The Lockdown and Smart Working Effects on Electric Energy Consumption: The Analysis for a Group of Employees

The Covid-19 pandemic, aside from its dramatic effects, has accelerated transformations of the society, some of them already underway. One of these accelerations, with the goal of limiting the spread of the infection, concerns the diffusion of smart working, that is the working life and the jobs carried out also outside the office, mainly at home, and not only inside the office buildings. In Italy, this opportunity has been allowed since 2017 by the legislation (Legge n. 81/2017), but it was very poorly exploited before the pandemic crisis. It is interesting to remark that this change in the job organization has consequences not only on health and social aspects, but also on the energy sector, by decreasing energy transport consumption and shifting electric consumption from the job offices (tertiary sector) to homes (domestic sector). In this study, in particular, the shift from consumption in public administration to the domestic sector is investigated.

The dramatic lockdown measures adopted in many countries represent a peculiar period (a sort of “unplanned lab”) where these effects can be observed at its maximum extent. Besides the Italian “strict lockdown” of March-May 2020, different time periods can be studied, with a different rate of job at the office and job at home. In future, a new and more appropriate equilibrium will be probably found, according also to enterprise and national policies. Smart and home working have interesting consequences on the total amount of energy consumption and on the shape of the hourly electric load in the domestic and tertiary sectors. Therefore, also the economy of PV installations, depending on self-consumption, is affected. The goal of this paper is to begin to investigate these consequences.

In section 2, the energy statistics in Italy for different sectors are reported, focusing on the average consumption of the Italian families. As case study, the electric consumption behavior of a group of 10 employees, working with classical office equipment, has been explored. The representativeness of the consumption of the selected users and of their families has been verified, comparing the data with the literature.

Section 3 concerns the energy data analysis at the offices. The data of the electric consumption in the building with the offices have been provided by appropriate monitoring electric systems. The average electric consumption (energy per employee) is estimated by monitoring the amount of people really present in the building itself.

Section 4 discusses the domestic consumptions of the chosen users. The average electric consumption is estimated by considering the number of residents. The electric load values have been analyzed during and after the lockdown restrictions introduced by the Italian government. The change of the total consumption and of habits, with respect to the previous year, has been analyzed and discussed.

The changes in the electric load values can affect, in case of adoption of PV panels, the net energy consumption, PV self-consumption, PV self-sufficiency and PV return of investments. Some of these issues are estimated in Section 5.

Thermal consumption due to heating is out of the scope of this study.

3.1 Description and method

The people of the sample under investigation work in the same building (in Bologna, Italy) as employees, using typical office equipment, with R&D and administrative tasks. Electric consumption is not monitored individually, but the data refer to the whole building, hosting, in about 70 rooms and 1600 m², about 100 employees (with an average of 75 employees really present daily).

The monitored data are collected in the years 2018-2020, separately, according to three different uses: lighting (annual contribution of 16%), air conditioning (17%) and Power supply at the Sockets (P.S.) (for PCs, servers, monitors, printers, drivers, copy machines, etc.) (contribution of 67%). The average contribution due to air conditioning, in July and August, turns out to be about 40-50%.

A software has been developed for the data monitoring and analysis: the data are collected using the Apache Camel routing and mediation engine, with a custom and internally developed Modbus connector. These components are deployed in a deep customized Apache Karaf container, named SignalMix. PostgresSQL has been used as DataBase Management System tier, and Java for introspection and data analysis.

The average quarter-hour electrical load curve was calculated adding the three electrical components (lighting, air conditioning and P.S.) and applying a spline interpolation to achieve a continuous load curve in case of lack of data for less than 6 hours.

Table 3 shows the number of PCs and of monitors per employee. The number of electronic devices per employee turns out to be about 4.

The annual electricity consumption for the building turns out to be about 52,268 kWh.

Since the total electric consumption of the buildings obviously depends on the number of people actually present at work, an evaluation of the building occupancy is necessary, as studied also in [6]. The number of people nominally employed in the staff, has not been considered a value precise enough, since the absence due to many factors (e.g. smart working, tasks outside the building, flexible timetables, holidays) make unpredictable, on daily basis, the number of people really present in the building.

