Governments around the world have taken measures to hinder the transmission of coronavirus disease 2019 (COVID‐19), some of which include mass lockdowns—however, there is substantial variation in how governments have adopted and how quickly physical distancing measures have been adopted. Beyond the global health crisis, the COVID‐19 pandemic is causing economic disruptions, and its impact is being felt across the energy system.
Energy consumption follows economic activity,1 and considering that electricity is employed in most economic activities, its consumption is a handy indicator of economic fluctuations. Electricity use can be therefore monitored to gauge the economic impact of COVID‐19,2 and Dutta et al.3 already provided evidence that COVID‐19 exerts substantial adverse effects on energy markets.
Electricity demands have decreased during lockdowns, with considerable reductions in services and industry only partially offset by higher residential use.4 Electricity‐related data over time can, therefore, offer some insight into the effects of the COVID‐19 crisis and associated lockdowns. The IEA has been publishing daily energy‐related data, and its mid‐April report showed that countries in full lockdown presented a reduction of 25% in weekly energy demands, while countries in partial lockdown presented a 18% decline.5 The Global Energy Review5 reported that the drop in energy demand in 2020 should be seven times the decrease after the 2008 financial crisis.
In Italy, the pandemic caused a reduction of up to 37% in electricity consumption compared to the same time in 2019, which reflected on daily profiles and market prices.6 In Ontario, Canada, the most significant daily electricity consumption drop was observed on weekends, with an average of 18% daily reductions and a peak of −25%.7 In Turkey, energy consumption values were compared with the same month in 2019, and demonstrated decreases of 5%, 20%, and 22% in March, April, and May.8 Decreases in energy consumption were identified earlier in China, as much as in January, due to the more advanced stage of the pandemic, with an average decrease of 1.5% in electricity consumption, which equals the consumption of Chile.9
In Brazil, the first trimester of 2020 showed a decrease of 0.9% in electricity consumption in comparison with 2019 values, and residential, industrial, and commercial sectors presented declines of 0.3%, 0.4%, and 2.2%, respectively.10 In the recent extraordinary review of the load forecast for 2020–2024, the impact of reduced consumption was experienced in all regions of Brazil, with the Southeast/Midwest region showing the most significant decrease (−3.6%), followed by the Northeast (−2.3%), South (−1.9%) and North (−1,5%) regions.11
According to the Brazilian Chamber of Commercialization of Electricity,12 the consumption of electricity in Brazil receded 11% in May, mainly due to mobility restrictions. Data demonstrated that, as mobility control measures continued throughout May and June, there was a stabilization of the drop in electricity consumption, between 10% and 13%. An increase was observed for the residential sector, while the commercial and industrial sectors remained with declines, except for the sanitary and food‐related divisions. Automotive and textile segments were the most affected, with 47% reductions.12
In light of the impact of the Covid‐19 pandemic and resulting lockdown measures on the energy system, the objective of the study presented herein is to present a timely analysis of electricity consumption trends. This study analyzes the effect of mobility restrictions due to the COVID‐19 pandemic on the energy consumption trends for the Brazilian energy system and its subsystems (Northeast, North, South, and Southeast‐Midwest).
To the best of the authors’ knowledge and after systematic reviews, this is the first study to employ joinpoint analysis for the calculation of energy consumption trends. Data presented herein is unique, in its focus on Brazil, which enables more accurate implications to be drawn for Brazilian policy makers. As Brazil begins to resume economic activity, a scientific approach to understand and predict the impact on the electricity sector is required.
2 MATERIALS AND METHODS
2.1 Energy demands
The analysis employed energy consumption data from the National Interconnected System (SIN), available from the operation history of the National Electric System Operator.13 All Brazilian states are connected to the ONS, except for Roraima (North region), which depends on electricity imports from Venezuela and electricity supplied by local thermoelectric power plants. Daily data considered electricity consumption plus losses, as reported by the daily electricity balance of ONS.
