Building and Environment: Tom Ben-David, Michael S. Waring

Building and Environment 104 (2016) 320e336
Contents lists available at ScienceDirect
Building and Environment

journal homepage: www.elsevier.com/locate/buildenv
Impact of natural versus mechanical ventilation on simulated indoor

air quality and energy consumption in ofces in fourteen U.S. cities
Tom Ben-David, Michael S. Waring*
Drexel University, Civil, Architectural, and Environmental Engineering, 3141 Chestnut St., Philadelphia 19104, United States
a r t i c l e i n f o a b s t r a c t
Article history: This study simulated the impacts of natural versus mechanical ventilation in ofces on indoor con-
Received 17 February 2016 centrations of key pollutants, as well as energy usage. A typical ofce building was modeled in Ener-
Received in revised form gyPlus in fourteen U.S. cities to assess the energy use and airows delivered by an ideal variable air
16 April 2016
volume (VAV) system in a range of climates. Two mechanical ventilation strategies (minimum;
Accepted 3 May 2016
minimum economizer control) were modeled, as well as two analogous natural ventilation strategies,
Available online 7 May 2016
which used a fan-driven recirculation hybrid system to maintain setpoints if necessary. Outputted
hourly ventilation, recirculation, and inltration rates were used in an indoor air model with city-
Keywords:
Natural ventilation
specic outdoor monitoring data to compute indoor concentrations of carbon monoxide, carbon diox-
Particulate matter ide, formaldehyde, nitrogen dioxide, ozone, and ne particles (PM2.5). Natural ventilation decreased
Filtration energy use, due to a wider temperature setpoint band for natural ventilation scenarios and somewhat
Indoor outdoor (I/O) ratio lower fan energy use. Indoor concentrations and indoor/outdoor (I/O) ratios of all pollutants were
Mechanical economizing similar for analogous strategies, except PM2.5, which was reduced by ltration in the supply air. Median
PM2.5 I/O ratios were a factor of 1.2, 2.2, and 6.3 larger for natural versus mechanical ventilation stra-
tegies with MERV 8, 11, and 16 lters, respectively. The ltration impact was so strong that PM2.5 I/O
ratios differed little between mechanical minimum and economizing strategies, especially as lter ef-
ciency increased. These results can be used to understand tradeoffs of energy and indoor air pollution
trends of natural versus mechanical ventilation.
2016 Elsevier Ltd. All rights reserved.
1. Introduction maintain acceptable indoor air quality [13].

Standards-setting organizations such as ASHRAE or the Euro-
Buildings demand large amounts of resources. For instance, they pean Committee for Standardization (CEN) prescribe minimum
consume roughly 40% of the energy used by the U.S. and 24% by the ventilation rates in buildings. For instance, the minimum ventila-
world [1], the residential and commercial building sectors are tion rate in U.S. ofces at default occupant densities is 8.5 L/s/per-
responsible for 40% of the U.S. carbon dioxide (CO2) emissions [2], son [13]. Traditionally, fans in air handling units (AHU) in HVAC
and building occupants account for 10% of the U.S. water con- systems supply ventilation air mechanically, though a newer design
sumption [3]. Heating, ventilating, and air conditioning (HVAC) paradigm attempts to use natural ventilation to supply air without
systems are used in buildings to maintain desired indoor condi- fans and instead by using wind-driven ows at the building faade
tions, and these consume two-thirds of the site energy for resi- [14]. This shift has occurred because natural ventilation can save
dential, commercial, and public buildings [4] in the developed energy by eliminating the need for fans, by sometimes supplying
world. Ventilation is the intentional exchange of outdoor air with larger airow rates than in mechanically ventilated spaces, and by
the indoor air, and conditioning ventilation air to the appropriate allowing larger allowable ranges of temperature bands in which
thermal state can potentially amount to about half of building occupants are comfortable [15]. Of course, natural ventilation alone
space-conditioning energy [4]. Ventilation rates may be increased is not feasible in all climates or at all times of year. However, hybrid
or optimized [5] to enhance occupant productivity [6,7], reduce sick approaches are also possible which use a mechanical recirculation
building syndrome [8e10] and absenteeism [10e12], and to system to complement natural ventilation to meet loads or which
may employ additional operational strategies such as thermal night
ushing [16].
