Content uploaded by Wangda Zuo
Author content
All content in this area was uploaded by Wangda Zuo on Dec 10, 2021
Content may be subject to copyright.
Abstract
To minimize the indoor transmission of contaminants, such as the virus that
can lead to COVID-19, buildings must provide the best indoor air quality
possible. Improving indoor air quality can be achieved through the building’s
HVAC system to decrease any concentration of indoor contaminants by
dilution and/or by source removal. However, doing so has practical downsides
on the HVAC operation that are not always quantified in the literature. This
paper develops a temporal simulation capability that is used to investigate
the indoor virus concentration and operational cost of an HVAC system for
two mitigation strategies: 1) supplying 100% outdoor air into the building
and 2) using different HVAC filters, including MERV 10, MERV 13, and
Preprint submitted to Building and Environment December 9, 2021
C. A. Faulkner, J. E. Castellini, W. Zuo, D. M. Lorenzetti, and M. D. Sohn
2022. “Investigation of HVAC Operation Strategies for Office Buildings
During COVID-19 Pandemic.” Buildings and Environment, 207 (B), pp.
108519, https://doi.org/10.1016/j.buildenv.2021.108519
Investigation of HVAC Operation Strategies for Office
Buildings During COVID-19 Pandemic
Cary A. Faulknera, John E. Castellini Jr.a, Wangda Zuoa,b,c, David M.
Lorenzettid, Michael D. Sohnd
aDepartment of Mechanical Engineering, University of Colorado Boulder, UCB
427, Boulder, 80309, CO, U.S.A.
bDepartment of Civil, Environmental and Architectural Engineering, University of
Colorado Boulder, UCB 428, Boulder, 80309, CO, U.S.A.
cNational Renewable Energy National Laboratory, Golden, 80401, CO, U.S.A.
dEnergy Analysis and Environmental Impacts Division, Lawrence Berkeley National
Laboratory, 1 Cyclotron Road, Berkeley, 94720, CA, U.S.A.
HEPA filters. These strategies are applied to a hypothetical medium office
building consisting of five occupied zones and located in a cold and dry
climate. We modeled the building using the Modelica Buildings library and
developed new models for HVAC filtration and virus transmission to evaluate
COVID-19 scenarios. We show that the ASHRAE-recommended MERV
13 filtration reduces the average virus concentration by about 10% when
compared to MERV 10 filtration, with an increase in site energy consumption
of about 3%. In contrast, the use of 100% outdoor air reduces the average
indoor concentration by about an additional 1% compared to MERV 13
filtration, but significantly increases heating energy consumption. Use of
HEPA filtration increases the average indoor concentration and energy consumption
compared to MERV 13 filtration due to the high resistance of the HEPA filter.
Keywords: COVID-19, Indoor Air Quality, Building Energy, Modelica.
1. Introduction
The COVID-19 pandemic has increased the need for buildings to improve
their indoor air quality to help reduce the risk of infection from airborne
transmission. A recent study [1] found that all 318 identified outbreaks of
three or more COVID-19 cases in China occurred in indoor environments.
Another study [2] found an outbreak of 55 cases among 81 attendees of
2
in-person exercise classes at an indoor facility in Chicago, IL. It was also
found that the direction of the airflow due to the air-conditioning system
played a large role in the infection of patrons inside a restaurant in Guangzhou,
China [3]. These studies demonstrate the significant risk of indoor infection.
As a result, organizations, such as ASHRAE in January 2021 [4], provided
recommendations for building operation during the pandemic to improve
indoor safety. Included were recommendations for the heating, ventilation,
and air conditioning (HVAC) system operation, such as providing necessary
ventilation and using filters that achieve minimum efficiency reporting value
(MERV) 13 or better. The recommended strategies have been shown to
be helpful in improving indoor air quality, but their impacts on the HVAC
system operation, such as energy consumption, are not quantified and need
to be investigated further.
Previous literature tried to study and compare strategies to improve
indoor air quality to better understand this issue. Zhang et al. [5] conducted
experimental studies to investigate the effectiveness of different HVAC filters
to remove airborne viral particles. They found the viral filtration efficiency
generally correlated with the filter MERV rating and high-efficiency filters
were effective at capturing airborne viral particles. Zaatari et al. [6] studied
3
the effect of filter pressure drop on energy consumption and clean-air-delivery-rate.
They found replacing a MERV 8 filter with MERV 13/14 filter for a system
in fan-only mode with fan speed control increased energy consumption by
11-18% but improved clean-air-delivery-rate by a factor of 2.9-3.8. Santos
and Leal [7] characterized the relationship between energy consumption and
ventilation rates in European buildings and found increasing ventilation rates
can significantly impact annual building energy use. Other studies [8, 9]
compared the impact of different levels of ventilation and filtration on indoor
PM2.5 concentration and financial cost. Azimi and Stephens [10] studied risk
reduction of influenza virus and associated operational costs for different
filter ratings and equivalent levels of outdoor air ventilation. In addition,
they found filtration improved risk reduction with lower costs compared to
increased ventilation. A study during the COVID-19 pandemic [11] found
that, when sleeping in the same room as an infected person, running the fan
of the air-conditioning system can lower the risk of infection by one-third
and improving system filtration can reduce the risk by two-thirds. Pease
et al. [12] investigated the effect of filtration, air change rate, and outdoor
air fraction on the concentration of COVID-19 virus and probabilities of
infection in a multi-room building. They found filtration was the best method
4
for reducing virus concentration. The study also suggested that although
increasing outdoor air rate or air change rate is proved to be beneficial, they
should be used with caution. For example, increasing the outdoor air rate
can increase the heating or cooling energy used by the HVAC system due to
the temperature difference between the indoor and outdoor air.