Consumption data were gathered on a daily basis, as mean GWh, for the period between January 1 and May 27, 2020. Data collection encompassed the Brazilian system and its regional subsystems (Northeast, North, Southeast‐Midwest, and South). Daily data were grouped into weeks.
2.2 Statistical Analysis
Data were analyzed by the Joinpoint Regression Program, version 220.127.116.11.14 The trends are assessed throughout time, according to significant modifications in their evolution patterns. Joinpoint regression has been applied to different fields, to evaluate changes in time series data, especially in health studies.15
Daily energy consumption data were grouped into weeks, and therefore the temporal unit employed herein was 1 week (7 days). The program identifies the joinpoint (time points when there is a significant change in trend) and calculates the percentage of change per time interval.
Weekly Percentage Change (WPC) was calculated to identify the statistical significance for each segment (p‐value<0.05), with a 95% confidence level. The analysis was developed considering heteroscedasticity and variance of Poisson assumptions. Significant modifications in the curve represent the joinpoints. The connection of linear elements, by a graph, enables a succinct characterization of trends.16 For the periods with a statistical significance of WPC, the trends can be classified as “increasing” or “decreasing”. For those values with no statistical significance, the term “stable” was employed. Models with zero to three joinpoints were analyzed, and the models that presented the best fit with observed data were selected.
A comparative analysis was also carried out between the period before the beginning of the isolation decrees in Brazil (until March 14) and the period after (as of March 15). Descriptive statistics values were calculated, and Student’s T and Mann‐Whitney’s U tests were carried out at a 95% confidence level.
3 RESULTS AND DISCUSSIONS
Regarding the routine of data capture, this step required intense attention and effort from the researchers: data was available in a graphic form, from the ONS website,13 and had to be manually extracted and introduced into a spreadsheet. Daily data were grouped into weeks to reduce the variability (across working days and weekend/holiday days) throughout the time series, which resulted in robust data assessment.
Regarding data analysis, the assessment herein presented demonstrates the utility of joinpoint regression for the energy sector, focusing on electricity. Figure 1 shows the behavior of the weekly electricity consumption in the Brazilian system, and Figure 2 depicts the weekly electricity consumption for the regional subsystems.
In Figure 1, for the Brazilian system, there are two joinpoints, at weeks 11 and 15. Between weeks 1 and 11, the WPC was 0.19, which represents a slightly increasing trend (although statistically non‐significant). For weeks 11–15, the WPC indicated a statistically significant decreasing trend with −5.37, followed by a relatively stable period for weeks 15–21 with WPC 0.25, statistically non‐significant.
The behaviors presented for the geographic regions in Figure 2 are similar to Figure 1, with two joinpoints. For the North region, all trends were statistically significant, with a rather stable trend between weeks 1 and 9 (WPC = 0.75*), a pronounced decreasing trend for weeks 9–16 (WPC = −2.04*), followed by an increasing trend for weeks 16–21 (WPC = 1.48*). For the Northeast region, none of the trends were statistically significant, with a rather stable behavior at the beginning (WPC = 0.47), between weeks 1 and 10, followed by a sharp decrease for weeks 10–13 (WPC = −4.70), and then with a less pronounced decline after week 13 (WPC = −0.63).
For the Southeast/Midwest subsystem, stable trends were identified for weeks 1–11 (positive, with WPC = 0.12) and 15–21 (negative, with WPC = −0.05), although with no statistical significance. A statistically significant decrease was detected between weeks 11 and 15, with WPC = −5.64*. Similar behavior was verified for the South region, with a rather stable trend for weeks 1–11 (WPC = 0.49, not significant), followed by a statistically significant drop (WPC = −6.94) between weeks 11 and 15. After week 15 there was a slightly increasing trend (WPC = 1.07, not significant).
