A statistical approach to identify asynchronous extreme events for multi-regional energy system models

Phuong Minh Khuong, Hasan Ü. Yilmaz, Russell McKenna, Dogan Keles

Research output: Contribution to journalJournal articleResearchpeer-review


Purpose: With the growing deployment of variable renewable energy sources, such as wind and PV and the increasing interconnection of the power grid, multi-regional energy system models (ESMs) are increasingly challenged by the growth of model complexity. Therefore, the need for developing ESMs, which are realistic but also solvable with acceptable computational resources without losing output accuracy, arises. The purpose of this study is to propose a statistical approach to investigate asynchronous extreme events for different regions and then assess their ability to keep the output accuracy at the level of the full-resolution case. Design/methodology/approach: To extract the extreme events from the residual demands, the paper focuses on analyzing the tail of the residual demand distributions by using statistical approaches. The extreme events then are implemented in an ESM to assess the effect of them in protecting the accuracy of the output compared with the full-resolution output. Findings: The results show that extreme-high and fluctuation events are the most important events to be included in data input to maintain the flexibility output of the model when reducing the resolution. By including these events into the reduced data input, the output's accuracy reaches the level of 99.1% compared to full resolution case, while reducing the execution time by 20 times. Originality/value: Moreover, including extreme-fluctuation along with extreme-high in the reduced data input helps the ESM to avoid misleading investment in conventional and low-efficient generators.
Original languageEnglish
JournalInternational Journal of Energy Sector Management
Issue number2
Pages (from-to)352-384
Publication statusPublished - 2021


Dive into the research topics of 'A statistical approach to identify asynchronous extreme events for multi-regional energy system models'. Together they form a unique fingerprint.

Cite this