Although significant progress has been made in the literature, further
improvements can be made. First, current studies lack detailed modeling
of the operation and control of the HVAC system, for example by assuming
constant ventilation rates. As a result, the dynamic effects of the HVAC
system, which are critical for real system operation, are lost. For example,
the controls of the outdoor air damper and supply fan can affect the dilution
and removal rate of indoor virus concentration as well as the HVAC energy
consumption. Also, many studies evaluate risk with steady-state concentrations
and constant occupancy, while both these values are dynamic in practice.
Additionally, the supply fan in the HVAC system is often assumed to be
sized to accommodate the additional pressure drop of high-efficiency filters.
However, in reality a building’s HVAC system may not be sized to replace
their existing HVAC filter with a more efficient filter that increases the
system pressure drop. Finally, while some of these studies perform annual
5
simulations, they do not compare the indoor air quality and their associated
operational cost during different times of the year. The effect of outdoor
conditions can determine the optimal operating strategy, which may vary
over the course of the year.
To address this gap, we use an equation-based, object-oriented modeling
language (Modelica) to develop a detailed model of an HVAC system and
enable temporal analyses. For example, Fu et al. [13] used Modelica-based
models to study cooling systems in data centers and investigate their performance
during normal and emergency conditions. Huang et al. [14] used Modelica-based
models to study control related faults for chiller and boiler plants. Tian et al.
[15] used computational fluid dynamics coupled with Modelica to optimize
thermostat placement in an office room based on thermal comfort and energy
consumption. The efficient dynamic modeling offered by Modelica language,
as well as the availability of the Modelica Buildings library [16, 17], were
enabling features to these studies. Additionally, our developed models for
HVAC filters and virus transmission, to our knowledge, have not yet been
created using a Modelica-based platform.
The remainder of this paper is organized as follows. First, a review of
HVAC operation strategies to reduce the risk of transmission of airborne
6
virus indoors is presented in Section 2. Next, the implementation of new
models for HVAC filters and virus transmission into an existing Variable
Air Volume (VAV) system model for a medium office building is detailed in
Section 3. Then, results for virus concentration and energy consumption for
the different strategies are shown and the combined results are analyzed in
Section 4. Finally, the paper is concluded in Section 5.
2. Review of HVAC Operation Strategies to Mitigate Indoor Disease
Transmission
There are several methods to improve indoor air quality by HVAC operation,
each with different benefits and drawbacks. Guo et al. [18] summarized and
compared HVAC operation guidelines during the COVID-19 pandemic for
buildings in different countries. Based on this review, common strategies
include: 1) increasing ventilation rates of outdoor air as high as possible,
2) running the HVAC system for longer periods to flush out lingering virus,
and 3) improving filtration of recirculating air. Increasing the supply rate
of outdoor air can dilute the indoor concentration of virus without the need
of purchasing and maintaining new equipment, as well as without increasing
the system pressure drop in the air handling system. However, increasing
7
the supply flow rate of outdoor air can significantly increase cooling or
heating energy consumption when the outdoor air temperature differs greatly
from the room temperature setpoint. It may also sacrifice the thermal
comfort as most HVAC systems are not sized for higher outdoor airflow rates
than the design. Secondly, increasing the runtime of the HVAC system, for
example running the system longer before or after occupants arrive, can help
flush out virus lingering in the room. Finally, improving the filtration of
recirculating air can reduce virus concentration without increasing heating
or cooling energy consumption due to increased ventilation. However, the
airflow resistance caused by HVAC filters may increase the pressure drop
through the air handling system, which could increase fan energy consumption
and/or reduce the system flow rate [6]. Furthermore, there are associated
financial costs of purchasing and installing the HVAC filters, as well as
replacing the filter after it accumulates particles over time.
This review suggests that mitigation strategies might benefit, or differ, if
an evaluation considered the typical change in occupancy, as well as cooling
and heating loads, over the course of a day. Thus, this paper studies the
strategies of supplying 100% outdoor air and using high-efficiency HVAC
filters, since these strategies can be used to improve the indoor air quality
8
throughout the day. There are many HVAC filters that are often defined
using MERV, which describes their ability to filter particles of different sizes
[19]. ASHRAE recommends [4] to use filtration that achieves at least MERV
13 during the pandemic due to their ability to filter at least 85% of airborne
particles with diameter between 1-3 µm. The most efficient filters are known
as high-efficiency particulate absorbing (HEPA) filters, which exceed MERV
16 and filter 99.97% of airborne particles with diameter at 0.3 µm [20].
Thus, this paper investigates the use of MERV 13 and HEPA rated filters,
representing the minimum recommended rating for use during the pandemic
and the maximum achievable rating. These strategies are evaluated against
MERV 10 filtration, which filters at least 50% of airborne particles with
diameter between 1-3 µm [19] and may be used in older buildings. It should
be noted the 100% outdoor air strategy is assumed to use a MERV 10 filter,
since buildings use HVAC filters to filter outdoor air as well.
3. Model Implementation
For the temporal simulations, we first developed a temporal simulation
of a variable-air-volume (VAV) HVAC supply system for a typical medium
office building. We based the building model along a prototype provided
9
in the Modelica Buildings library [21]. For our simulations, we developed
modules to supplement the prototype to represent HVAC filtration and virus
transmission. In this section, we describe the features of the model of the
medium office building and our qualitative verification of the new models to
support this study. Finally, we describe the whole building model with the
new modules incorporated.
