(749b) Quantitative Assessment of the Impact of COVID 19 in the Indian Power Sector Via Machine Learning

The COVID-19 pandemic has spread rapidly around the globe and caused significant challenges to the energy industry. India is one of the most affected countries by COVID-19, with over 13 million cases and 170K deaths [1]. Besides, India is also experiencing the second wave in 2021, which is much worse than the first one. In 2020, the pandemic forced governments worldwide to impose restrictions to prevent the spread of the virus. In India, a nationwide lockdown was imposed from March to May 2020 [2]. With a series of nationwide lockdown and social-distancing measures, the country gradually resumed economic activity after the lockdown in 2020. Policymakers and power system operators must take a scientific approach to understand and evaluate the impact of COVID-19 on the electricity sector and prepare for the second wave in India. The Indian electricity distribution companies (DISCOMs) have already been witnessing financial losses for the past decade (due to India’s growing energy demand as well as demand-supply imbalances). With the steep dip in the overall electricity consumption across various Indian states (as we have statistically quantified), the incurring losses of DISCOMs will only exacerbate. This could threaten the energy security of India going forward.

Several reports have been published, both peer-reviewed and non-peer reviewed articles have analyzed the impact of COVID-19 on electricity consumption [3-5]. However, these assessments have not calibrated a baseline electricity consumption profile in the absence of the pandemic. Also, no studies have quantified the effect of COVID-19 dependent factors on electricity consumption.

In this work, we have developed a backcast model used to determine the electricity consumption profile if no COVID-19 occurred. The model gives us a fair estimation of the baseline and is used to determine the percentage of deviation in the electricity consumption. After the backcast model, we developed a regression model to capture the effect of various factors on electricity consumption in 2020. We analyze the impact of the three lockdown phases (partial lockdown, strict lockdown, lifted lockdown) on India’s power sector for seven states (i.e. Delhi, Maharashtra, Gujarat, West Bengal, Uttar Pradesh, Tamil Nadu, and Karnataka), constituting 65% of the total electricity consumption. Such state-level quantifications as analyzed in this study, can be directed to National Government to assist in the relevant state level policy making and tariff related support to state DISCOMs to ensure the nation-wide energy security in coming years.

The backcast model is developed to forecast the energy consumption of 2020, based on historical data for the past three years (2017-2019). It is a classical time-series model to capture the trends and seasonality. The model is expressed as a function of the factors that affect the power consumption, including weather parameters (temperature, humidity, and wind speed) and economic parameters, i.e., GDP. Using the forecast for 2020 by the backcast model as our baseline, the analysis shows that the various states experienced anywhere between 17-44% reduction in energy consumed than the same time period from the last two years and baseline 2020. The state-wise dip in electricity consumption eased during phases 2 and 3, ranging between 6%-21%, and gradually reaching the regular norms by the end of August 2020.

In the second model, we developed a support vector regression (SVR) model to map the state-wise energy consumption as a function of weather parameters (e.g. temperature, humidity, wind speed) and social distancing factors (reduction in work population, increase in homestay). To estimate the power consumption during the second wave of COVID-19, we then devised energy consumption

scenarios for each state for the months from May – August 2021, by varying the lockdown parameters. This model enables us to use the SVR model to create May onward scenarios to estimate energy consumption across states under varying degrees of lockdown. Certain states in India are going through the second lockdown (partial). A stricter lockdown measure in Delhi, from May onwards, may inflict a reduced energy consumption for the period from May to mid-June.

We also studied the contribution of renewable sources in the grid mix. India managed to increase the % RES (Renewables) in the grid mix by 4.83 % compared to 2019, despite generating 2.37% less energy for the same year. Therefore, India managed to keep the RE mix of the grid at 10-11% in 2020 compared to 8-9% in the preceding years of 2018 and 2019. The current crisis and studies such as these present an opportunity for policy analysts to convene and lend a common voice to long-pending structural changes needed in the power sector in India.

References:

1. Worldometer. Worldometer-India. 2021; Available from: https://www.worldometers.info/coronavirus/country/india/.

2. Lancet, T., India under COVID-19 lockdown. Lancet (London, England), 2020. 395(10233): p. 1315.

3. IEA, Covid-19 impact on electricity. 2021.

4. EnergyWorld. Coronavirus impact: Within ten days, 26 per cent fall in India’s energy consumption. 2020; Available from: https://energy.economictimes.indiatimes.com/news/power/coronavirus-impac....

5. Aruga, K., M. Islam, and A. Jannat, Effects of COVID-19 on Indian energy consumption. Sustainability, 2020. 12(14): p. 5616.