Therefore, an algorithm has been developed in order to evaluate, in real time, the actual occupancy of the building, by monitoring and recording (anonymously) when each person enters and exits.

Table 3. PC and monitors normalized to the number of employees (87 in staff)

In this way, the electric load and the consumption, day by day, can be normalized to the actual number of employees in the building and an average value per employee can be estimated. Only normalized values can be somehow useful in evaluating the energy efficiency of the devices and of the behaviors and in planning interventions to improve them. This methodology becomes particularly interesting when smart working becomes an important time fraction of the job, as happened in many months of the years 2020-21.

In order to visualize the correlation between occupancy and power load, both of these quantities are displayed in the same chart in Figure 4: the loads due to lighting and to P.S. are separated during a typical working day. It is clear how the load increases with higher occupancies and how it decreases during lunch time.

4.jpg

Figure 4. Power load vs building occupancy

In this way, it is possible to easily separate the contribution due to the base load (independent of the occupancy) from the contribution due to the presence and work in the office.

With this approach, the building occupancy has been calculated as integration in time of the number of people: Occupancy =$\left(\sum_{i} p\left(t_{i}\right) \cdot \Delta t_{i}\right)$, where p(t_i) denotes the number of people present in the building at time t_i and t_i the time of presence of p(t_i) people.

The average load per employee can thus be obtained by subtracting from the total daily consumption [kWh_d], the base load kW_b times 24h (obtaining therefore the consumption due to all the employees) and dividing then by the occupancy of the day (people·hour):

$\frac{W}{\text { employee }}=\frac{\left[k W h_{d}-24 \cdot k W_{b}\right]}{\left(\sum_{i} p\left(t_{i}\right) \cdot \Delta t_{i}\right)}$              (1)

The value of kW_b is assumed to be constant during the whole day.

The base load due to the P.S. varies in the range 2.1 kW (min) – 4.2 kW (max), with an average of about 3.2 kW. The variability of the base load is mainly due to the use of several servers. Normalizing to the number of employees, the P.S. base load turns out to be 36 W/employee on average. The contribution of the base load is high: the base load for the P.S. weighs about 45-50% on the total P.S. consumption and the percentage of the consumption out of the working time (from 9 p.m. to 7 a.m. and during weekends and holidays), summing up all the components, turns out to be 35% on the total consumption.

The power load, averaged per employee (job in presence), summing up P.S. and lighting (without considering air conditioning), turns out to be about 100-110 W/employee in July and about 200-230 W/employee in December (2018). The motivation of this season dependency is due to the use of heating and lighting devices.

All these contributions drive a daily electric consumption per employee, considering 365 days and the actual presence of the employees in the office, of 2.2 [kWh/day/person]. When excluding air conditioning and the base load, one can estimate that the consumption of the employee is 1-1.2 kWh/day/person. Obviously, these data refer to average values and high fluctuations are present due to the heterogeneity of the electronic devices, job tasks of the employees, day of the week, external conditions of illumination and temperature.

3.2 Variation in the lockdown periods

The average quarter-hour electrical load curve of the 2020 “strict lockdown” period (16 March-17 May) has been analyzed and compared to that of the previous year. Figure 5 shows the downfall of the average values throughout the day due to the absence of most of the employees, leaving almost only the base load consumption.

Table 4. Electricity consumption change in the offices

Note: d = day and p = person.

The average values of the electricity consumption in the building during the 2020 “strict lockdown” was investigated as well: the results, normalized per day (d) and person (p), are reported in Table 4. The consumption turns out to be lower by about 47% and the value of variation (1.01 kWh/d/p), is consistent with the value obtained in the previous section, referring to the consumption without air conditioning and the base load (1-1.2 kWh/d/p).

5.png

Figure 5. Variation of the average electrical load curve in the office buildings during weekdays (“strict lockdown” period)

5.1 Description and methods

In this section PV panels are supposed to be installed, without energy storage, over the building where people work and over the homes of the people under investigation. This hypothesis is useful to evaluate the impact of lockdown and smart-working on PV self-consumption and PV self-sufficiency in this case study, as an example of tertiary and domestic sector.

The maximization of self-consumption and self-sufficiency represents a common goal to increase energy independency, to lower the stress on the electricity distribution grid and to reduce the dimension and losses of a possible storage [8].