3.1. Overview of the Building System
This building was based on the DOE commercial reference medium office
building [22], with focus on the bottom floor building prototype. The floor
contains five zones, including a core zone and four perimeter zones, as shown
in Figure 1. These zones are assumed to be well-mixed in the model, with
volumes of 2,698 m3for the Core Zone, 569 m3for the North and South zones,
and 360 m3for the East and West zones. We have a central air handling unit
with heating and cooling coils with VAV terminal boxes containing reheat
coils in each zone. An outdoor air economizer is used to provide fresh,
outdoor air to the building and is controlled to supply at least the minimum
outdoor airflow based on the ASHRAE standard [23]. Additional outdoor
air may be supplied to provide free cooling. The HVAC system is sized for
the location of Denver, Colorado. Cooling is provided using chilled water
10
with coefficient of performance of 5, which is defined as the ratio of the rate
of cooling provided to the input electrical power. Heating is provided using
a hot water system with efficiency of 0.8, which is the ratio of the rate of
heating provided to the input required power. The system is controlled based
on the control sequence VAV 2A2-21232 from the Sequences of Operation for
Common HVAC Systems described in [24].
Figure 1: Floor layout of the medium office building with five zones.
3.2. Implementation of New Models
The development of the new models to support this study are detailed
next. First, the new HVAC filter component model is described, followed
by the models for virus generation and decay. The new models are also
11
qualitatively verified.
3.2.1. HVAC Filter Model
An HVAC filter model was developed to support the work for this paper,
since such a model was not included in the original VAV system model.
The model includes two main components: removal of virus based on a
defined efficiency and static pressure drop depending on the mass flow rate
and defined nominal flow conditions.
The removal of virus can be described as:
cout = (1 −ηf ilter)cin,(1)
where cout is the virus concentration exiting the filter, ηf ilter is the filter
removal efficiency in terms of percentage of virus removed, and cin is the
virus concentration entering the filter. The filter efficiency can be between
0-100%, where ηf ilter = 100% describes a filter that completely removes all
virus in the airflow.
The removal of virus for the HVAC filter model was qualitatively verified
with a simple unit study, by supplying air with concentration c0to a room
initially virus free for different filter efficiencies. The case was run for 500
seconds, when the concentrations approach their equilibrium values. The
12
results in Figure 2 show the normalized concentrations approach their expected
steady-state values of (1−ηfilter)c0, since the filter removes a fraction of ηfilter
from the supply flow with concentration c0. This confirms the HVAC filter
model removes virus as expected based on Equation 1.
Figure 2: Normalized room virus concentrations over time with different
HVAC filter efficiencies.
Next, the static pressure drop caused by the resistance of the filter is:
∆pfilter =kf ilter ˙m2
filter ,(2)
13
where ∆pfilter is the static pressure drop caused by the filter, ˙mf ilter is the
mass flow rate of air though the filter, and kfilter is:
kfilter =∆pnom
˙m2
nom
,(3)
where ∆pnom is the nominal pressure drop at the nominal mass flow rate,
˙mnom. These two values are inputs to the filter model. The quadratic relation
between static pressure drop and mass flow rate can be approximated using
the Bernoulli equation and captures the general trend from experimental data
[19]. It should be noted that the filter pressure drop increases over time as
the filter collects particles [25], but the nominal pressure drop was assumed
to be constant in this study for simplicity.
The pressure drop as a function of flow rate for the HVAC filter model was
demonstrated by supplying air at rates of 0.5 kg/s - 1.5 kg/s with different
filter nominal pressure drops. The nominal mass flow rate for the filter was
held constant at 1.0 kg/s for all the cases. The results output from the model
are shown in Figure 3.
14
Figure 3: Filter pressure drop vs mass flow rate for different nominal pressure
drops.
The results confirm the expected quadratic relation between pressure drop
and mass flow rate based on Equation 2. It can also be seen that the pressure
drops for the three cases pass through their nominal values at the nominal
mass flow rate of 1.0 kg/s.
The settings used for the filters used in this study are shown in Table 1.
The filter efficiencies come from ASHRAE technical resources [19] and the
nominal pressure drop values are chosen based on data for MERV 10 [26],
15
MERV 13 [27], and HEPA [10] filters based on the nominal flow rate of the
studied system. The pressure drop can vary for filters with the same rating
based on the filter type and depth of the filter [9]. The pressure drop also
increases over time as the filter accumulates particles. Thus, the nominal
pressure drops shown in Table 1 are chosen based on the average between
the typical initial pressure drop of the clean filter and the final pressure drop
recommended by the manufacturer.
Filter Nominal Pressure Drop (Pa) Efficiency
MERV 10 143 50%
MERV 13 162 85%
HEPA 373 99.97%
Table 1: HVAC filter simulation settings.
3.2.2. COVID-19 Virus Modeling
Additionally, models for the generation and decay of virus in the rooms
were developed for this study. First, we simulated the presence of one
“sick” person in each zone working from 9:00 AM to 5:00 PM, Monday
through Friday throughout the year. The purpose of the presence of sick
16
people throughout the year was to study the concentrations and effect of
strategies in all the zones during the different seasons. Generation of virus
was described in terms of quanta emission rates, where a quantum is the dose
of airborne droplet nuclei expected to cause infection in 63% of susceptible
people. Quanta emission rate of COVID-19 virus is difficult to quantify and
dependent on many uncertain values, such as viral load in the mouth and
activity level of the sick person [28]. We use a value of 25 quanta/hr for
the majority of the study based on literature [28, 29], although different
quanta emission values are considered in Section 4.1.3 to compare the risk
of infection for different emission rates. The virus was generated directly in
each well-mixed zone when the sick people were present based on the quanta
emission rate.