* Corresponding author.
http://dx.doi.org/10.1016/j.buildenv.2016.05.007
0360-1323/ 2016 Elsevier Ltd. All rights reserved.
T. Ben-David, M.S. Waring / Building and Environment 104 (2016) 320e336 321
Though standards such as ASHRAE 62.1e2013 try to safeguard pooled analysis, the pollutant indoor/outdoor (I/O) ratio was our
indoor air quality (IAQ) with ventilation, its actual impact on IAQ is primary parameter of judgment of ventilation strategy impacts,
more complex. By its act of bringing in more outdoor air, ventilation since it is independent of city-to-city pollution variation, though
has the potential to increase indoor concentrations of outdoor summary statistics of absolute concentration distributions in each
originating pollutants, while decreasing indoor emitted ones city are provided as well. The simulation framework is generally
[17e19]. In ofces, for example, important outdoor sourced pol- introduced in this sub-section, and specic details are given
lutants can include particulate matter (PM), ozone (O3), and ni- afterward.
trogen dioxide (NO2) [20e23], and important indoor emitted A complete simulation herein consisted of a coupled building
pollutants include volatile organic compounds (VOCs) and carbon energy model using U.S. Department of Energy (DOE) EnergyPlus
dioxide (CO2) [18,24]. Since Americans spend ~87% of their time in Energy Simulation Software and an indoor concentration model
buildings [25], indoor exposure to these pollutants can be a using standard, well-mixed mass balances formulated for ofces.
dominant route [21,26e29]. This exposure is important, as studies EnergyPlus is a validated, physics-based, research-grade energy
suggest that exposures to PM [30e36], ozone [37,38], NO2 [39,40] analysis and thermal load modeling program [50,51] that, for a
and certain VOCs [40,41] are associated with increased morbidity particular building and climate, calculates the year-long transient
and mortality. heating and cooling necessary to maintain thermal and ventilation
There is one major distinction between the impacts of natural setpoints, as well as operational states of the HVAC equipment.
versus mechanical ventilation on indoor exposures. Namely, me- First, we used EnergyPlus to model a standard ofce building with
chanically ventilated buildings have the potential to reduce indoor the four ventilation strategies to quantify hourly energy usage and
PM concentrations in a way naturally ventilated spaces cannot, air exchange rates (AER). Then, along with hourly ambient pollution
even at high ventilation rates, since those buildings near- data from 2013 or prior [52], those AERs were used in mass bal-
universally employ PM lters in the supply airstream [13], ances with other typical parameters to predict the yearlong, hourly
through which mechanically ventilated air passes. For instance, indoor concentrations and I/O ratios of pollutants of interest (e.g.,
Quang et al. [22] demonstrated in three mechanically ventilated as in Rackes and Waring [18,24]).
ofces that lters were efcacious at controlling indoor concen- For the modeled ofce in each city, two analogous mechanical
trations of PM of outdoor origin, measuring efciencies of 26e47% and two natural ventilation strategies were simulated, which were
for ne and ultrane particles. Modeling suggests the same chosen based on adherence to current standards [13] or charac-
conclusion for a variety of building types [24,42e48]. Since non- teristic operational regimes, as well as to facilitate comparisons.
hybrid natural ventilation supplies air only through the building The two modeled mechanical ventilation strategies are frequently
envelope, rather than through an AHU, the supply air cannot be implemented in real buildings, and they were:
ltered before it reaches the building occupants. As a result, one
study predicted that locating a naturally ventilated building next to Mech_min: mechanical ventilation minimum strategy, for which
a roadway may lead to indoor PM concentrations near those of the ventilation rate was constant at the minimum ofce rate
outdoors or up to ~20% higher due to the lack of ltration [49]. As (2.5 L/s/person plus 0.3 L/s/m2) according to ASHRAE Standard
such, the concentration of and exposure to indoor PM of outdoor 62.1e2013 [13].
