Associate Professor Zhong Haiwang of EEA, Tsinghua University and Professor Xie Le (an alumnus admitted to EEA in 2000) of Texas A & M University published "A Cross-Domain Approach to Analyzing the Short-Run Impact of COVID-19 on the US Electricity Sector" on Joule on November 18, 2020. The research team developed a set of public data sets to track the impact of COVID-19 in real time, strictly quantified the decline in power load caused by COVID-19, and found that retail mobility is the most evident to explain load changes. The research result expands the traditional understanding of the influencing factors on power load and helps to improve the operation efficiency of power system during pandemic prevention and control.

Since the outbreak of COVID-19, it has spread rapidly all over the world, which poses a serious threat to human life and health. At the end of March 2020, the United States became the country with the most confirmed cases in the world, and the pandemic situation in the United States developed rapidly and gradually got out of control in the following few months. In response to this public health crisis, various state governments have taken a series of measures, such as social restriction and work from home. These restrictions have changed people's production and life behavior, thus impacted the power load characteristics and the power system operation. How to strictly quantify the load changes caused by COVID-19 and how to explore the leading factors on load changes have become general concerns in the scientific and industrial fields.

In this context, the research team established and released a set of public databases (COVID-EMDA+) to track the impact of the pandemic on the power system in real time. The database has realized cross-platform aggregation, multi-time and multi-space collaboration, quality management and optimization of data from different industries, so as to help experts and policy makers in different fields to carry out further research.

Fig. 1: Flow chart of data aggregating and processing of the public data set

The key to strictly quantifying the impact of the pandemic is to assess the "counterfactual scenarios", that is, to control other conditions unchanged and estimate the power load level assuming there hadn't been COVID-19. The research team proposed an integrated backtracking model to assess the estimated value and uncertainty interval of the load in counterfactual scenarios. Tab. 1 shows the assessment result of the seven major electricity markets and seven typical cities in the United States. The result showed that peak decline rate of power load was 14.77% (New York City in May); the load changed rapidly in all markets and cities in February and March, reaching the peak in April and May, and gradually declined in June; horizontally comparing, the northeastern and central regions of the United States were the most seriously impacted.

Tab. 1 Load decline rate caused by COVID-19 in seven major electricity markets and seven typical cities

The change of power load is closely related to the development of pandemic, economic shutdown and work resumption arrangement. The research team further analyzed the causal relationship between load changes and a series of high-frequency dynamic variables (including the number of new confirmed cases, retail mobility, ratio of staying at home, etc.) in order to deepen the understanding of the impact mechanism of the pandemic and dig out the leading factors on load changes.

To deal with the cross-variable and cross-time series coupling relationship among related factors (see Fig. 2), the research team evaluated the influence degree of the related factors with the restricted vector autoregressive model and ensure the reliability of the results with a series of statistical tests. The final result showed that retail mobility was the most powerful to explain the load decline, which indirectly indicated that the decline of business load was the main factor on load change, and the conclusion is true in different time periods and different places. The number of confirmed cases was the core data for assess the severity of pandemic situation, but it's not ideal to explain the load change.

Fig. 2: Cross-variable and cross-time series coupling relationship among related factors

The preprint of the research achievement was released on arXiv and EnerarXiv platforms in May 2010. Since the achievement was released, it has been specially reported on IEEE Spectrum magazine and American media KBTX-TV, and the research team has been invited to give special reports in MIT, PSERC and other well-known universities and university alliances. The first author of this paper is Ruan Guangchun, a doctoral student of EEA, Tsinghua University (visited at Texas A & M University from 2019 to 2020). Professor Kang Chongqing is the co-author and Professor Xie Le is the correspondent author of this paper.

Paper link: https://www.cell.com/joule/fulltext/S2542-4351(20)30398-6

Public database link: https://github.com/tamu-engineering-research/COVID-EMDA

Joule is the first flagship energy journal of the publishing house Cell. It is positioned as the sister journal of the well-known journal Cell, with an impact factor of 27.054. Joule has been widely concerned and recognized by international academic circles since its publication, and it mainly includes high-level research achievements in energy-related fields.