In this study, PV self-consumption (kWh) (also called absolute self-consumption [8]), refers to the consumption of self-produced energy (i.e., the minimum between load and production), while the PV self-consumption rate refers to the ratio between PV self-consumption and the total PV production. On the other side, the PV self-sufficiency rate refers to the ratio between PV self-consumption and the total consumption.

The results are obtained simulating, hour by hour, the PV production, following method [9], according to real solar radiation data of 2019 [10] and assuming system losses of 10%. The nominal PV power (kW_p) has been chosen in such a way that the PV annual production (in kWh) is equal to the annual consumption (of the office building or of the average behavior of homes). This criterion, detached from other economical choices, is useful to analyze and compare the two situations (home and work). The PV nominal powers thus turn out to be:

• 1.6 kW_p for the household (0.73 kW_p/person);
• 42 kW_p for the office building (0.56 kW_p/person).

The location of the panels has been chosen in Bologna (Italy), with 30° of tilt, south direction. The data analysis was performed for all the days of the week (not only workdays). Figure 9 shows some days of example for the electrical load, PV production and PV self-consumption curves in the office and in household. As expected, during the night, consumption is higher than production. The reverse happens during the daytime, where production is much higher than consumption.

9.png

Figure 9. Week example of load (black), PV production (red) and PV self-consumption (blue) curves in 2019

5.2 Variation in the lockdown periods

The load curve average behavior during the day, of the PV production and of the self-consumption, for the offices building, during March-May 2019 and during March-May 2020 (“strict lockdown” period of Table 5), is shown in Figure 10 and Figure 11 respectively.

The same comparison was made for the households in Figure 12 and Figure 13.

In these figures the self-consumption curve appears a bit lower than the load curve during PV production because the charts represent an average with different levels of PV production (sunny and cloudy days).

10.png

Figure 10. Average curves in offices without lockdown

11.png

Figure 11. Average curves in offices with strict lockdown

12.png

Figure 12. Average curves in households without lockdown

13.png

Figure 13. Average curves in households with strict lockdown

The self-consumption computed over 56 days in the March-May period (excluding some incomplete days with respect to 63 total days) turns out to be 4,031 kWh in 2019 and 2,228 kWh in 2020 for office building, whereas it turns out to be 112 kWh in 2019 and 175 kWh in 2020 for the average household.

Table 7. Self-consumption without and with lockdown

Note: d = day and p = person.

Table 8. Self-consumption rate without and with lockdown

Table 9. Self-sufficiency rate without and with lockdown

Table 7 and Table 8 show the results, respectively in normalized and percentage values, for the PV self-consumption, whereas the Table 9 shows the results, in percentage values, for the PV self-sufficiency.

In smart working, in the case study here analyzed, every person shifts, from the office to home, about 1.0-1.2 kWh/day. This consumption concerns mainly typical office equipment, electronic devices and lighting, but also some no-working related consumptions (family components at home). At the office building remains the base load, which in the case analyzed, turns out to be 0.8 kWh/d/p.

With the approximations with which we worked, no variation of the total electric consumption has emerged. Of course, the main advantage of smart working remains the energy saving in transport to move from home to the workplace.

These results have been obtained thanks to a methodology developed to estimate the actual office building occupancy, based on recording the transit of people at the building entrance. This approach becomes particularly interesting when smart working increases and therefore the building occupancy is difficult to estimate a priori.

Since the “shifts” are concentrated during the working hours, which are daylight hours, this impacts on PV self-consumption of supposed PV panels installed at the office building and at home. In the case analyzed, and assuming the nominal power kW_p such that PV annual production is equal to annual consumption, homes do not have high values for the self-consumption (32%). This is due to the fact that most of the people work outside home during the day. With “full” smart working, self-consumption acquires values slightly higher than that of the office building without smart working (50%). About 42% of the increase of the household total consumption becomes self-consumption.

In the office building, self-consumption drops from 44% to 24%, while self-sufficiency remains at a level of about 50%.

In the next years, with “soft” smart working, a value between these two extremes could be envisaged. These results are, on the other hand, dependent on the chosen PV nominal powers and need to be generalized.

Future investigations will involve also the extension of the study to the whole year, a wider sample of people, and a quantification of the economic variables.