The viral decay in the room due to gravitational settling and the death
of airborne viruses is modeled based on Equation 4, which has been used in
literature to model viral decay in well-mixed zones [12]. This is described as:
˙cdecay,zone =kdecay cz one,(4)
where ˙cdecay,zone is the rate of viral decay in the zone, kdecay is a defined
constant rate of viral decay, and czone is the virus concentration in the zone.
17
The decay model was qualitatively tested by examining the viral decay in
a room with initial concentration c0. There were no other means to produce
or remove virus, other than loss due to a holistic decay rate. For this case,
the virus concentration in the room can be derived analytically as:
c(t) = c0exp(−kdecayt),(5)
where c(t) is the transient virus concentration in the room, c0is the initial
virus concentration, kdecay is the viral decay rate, and tis time. The results
for virus concentration over time with different viral decay rates are shown
in Figure 4. The results match the expected trend of the virus concentration
decaying exponentially and faster for larger values of kdecay.
18
Figure 4: Normalized virus concentration over time with different viral decay
rate values.
To quantify the impact of the virus concentrations, risk of infection is
calculated using the Wells-Riley approach, which determines this risk based
on the amount of virus inhaled by an occupant. Risk of infection is calculated
as:
R(t)=1−exp(−IR Zt
t0
c(t)dt),(6)
where R(t) is risk of infection in terms of percentage, IR is the volumetric
19
inhalation rate of air for an occupant, and Rt
t0c(t)dt is the integral of virus
concentration in the room with respect to time since initial time t0. The
predicted number of infections, R0, can be calculated based on the risk, R.
The predicted number of infections over time, R0(t) is calculated accounting
for the variable occupancy in the zone for this study. This is done by
calculating R0(t) for a given time interval when the occupancy is constant
and adding the predicted number of infections calculated from the previous
time interval. This can be described as:
R0,T (t) = ST[1 −exp(−IR Zt
t0
c(t)dt)] + R0,T −1(t0),(7)
where R0,T (t) is the predicted number of infections in the zone for time
interval T,STis the number of susceptible occupants in the zone during T,
t0is the time at the beginning of interval T, and R0,T−1(t0) is the predicted
number of infections from the previous time interval, T−1, ending at time t0.
Susceptible occupants is determined as S=N−1, where Nis the number of
occupants. This way Sdoes not account for the sick person, since they cannot
infect themselves. Figure 5 shows the predicted number of infections for a
sample day based on the virus concentration and occupancy. A sample time
interval, T, is highlighted to show how R0is calculated based on the number
20
of susceptible occupants and amount of inhaled virus during this time. At
the beginning of this time interval, the slope of R0(t) initially increases due
to the increase in occupancy as the susceptible occupants return from lunch.
R0(t) then steadily increases as occupants constantly inhale virus in the zone.
Finally, R0(t) quickly flattens out after this interval as occupants leave and
the virus concentration decreases. For this study, the values for viral decay
and inhalation rate were chosen to be 0.48 hr−1and 0.48 m3/hr based on
literature [12].
21
(a) Occupancy during the day.
(b) Normalized virus concentration.
(c) Predicted number of infections.
Figure 5: Predicted number of infections based on the occupancy and virus
concentration for a sample day.
22
3.3. Whole Building Model
The newly developed models were added to the VAV system model to
perform this study and the final Modelica model capability is shown in Figure
6. The entire model can be divided into four sections: (1) the multizone
airflow model for the five zones, which includes the generation and decay
of virus in the zones; (2) the VAV system model which includes the central
air handling unit, as well as VAV terminal boxes and the return duct; (3)
the control system which includes PI controllers for the heating and cooling
coils, outdoor air economizer, and supply fan; and (4) the weather conditions,
including dry bulb temperature, wind speed, and radiative exchange.
Figure 6: Modelica model of the medium office building.
23
4. Results and Discussions
In this section, the results for indoor virus concentration are presented
first, followed by the results for energy consumption. Finally, analysis of the
combined results is performed to consider best overall strategies based on
both indoor air quality and operational cost.
4.1. Virus Concentration Results
The virus concentration results are presented for different time scales in
this section. First, the annual average virus concentration results in the five
zones for the different strategies are presented. Next, the monthly average
results are presented to show how the concentrations vary throughout the
year. It is worth noting that the annual and monthly averages only account
for the concentrations during occupied hours. Finally, results from two
sample days are presented including risk analysis based on predicted number
of infections.
4.1.1. Annual Virus Concentration Results
The results of the annual-average virus concentration by four different
strategies in five different zones are shown in Figure 7. The indoor virus
concentration results in this figure and throughout this section are normalized
24
by c0, which is the building average virus concentration for the MERV 10 case
(first blue bar on the left in Figure 7). The results show that the strategy of
supplying 100% outdoor air provides the lowest annual building-average virus
concentration and reduces the annual building-average virus concentration by
about 11% compared to using MERV 10 filtration. While using MERV 13
filtration does not reduce the virus concentration as much as supplying 100%
outdoor air, it still reduces the annual building-average virus concentration
by about 10% compared to using MERV 10 filtration. Use of HEPA filtration
only reduces the annual building-average virus concentration by about 5%
compared to use of MERV 10 filtration, despite the high-efficiency of the
HEPA filter. This occurs because the supply fan is not sized for the increased
pressure drop when using the HEPA filter, so the airflow supplied to the
building is reduced compared to the other strategies. In terms of the concentrations
in each of the zones, the Core zone has the lowest average concentration. In
large part this is because it is the largest zone but has the same number of
sick people as the other zones. The North zone has the highest concentrations
for all the strategies due to having the lowest nominal flow rate based on the
system sizing. This zone will be used for further analysis in Section 4.1.3.
