainable future, the variability and uncertainty of wind/solar generation at high penetration levels pose unprecedented challenges to grid reliability. In today's power systems, reliability is assessed by analyzing a set of contingency events, such as the loss of generators or transmission lines. However, these contingency-based methods are insufficient under large wind/solar ramping events, which become increasingly frequent and significant at high renewable penetration. Specifically, a large wind/solar ramping event should not be treated as a single contingency event. Rather, it must be modeled as an uncertain event-sequence that is sequentially revealed in time; yet the system operator must take immediate actions in the midst of such an event-sequence despite significant future uncertainty. Due to this critical difference, traditional contingency-based methods can no longer accurately quantify the resource requirement for reliability at high wind/solar penetration, leading to more costly dispatch of resources to offset poorly-managed uncertainty.
To address this open challenge, this CyberSEES project develops a new and mathematically rigorous computational framework for procuring and dispatching constrained grid resources, which aims to provably ensure reliable and economic grid operations under high renewable penetration. The project integrates reliability analysis at both the min/hour time-scales (for balancing demand and supply) and the second/sub-second time-scales (for dynamic security assessment). Specifically, the project team develops reliability-assuring online algorithms for day-head unit commitment and real-time economic dispatch at the min/hour time-scales, which provably balance demand and supply under large and uncertain wind/solar ramping events. Further, the online algorithms are tightly integrated with computationally-efficient set-based reachability analysis for assessing dynamic system security at the second/sub-second time-scales, which provides guaranteed bounds for dynamic system states in the presence of wind/solar uncertainty.
The success of this project will lead to paradigm-shifting advances in our ability to ensure power system reliability under high renewable penetration. At a broader scale, the results will contribute to the increased adoption of renewable energy sources worldwide, and will enable a smooth transition path to a sustainable future grid. The developed computational framework will also be applicable to small-footprint power systems, such as remote microgrids. Further, the results are expected to contribute to computation by advancing online algorithm design and reachability analysis, which will also be useful to other disciplines. The results of this work will be widely disseminated through conference and journal publications and actively shared with industry.