Table 1 shows a comparison of electricity consumption data before (January 1 toMarch 14) and after (March 15 to May 27) the beginning of the isolation decrees in Brazil. A comparison of both periods reveals a statistically significant decrease in the consumption of electricity in Brazil and its geographic regions.
|Region||Period||Minimumpower (MW)||Maximumpower (MW)||Mean (MW)||SD (MW)||Median (MW)||p‐value|
|Brazil||1||10 678||11 950||11 520||381.58||11 630||<0.001b|
- a Student’s T,bMann‐Whitney’s U.
The results demonstrate a reduction in the consumption of electricity in Brazil and its geographic regions when comparing the periods before and after the beginning of mobility restrictions due to COVID‐19. Because the Brazilian geographic regions present differentiated profiles of electricity consumption, the decrease dynamics were also different. The Southeast‐Midwest subsystem experienced the most pronounced decrease (20% difference between the medians before and after implementation of the physical distancing decree), followed by the South region, which presented a 18% reduction in medians. The North and Northeast regions presented lower reduction rhythms, with 14% and 7%, respectively. In the Northeast, the majority of electricity consumption is associated with the residential sector, which has been the least affected by COVID‐19. The North region presents elevated participation of the industrial segment (concentration of metallurgical industries), which was also less affected than other sectors.17
With the increase in home office arrangements for a share of the Brazilian population, residential electricity consumption has increased. An accentuated decline in the consumption of the industrial, commercial, and transportation‐related sectors has also been observed. This significant decrease in the consumption of electricity in Brazil, due to the COVID‐19 pandemic, could leave electricity distributors with contractual electricity surpluses and some distributors could even face more significant issues.18, 19
The increase in electricity consumption at home is mainly related to the more prolonged use of computers, televisions, and air conditioning units in the warmer cities that do not present considerable environmental temperature fluctuation throughout the year. In the short term, the increase in electricity consumption due to the pandemic and home office schemes entails higher costs to households. This additional cost could generate an increase in failure to pay electricity bills and an overall decrease in well‐being levels. The Brazilian National Agency of Electrical Energy (ANEEL) has predicted this scenario and has determined that non‐payments will not lead to the suspension of electricity supply during the duration of the state of emergency ‐ this has been extended until July 31, 2020.20 A list of countries and their corresponding energy action plans was published by Qarnain et al.,21 showing temporary relief measures at the time of financial loss.
It must be highlighted that the dynamics of electricity consumption through the remainder of 2020 will depend on the duration, degree of strictness, and geographical coverage of lockdowns. Some factories are not operating at all, while others are operating at 50% or even 30% capacity. Goods are not being delivered, and people are staying at home (many suffering hardships). The speed of recovery will also affect the evolution of electricity consumption. In Brazil, the implementations of physical distancing measures were the responsibility of governors and mayors, and therefore there have been different degrees of mobility restrictions, which affect the consumption curves.
On a global level, electricity consumption decreased by 2.5% in the first trimester of 2020, but it must be mentioned that lockdown measures were enforced for less than a month in most countries.5 When full lockdowns were implemented, electricity consumption declined by at least 20%, with smaller decreases for partial lockdowns: initial IEA analyses indicate that full‐year electricity demand could drop by 6%, equivalent to the combined electricity demands of France, Germany, Italy and the United Kingdom in 2019.5 The impact of different containment measures on electricity consumption was verified by Bahmanyar et al.,22 showing that different lockdown measures and their effects on population activities have considerably changed electricity consumption profiles: in Spain, Italy, Belgium and the UK with severe mobility restrictions, electricity consumption was dramatically reduced—however, in the case of Sweden, with no lockdowns imposed, only a slight decrease was observed.