25
Figure 7: Annual-average, normalized virus concentration results for the
different strategies and zones.
Figure 8 shows the variation of concentration for the four strategies during
the day throughout the year. For each weekday, the concentration begins to
increase at 9:00 AM when the sick people arrive, then decays quickly when
they leave at 5:00 PM. The MERV 10 case reveals the increased concentration
during the summer and some weeks during the winter when the HVAC
system tends to supply the minimum outdoor airflow. In contrast, more
outdoor air can be supplied during the mild shoulder seasons to provide free
cooling and/or increase ventilation. While the total supply flow rate does
26
not vary significantly throughout the year (for a given strategy), the fraction
of supplied outdoor air can vary significantly. The other cases do not show
as significant variation during the year, since they are less sensitive to the
outdoor air fraction due to their ability to remove virus efficiently.
(a) MERV 10 (b) 100% outdoor air
(c) MERV 13 (d) HEPA
Figure 8: Heat maps showing the magnitude of normalized concentration
during the days over the course of the year.
4.1.2. Monthly Virus Concentration Results
The monthly results for building-average virus concentration for the four
strategies are shown in Figure 9. This further reveals the implications of
27
changing seasons on the virus concentration for the different strategies. While
the average concentrations for the use of 100% outdoor air and HEPA filtration
vary slightly month-to-month, the MERV 10 case varies more due to its
sensitivity to outdoor air fraction. Also, the variation of concentration for
the MERV 13 case is more apparent in this figure, since it still does not
supply as clean air as the 100% outdoor air and HEPA cases. The results
show the lowest average virus concentrations for the MERV 10 and MERV 13
cases occur during April, October, and November when the weather is most
mild and the HVAC system tends to supply more outdoor air. Similarly, the
highest concentrations for these cases occur during the hot summer months
when the system often supplies the minimum outdoor air. This reveals the
advantage of supplying 100% outdoor air, relative to using MERV 10 and
MERV 13 filtration, during the summer based on the virus concentration.
The reduced advantage of the 100% outdoor air case during mild weather,
based on the virus concentration, is also apparent.
28
Figure 9: Monthly building-average virus concentration results for the
different strategies.
4.1.3. Sample Day Virus Concentration Results
Transient results for virus concentration and risk in the worst zone (North
Zone) for two sample days are shown in this section. The emission of
virus depends on the individuals and their activities, for example if they
are speaking or exercising. To capture that variation, three different virus
generation rates (q= 2 quanta/hr, 25 quanta/hr, and 50 quanta/hr) were
considered. These generation rates roughly span low, moderate, and higher
activity of a sick person.
29
First, virus concentration results for a hot summer day are shown in
Figure 10. This shows the differences among the strategies when the filter
cases use the minimum outdoor supply flow rate. All the strategies follow
a similar trend throughout the day. The virus concentration increases at
9:00 AM when the sick people arrive, tends towards an equilibrium during
the middle of the day, then sharply decreases when the sick people leave at
5:00 PM. The 100% outdoor air strategy reduces the virus concentration for
this day by up to 22% compared to the MERV 10 case. Use of MERV 13
and HEPA filtration reduces the virus concentration by up to 17% and 14%,
respectively, compared to MERV 10 filtration for this day. When the virus
generation rate is very low, as shown in Figure 10a, then the impact of the
different strategies is very small due to the low levels of virus concentration.
30
(a) q= 2 quanta/hr (b) q= 25 quanta/hr
(c) q= 50 quanta/hr
Figure 10: Normalized virus concentration in the worst zone for a hot day
with different virus generation rates.
To better understand the implication of these virus concentrations, the
predicted number of infections over time, R0(t), for this day are shown in
Figure 11. The higher generation rates increase the predicted number of
infections, as expected due to the higher virus concentrations. This is because
the predicted number of infections accounts for the amount of virus inhaled
31
by the susceptible occupants throughout the day. Both the occupancy and
concentration vary throughout the day, so the predicted number of infections
is calculated and summed for each hour to account for these dynamic effects.
For example, R0(t) begins to increase at 9:00 AM as the concentration
increases and susceptible occupants are exposed to the virus. The slope
of R0(t) then decreases at 12:00 PM when occupants leave for lunch, but the
slope increases again when they return at 1:00 PM. Finally, R0(t) flattens
out at its final value when the virus concentration decays to zero at the end
of the day.
