2.2. Choice of relevant factors and time-windows of interest

In order to analyze in what time-windows the residential electricity usage has changed most significantly due to the pandemic in 2020, an electricity-diurnal analysis was carried out. For brevity, we henceforth refer to the times before March 21st 2020 as the “pre-stay-at-home” period, and the times after that as the “stay-at-home” period.

First, it can be noted from Fig. 3 (a) that there are shifts in demand during the morning hours on weekdays, as previously described by Meinrenken et al. [15]: Pre-stay-at-home, the early-morning load ramp-up started at about 6.00am and peaked at 8.30am, followed by a decline, with no second ramp-up until the early evening. In contrast, stay-at-home usage exhibited a smoother ramp-up that started between 6.00am and 6.30am, reached the height of the pre-stay-at-home morning demand peak only at 9am, and then continued to increase through the morning and early afternoon [15].

(a) Stay-at-home and pre-stay-at-home electricity diurnals of one week in early April of 2019 and 2020, respectively. (b) Same for one week in July. Diurnals are shown separately for weekdays and weekends. Data tables provide the comparisons of electricity usage in 2020 versus 2019, namely: 24-hour electricity-use increase of weekdays and weekend, and 8-hour electricity-use increase (9am-5pm) of weekdays and weekend.

Regarding electricity use, Fig. 3 (a) shows that, overall, 2020 weekday electricity usage of apartments (24 h) shows a more significant increase (7% increase) versus 2019 use than on weekends (4% increase). These increases became more pronounced once advancing into warmer weather in July, where the increase in 24 h weekday-use above 2019 reached 13%, probably due to higher loads from air-conditioners (Fig. 3 (b)).

Studies for commercial buildings in the U.S. have shown that their principal electricity use is mostly concentrated in the worktime period (usually 9am – 5 pm) on weekdays [26]. Focusing on the same time window in the residential sector, when many residents would usually be at work/school or otherwise outside of their homes, the stay-at-home usage increases are even larger than over the 24 h period: Comparing 2020 to 2019 usage during 9am to 5 pm, one can see a 22% increase in average electricity use in early April and an even larger increase of 27% in early July.

Fig. 4 shows the overall trends in the 24-hour-electricity-use and 8-hour-electricity-use (9am—5pm) as percentage increases from 2019 to 2020, over the same period of Jan 1st – Aug 31st. Percentage increases in the hourly peak demand on weekdays between 12 pm and 5 pm are also shown (see rationale in section 2.2.3). It can be observed that the three characteristics, especially the hourly peak demand between 12 pm and 5 pm and the 8-hour electricity use, are correlated with two metrics, i.e., the outdoor wet-bulb temperature and the number of new confirmed Covid-19 cases in every month: During the stay-at-home period, the pandemic led to significant increases in residential electricity use, even when temperatures had not yet reached levels where air conditioning was required. These increases were therefore most likely due to an increased use of lights, appliances for food preparation, computers, and entertainment systems because more residents worked/studied from home. Once entering Phase 1 of the gradual reopening, new daily Covid-19 cases in NYC were declining, and the portion of residents remaining in their homes during the day was likely declining as well [27]. However, due to the higher outdoor temperatures now requiring increased cooling loads, the 8-hour electricity usage exhibits a notable further increase during the summertime in 2020.

(a) Increases in 24-hour-use, 8-hour-use, and 5-hour peak-demand (weekdays) between 2019 and 2020, by month. (b) Total monthly new confirmed Covid-19 cases in NYC in 2020, by month. (c) Average monthly wet-bulb temperature in 2019 and 2020, by month. The three areas in (a), denoted by blue, red, and yellow shading, represent the three degrees of government shelter-in-place orders in 2020 due to the varying pandemic severity in NYC. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

In summary, both the outdoor temperature and Covid-19 cases should be considered when explaining differences in electricity usage between 2019 and 2020.

According to the average weekday load profiles in NYC (NYISO [28], Fig. 5 ), system-wide demand in Jan. – Aug., 2019 and 2020 (i.e., residential, commercial, and industrial electricity usage combined) ramped up during the morning hours and reached 98% or higher of daily peak levels from 12 pm onwards until well into the afternoon. System-wide demand went down from 2019 to 2020, with the largest reduction (about 700 megawatts or 10%) around 4 pm to 5 pm. In contrast, based on our NYC residential electricity dataset (Fig. 3), there were substantial increases in hourly demand peaks from 2019 to 2020 during the time window of 12 pm – 5 pm (up to 50% on some days; not shown) as well as substantial increases in residential electricity consumption from 2019 to 2020 during the time window of 9am – 5 pm (up to 55%).

Weekday average hourly load (residential, commercial, and industrial; MW) in NYC from Jan. to Aug. in 2019 and 2020 (NYISO).

Based on the observations in sections 2.2.1 and 2.2.2, for the remaining analyses, we therefore focus on the following two characteristics of electricity usage, which capture different time windows and different electricity metrics:

(i) Average per-apartment electricity consumption (kWh) cumulatively from 9am to 5 pm on a given weekday, for brevity also referred to in this study as “8-hour-electricity-use”. This was analyzed in order to gauge the electricity usage (and associated costs) that can shift from the commercial sector (such as office buildings and schools) to the residential sector because of “stay-at-home” and/or “work-from-home” guidelines.

(ii) Hourly peak demand (Watt) for an average apartment at any time between 12 pm and 5 pm on a given weekday, defined at 1-hour resolution, for brevity also referred to in this study as “5-hour-peak-demand”. “Peak demand” was defined as the highest of the hourly average load (in Watts) between any two consecutive full hours in the time window of interest. To establish these, first, the hourly average Watts between 12 and 1 pm, 1–2 pm, … and 4–5 pm on a given day were determined, and then the “5-hour-peak-demand” on that day was taken to be the maximum of these five, hourly values. The peak demand was analyzed in order to gauge potential stress on local or region-wide grid infrastructure when increased residential demand coincides with (still) high system-wide demand. The peak demand during full or partial stay-at-home orders was further compared to the highest ever hourly residential peak in a no-pandemic condition in 2019. This peak typically occurs in the evenings of hot/humid days. The comparison was carried out in order to gauge whether the increased afternoon peak demand during widespread stay-at-home conditions could lead to black-outs or brown-outs of the local substations and distribution system in predominantly residential regions of a city (because the demand is larger than what the system was designed to handle), even if other areas of the city with higher commercial and industrial usage experience reduced system-wide electricity use (see Conclusions and Discussion).