In Germany, the weekly consumption pattern remained almost the same (although at a lower level), but Italy, France, Spain, and Poland presented a smoother morning peak and preserved the evening peak.23 These lower levels of electricity consumption, if sustained, could place fossil fuel generators into a challenging position, due to constrained output and decreasing revenues. Some privately‐owned fossil fuel power plants (especially older units) could close, and government power utilities could require subsidies to survive.24 Because renewable outputs are usually dispatched first, they have no impediments to operate, and as a result, renewable energy penetration has reached its highest levels.24
Although shutdowns and lockdowns will be eventually eased, home office schemes could become a reality for many workers, and the International Labour Organization25 verified that 20%–30% of occupations can be performed remotely. Home energy use would increase, and there could be a substantial drop in commercial building use if telecommuting becomes widespread.26 In Italy, the forced closure of economic activities flattened the working hour consumption curves in comparison with pre‐lockdown curves, but the increase in residential consumption did not compensate for the general reduction.6
While the decline in electricity consumption will not be permanent, the shape and pace of recovery are still uncertain. Another aspect to consider is whether the changes in consumption patterns will persist post‐pandemic. The COVID‐19 crisis is having a significant impact across the energy sector. Although energy transition will slow down considerably, Leach et al.27 mentioned that it is unlikely that the pandemic will materially affect the energy transition on the long‐term and it has shown the importance of flexible supply. Energy consumptions patterns have changed during the enforcement of stay at home orders, leading to the unfolding implications on demand response programs. Some flexible management strategies considering multiple uncertainties at different time scales were reported by Ju et al.,28 and an integrated operation model for distributed generators, energy storage, and demand response was presented by Lin et al..29 Recommendations for energy policymakers on how to navigate the COVID‐19 crisis were reported by Steffen et al.,30 based on three principles: 1) avoiding overreaction in the short term; 2) searching for new opportunities for the energy transition in the mid‐term, and 3) developing new policy designs that can withstand future shocks.
Finally, governments are starting to formulate stimulus plans to help alleviate the adverse economic effects of the pandemic. Dincer31 adds that task forces should be launched immediately, focusing not only on health‐related issues, but also on energy and sustainability, and suggests that the coronavirus pandemic could be the closure of the carbon energy age. One of the repercussions of energy‐related issues associated with the pandemics is the deceleration of low‐carbon energy transitions, but this should be viewed as an opportunity (instead of a risk) to incentivize energy efficiency schemes. Co‐ and tri‐ generation systems can help boost economic competitiveness and supply more affordable energy, with lower associated impacts.
The outbreak of the new coronavirus (COVID‐19) has caused instabilities in the global economy and significant changes in the patterns of electricity consumption and production levels. This has had a direct impact on energy demands and carbon emissions, at least in the short term, for several countries. Naturally, these effects are not homogeneous on the population, varying by income, gender, type of employment, and nature of work, besides other aspects related to urbanization and access to essential basic services.
The mobility restrictions enforced due to the COVID‐19 pandemic have affected the levels of electricity consumption in Brazil and its weekly patterns, with statistically significant decreases. Because the Brazilian geographic regions present different profiles of electricity consumption, the reductions identified were also different. The Southeast‐Midwest and South subsystems depicted the most pronounced drops: −20% and −18%, respectively, when comparing the medians before and after implementation of mobility restrictions. The North subsystem presented a less marked decrease, −14%, because the industrial sector, constituted mainly of metallurgies, was less affected. The Northeast subsystem experienced a variation of −7% because most of its consumption is associated with the residential sector.
Further research is necessary to explore the duration of social changes (e.g., home office arrangements, video conferencing, e‐commerce) and the resulting dynamics of these behavioral patterns and energy consumption. The analysis of electricity consumption trends due to the COVID‐19 enables the study of scenarios regarding changes in behavior, in the short‐ and long‐ terms. Also, further research can focus on the supply‐side of electricity markets, showing the degree to which trends in the shifting mix of electricity supply are being affected.
The authors wish to acknowledge the support of the National Council for Scientific and Technological Development (CNPq, Research Productivity grant 307394/2018‐2) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior ‐ Brasil (Capes) ‐ Finance Code 001.