Even for the most effective strategies, it is possible or even very likely
that one infection will occur in this zone if the generation rate is high, as
shown in Figure 11c. For example, R0reaches 0.75 at the end of the day for
the highest generation rate case with use of 100% outdoor air. This suggests
that there is a 75% chance that one person in the zone that day is exposed to
a level that could result in an infection. Use of 100% outdoor air offers the
benefit of reducing R0by about 0.20 at the end of the day when compared
to use of MERV 10 filtration. Use of MERV 13 and HEPA filtration reduces
R0at the end of the day by about 0.15 and 0.13, respectively, compared to
MERV 10 filtration. For the medium generation rate, the probability that
32
one infection occurs in the zone for this day is slightly under 50% for the
four strategies. R0is reduced by about 0.10 at the end of the day for the
100% outdoor air case compared to the MERV 10 case. Use of MERV 13 and
HEPA filtration reduces R0at the end of the day by about 0.08 and 0.06,
respectively, compared to MERV 10 filtration. The relative differences among
the strategies are more significant as the generation rate increases, and the
relative differences are essentially negligible for the lowest generation rate as
shown in Figure 11a. For the lowest generation rate, the R0at the end of the
day for the four strategies is between 0.03 and 0.04, meaning there is a 3-4%
chance one infection occurs in the zone for this day for all the strategies.
33
(a) q= 2 quanta/hr (b) q= 25 quanta/hr
(c) q= 50 quanta/hr
Figure 11: Predicted number of infections in the worst zone for a hot day
with different virus generation rates.
Next, the results for virus concentration in the worst zone for a mild
spring day are shown. For this day, all the cases supply 100% outdoor air
due to the control of the outdoor air economizer during the mild weather.
This causes the MERV 10, MERV 13, and 100% outdoor air cases to overlap
with each other for these plots. The HEPA filter case does not overlap with
34
the other cases due to the reduced supply flow rate to the zone caused by
the increased pressure drop of the HEPA filter. There is also an overshoot in
virus concentration around 10:00 AM for the HEPA case. This is due to an
initially lower amount of clean airflow to the zone, which eventually increases
and settles during the day.
(a) q= 2 quanta/hr (b) q= 25 quanta/hr
(c) q= 50 quanta/hr
Figure 12: Normalized virus concentration in the worst zone for a mild day
with different virus generation rates.
35
The predicted number of infections for this day are shown next in Figure
13. Since the concentrations overlap for the cases (excluding HEPA filter)
in Figure 12, it is expected the predicted number of infections also overlap
for these cases. Once again, even though these cases supply 100% clean air
throughout the day, it is still likely at least one infection occurs during the
day in this zone for the highest generation rate. The final value of R0for
the cases excluding HEPA is about 0.85, meaning there is still about an 85%
chance an infection occurs in the zone during this day for the high generation
rate. This corresponds to a reduction of R0by about 0.15 compared to use
of HEPA filtration for this day. For the medium generation rate, the final
value of R0for the non-HEPA cases is around 0.44 and is about 0.07 lower
compared to the HEPA case. Finally, the final value of R0is around 0.04 for
the four cases with the low generation rate.
36
(a) q= 2 quanta/hr (b) q= 25 quanta/hr
(c) q= 50 quanta/hr
Figure 13: Predicted number of infections in the worst zone for a mild day
with different virus generation rates.
4.2. Energy consumption results
The total annual energy consumption for the four different strategies are
shown in Table 2, including the breakdown of energy consumption in terms
of fan, cooling, and heating energy. The total source energy is also included
to show the energy required at the source to provide the site energy. In this
37
study, the fan and cooling devices use electricity while heating is provided
with natural gas. The source to site conversion factor is 1.05 for natural gas
heating [30] and 2.25 for electricity from the grid based on the breakdown
of electricity sources for Denver, CO [31] and their respective conversion
factors [32, 33]. For this system, the fan energy consumption tends to be
most dominant and this energy consumption increases for both the MERV
13 and HEPA filter cases due to the increased pressure drops. The heating
energy also tends to be more significant than cooling energy due to Denver’s
cold climate. This is shown by the large increase in heating energy for the
100% outdoor air case in order to heat the cold outdoor air during the winter,
compared to a smaller increase in cooling energy. There is also a decrease in
heating energy for the MERV 13 and HEPA filter cases. This is because heat
is dissipated to the airflow based on the power drawn from the fan. Thus,
more heat is dissipated by the fan as it draws more power for the MERV 13
and HEPA filter cases, which saves some heating energy. This also results in
an increase in cooling energy for the MERV 13 and HEPA cases compared
to the MERV 10 case.
Due to the increase in heating energy, the 100% outdoor air case consumes
about 54% more total site energy than the MERV 10 case and has the highest
38
total site energy consumption among the four strategies. In comparison, use
of MERV 13 and HEPA filtration increase the total site energy consumption
by about 3% and 12%, respectively, compared to MERV 10 filtration. However,
the total source energy for the cases gives a better representation of the
cost of energy for the building. Since the MERV 13 and HEPA cases use
more electricity to power the fan, they increase the total source energy
consumption by about 6% and 27%, respectively, compared to MERV 10
filtration. In contrast, the energy increase for the 100% outdoor air case
is mostly from natural gas heating, so the increase in total source energy
consumption is instead about 32%.
39
Case Fan
Energy
(MWhr)
Cooling
Energy
(MWhr)
Heating
Energy
(MWhr)
Total Site
Energy
(MWhr)
Total
Source
Energy
(MWhr)
MERV
10 filter
32.8 13.2 26.3 72.3 131.0
100%
outdoor
air
32.1 14.7 64.3 111.1 172.8
MERV
13 filter
36.7 13.5 24.3 74.5 138.4
HEPA
filter
53.4 14.4 12.9 80.7 166.0
Table 2: Annual HVAC energy consumption for the different strategies.
The monthly breakdown of energy consumption is shown next in Figure
14 to compare the operational strategies throughout the year. Both monthly
site and source energy are shown. All the cases excluding the HEPA case
40
consume less energy in the warmer months due to the dominance of heating
energy for this HVAC system and climate. This is not true for the HEPA
case due to the significant amount of heat dissipated by the fan for the high
pressure drop filter. The plots show a massive increase in heating energy for
the 100% outdoor air case during the colder months, with a less significant
increase in cooling energy for the 100% outdoor air case during the summer.
The source energy plot shows the more significant increase in energy for the
MERV 13 and HEPA filter cases, especially during the summer when mostly
electrical energy is used.
41
(a) Site energy.
(b) Source energy.
Figure 14: Monthly breakdown of HVAC energy consumption for the different
strategies.
42
4.3. Analysis of Combined Results
Based on the results for virus concentration, use of 100% outdoor air
provides the best overall indoor air quality. Although MERV 13 filtration
is slightly less effective, it offers similar levels of improvement in indoor
air quality with small increases in risk of infection compared to the 100%
outdoor air case. The energy consumption results show similar levels of
site energy consumption for the three filter cases, while the 100% outdoor air
case consumes more energy at the site compared to the other cases due to the
significant increase in heating energy. However, the increase in source energy
is more significant for the MERV 13 and HEPA filter cases due to the increase
in electricity to power the fan. The increase in source energy consumption is
less significant for the 100% outdoor air case since the additional energy is
mostly heating provided by natural gas. The final consideration is the cost of
the filters. The price of MERV 13 filters can range from $12-$190 depending
on the depth and style of filter used [9], while HEPA filters can range from
$250-$350.
Based on all these considerations, ASHRAE-recommended MERV 13
filtration performs best due to its improvement in indoor air quality with
relatively low operational cost. HEPA filtration could potentially be beneficial
43
if the system is sized to accommodate the increased pressure drop. However,
HEPA filtration can have negative effects on both indoor air quality and
energy consumption if the system is not sized for it. Additionally, the cost of
a HEPA filter is higher than a MERV 13 filter. Use of 100% outdoor air can
be used to provide slightly better indoor air quality compared to MERV 13
filtration. For this climate, it is most beneficial to use in the warmer seasons
to avoid the significant increase in heating energy during the winter.
5. Conclusion
Different strategies to improve indoor air quality during the COVID-19
pandemic are investigated for a medium office building in a cold and dry
climate. Specifically, the supply of 100% outdoor air and use of filtration with
MERV 10, MERV 13, or HEPA ratings are investigated throughout the year
using Modelica-based models. The building is modeled using the Modelica
Buildings library and new models for HVAC filtration and transmission of
virus are developed to support this study.
The results show the 100% outdoor air case reduces average virus concentration
by about 11% compared to MERV 10 filtration, but consumes significantly
more energy at the site compared to the other cases due to the large increase
44
in heating energy during the winter months. Use of MERV 13 filtration
reduces the average virus concentration by about 10% compared to MERV
10 filtration and shows similar results for risk compared to the 100% outdoor
air case. Use of HEPA filtration did not improve the indoor air quality
compared to MERV 13 filtration because of the reduced system flow rate,
since the system was not sized for a HEPA filter. The HEPA filter case also
used more energy compared to the MERV 13 case because of the increased
fan energy. Thus, using ASHRAE-recommended MERV 13 filtration can
achieve a good balance between the indoor air quality and operational cost.
In this paper, we develop computational modules and allow for temporal
assessment of exposure and risks of indoor occupants. We demonstrate
how such an approach allows one to consider the various tradeoffs between
exposure risk, HVAC capacity, and energy use. We also show how to consider
the marginal benefits of such tradeoffs for current crises and will thus be
available if future what-if scenarios are to be considered.
Acknowledgements
This research was supported in part by the U.S. Defense Threat Reduction
Agency and performed under U.S. Department of Energy Contract No. DE-AC02-05CH11231.
45
This work emerged from the IBPSA Project 1, an international project
conducted under the umbrella of the International Building Performance
Simulation Association (IBPSA). Project 1 will develop and demonstrate
a BIM/GIS and Modelica Framework for building and community energy
system design and operation.
References
[1] H. Qian, T. Miao, L. Liu, X. Zheng, D. Luo, and Y. Li. Indoor
Transmission of SARS-CoV-2. Indoor Air, 31(3):639–645, 2021.
[2] F.R. Lendacki, R.A. Teran, S. Gretsch, M.J. Fricchione, and J.L.
Kerins. COVID-19 Outbreak Among Attendees of an Exercise Facility
— Chicago, Illinois, August–September 2020. Morbidity and Mortality
Weekly Report, 70(9):321–325, 2021.
[3] J. Lu, J. Gu, K. Li, C. Xu, W. Su, Z. Lai, D. Zhou, C. Yu, B. Xu,
and Z. Yang. COVID-19 Outbreak Associated with Air Conditioning
in Restaurant, Guangzhou, China, 2020. Emerging Infectious Diseases,
26(7):1628–1631, 2020.
46
[4] ASHRAE Epidemic Task Force. Core Recommendations for Reducing
Airborne Infectious Aerosol Exposure. ASHRAE, 2021.
[5] J. Zhang, D. Huntley, A. Fox, B. Gerhardt, A. Vatine, and J. Cherne.
Study of Viral Filtration Performance of Residential HVAC Filters.
ASHRAE Journal, 62(8):26–32, 2020.
[6] M. Zaatari, A. Novoselac, and J. Siegel. The relationship between filter
pressure drop, indoor air quality, and energy consumption in rooftop
HVAC units. Building and Environment, 73:151–161, 2014.
[7] H.R.R. Santos and V.M.S. Leal. Energy vs. ventilation rate in buildings:
A comprehensive scenario-based assessment in the European context.
Energy and Buildings, 54:111–121, 2012.
[8] M. Zaatari, A. Novoselac, and J. Siegel. Impact of ventilation and
filtration strategies on energy consumption and exposures in retail
stores. Building and Environment, 100:186–196, 2016.
[9] T. Ben-David and M.S. Waring. Interplay of ventilation and filtration:
Differential analysis of cost function combining energy use and indoor
exposure to PM2.5 and ozone. Building and Environment, 128:320–335,
2018.
47
[10] P. Azimi and B. Stephens. HVAC filtration for controlling infectious
airborne disease transmission in indoor environments: Predicting
risk reductions and operational costs. Building and Environment,
70:150–160, 2013.
[11] H. Bohanon and M. Zaatari. Effect of Ventilation And Filtration on
Viral Infection in Residences. ASHRAE Journal, 62(12):38–45, 2020.
[12] L.F. Pease, N. Wang, T.I. Salsbury, R.M. Underhill, J.E. Flaherty,
A. Vlachokostas, G. Kulkarni, and D.P. James. Investigation of potential
aerosol transmission and infectivity of SARS-CoV-2 through central
ventilation systems. Building and Environment, 197:107633, 2021.
[13] Y. Fu, W. Zuo, M. Wetter, J.W. VanGilder, and P. Yang.
Equation-based object-oriented modeling and simulation of data center
cooling systems. Energy and Buildings, 198:503–519, 2019.
[14] S. Huang, W. Zuo, D. Vrabie, and R. Xu. Modelica-based system
modeling for studying control-related faults in chiller plants and boiler
plants serving large office buildings. Journal of Building Engineering,
44:102654, 2021.
[15] W. Tian, X. Han, W. Zuo, Q. Wang, Y. Fu, and M. Jin. An optimization
48
platform based on coupled indoor environment and HVAC simulation
and its application in optimal thermostat placement. Energy and
Buildings, 199:342–351, 2019.
[16] M. Wetter, W. Zuo, T.S. Nouidui, and X. Pang. Modelica Buildings
library. Journal of Building Performance Simulation, 7(4):253–270,
2014.
[17] M. Wetter, M. Bonvini, T.S. Nouidui, W. Tian, and W. Zuo.
MODELICA BUILDINGS LIBRARY 2.0. In Proceedings of BS2015:
14th Conference of International Building Performance Simulation
Association, 2015.
[18] M. Guo, P. Xu, T. Xiao, R. He, M. Dai, and S.L. Miller. Review and
comparison of HVAC operation guidelines in different countries during
the COVID-19 pandemic. Building and Environment, 187:107368, 2021.
[19] ASHRAE. Standard 52.2. Method of Testing General Ventilation
Air-Cleaning Devices for Removal Efficiency by Particle Size. American
Society of Heating, Refrigerating and Air-Conditioning Engineers, 2017.
[20] R.W. Fox. Air cleaners: a review. Journal of allergy and clinical
immunology, 94(2):413–416, 1994.
49
[21] Buildings.Examples.VAVReheat. https://simulationresearch.lbl.gov/
modelica/releases/v5.0.0/help/Buildings Examples VAVReheat.html#
Buildings.Examples.VAVReheat, 2013.
[22] Department of Energy. Commercial Reference Buildings. https://www.
energy.gov/eere/buildings/commercial-reference-buildings.
[23] ASHRAE. Standard 62.1 Ventilation for Acceptable Indoor Air Quality.
American Society of Heating, Refrigerating and Air-Conditioning
Engineers, 2015.
[24] M. Wetter, J. Hu, M. Grahovac, B. Eubanks, and P. Haves.
OpenBuildingControl: Modeling Feedback Control as a Step Towards
Formal Design, Specification, Deployment and Verification of Building
Control Sequences. In Proceedings of 2018 Building Performance
Modeling Conference and SimBuild, 2018.
[25] T. Xia and C. Chen. Evolution of pressure drop across electrospun
nanofiber filters clogged by solid particles and its influence on indoor
particulate air pollution control. Journal of Hazardous Materials,
402:123479, 2021.
50
[26] Dwyer. MERV 10 Pleated Filters. https://www.dwyer-inst.com/PDF
files/Priced/DF10 cat.pdf.
[27] Dwyer. MERV 13 Pleated Filters. https://www.dwyer-inst.com/PDF
files/Priced/DF13 cat.pdf.
[28] G. Buonanno, L. Stabile, and L. Morawska. Estimation of airborne
viral emission: Quanta emission rate of SARS-CoV-2 for infection risk
assessment. Environment International, 141:105794, 2020.
[29] G. Buonanno, L. Morawska, and L. Stabile. Quantitative assessment of
the risk of airborne transmission of SARS-CoV-2 infection: prospective
and retrospective applications. Environment international, 145:106112,
2020.
[30] Energy Star. Source Energy. Energy Star Portfolio Manager, pages
1–19, 2020.
[31] U.S. Energy Information Administration. Colorado Net Electricity
Generation by Source. https://www.eia.gov/state/?sid=CO#tabs-4,
2021.
[32] L.N. Troup, D.J. Fannon, and M.J. Eckelman. Spatio-temporal changes
51
among site-to-source conversion factors for building energy modeling.
Energy and Buildings, 213:109832, 2020.
[33] X. Zhang, N.P. Myhrvold, and K. Caldeira. Key factors for assessing
climate benefits of natural gas versus coal electricity generation.
Environmental Research Letters, 9(11):114022, 2014.
52