origin may be greater in naturally versus mechanically ventilated Mech_econ: mechanical ventilation economizing strategy, for
spaces. which more outdoor air is introduced into the indoor space
To greater understand this impact, as well as the impact of during times when it is thermodynamically favorable (i.e., free
natural versus mechanical ventilation on indoor pollutant con- cooling), but the ASHRAE Standard 62.1e2013 minimum rate is
centrations, this work explored the differences in indoor concen- the operational state otherwise.
trations of PM and other important IAQ indicators within ofces
with mechanical versus natural ventilation, as well as energy use The two natural ventilation cases were chosen to mimic the
differences. Ofces were chosen because they are a typical building mechanical ventilation cases such that they were tuned to provide
type that is operationally amenable to both ventilation types. To nearly the same amount of ventilation air on average, but by using
assess the impact of these different ventilation paradigms on IAQ, wind-driven ow through the envelope rather than through the
we simulated the transient energy use and ventilation rates using a AHU. The natural ventilation strategies were:
building energy model of a typical ofce in 14 U.S. cities, separately
considering two natural and two mechanical ventilation strategies. Nat_min: natural ventilation minimum strategy, which had rates
Outputted building airow rates were used with outdoor moni- similar to Mech_min.
toring data in mass balances to predict the indoor concentrations of Nat_econ: natural ventilation economizing strategy, which had
the indicator pollutants. These results were analyzed to allow more rates similar to Mech_econ.
informed decision-making about the impacts of each ventilation
type on IAQ and energy in ofces. Strictly achieving natural ventilation ows at this level of control
is currently operationally infeasible in buildings. However, the
2. Methodology strategies modeled herein were designed to facilitate direct com-
parison of the energy consumption and indoor air concentrations
2.1. Simulation overview resulting from the ideal implementation of these mechanical and
natural ventilation analogues that exist at the ventilation airow
Annual simulations at an hourly timescale were conducted to boundaries of low, minimum ow (Mech_min and Nat_min) and
assess the impacts of four ventilation strategiesdtwo mechanical high, economizing ow (Mech_econ and Nat_econ). Of course,
and two naturaldon energy consumption and IAQ in a typical of- natural ventilation is not ideal in every climate from a thermal
ce in 14 representative locations, using their typical meteorolog- perspective and sometimes may not be capable of meeting the
ical years (TMY) and recent outdoor pollution data. Then, those thermal loads. During these times, a hybrid system was automati-
annual hourly transient results from simulations in the 14 cities cally employed in which a recirculating air system was used to
were pooled and analyzed as one combined dataset, and outcome provide conditioning to help meet the thermal loads. Though the
trends as functions of inuential variables were explored. For the primary purpose of this hybrid system was to condition the
322 T. Ben-David, M.S. Waring / Building and Environment 104 (2016) 320e336
recirculated air to meet thermal loads, that recirculation air passes variable air volume (VAV) system [50]. Due to the ideality of the
through an AHU and therefore a supply air PM lter. system, the loads were met with 100% efciency and the percent of
outdoor air in the supply airstream was constrained to meet the
2.2. Typical ofce building locations, properties, and systems desired ventilation rate for each ventilation strategy. Although
practically unrealistic, this system is designed to capture the dif-
To select the typical size of the modeled ofce building for this ference between mechanically versus naturally ventilated build-
study, the 2012 Commercial Buildings Energy Consumption Survey ings for two similar cases for each ventilation type. EnergyPlus
(CBECS) was consulted to estimate that the median ofce building performed the sizing calculations in order to determine the cooling
has a oor area of ~450e500 m2 per oor and is two stories in and heating capacities as well as the design airow rates of the
height [53]. However, to be able to conduct a oor-by-oor analysis system. Since no specic equipment of particular efciency was
in future work, a three-story building was modeled instead, and the simulated, the output energy consumption was measured by units
oor-to-oor height was set at 4.0 m [54]. To mimic a building of annual energy per square meter from generic District Cooling
shape that would be well designed to maximize natural ventilation, and District Heating [51]. For the natural ventilation cases, the
guidance from ASHRAE Research Project 448 [55] indicated that the HVAC system operation was only required when natural ventilation
optimal aspect ratio of the building should be 0.6, so the footprint was insufcient to meet the buildings temperature setpoints. In
dimension of the building was designed as 17.0 28.5 m. these cases, this hybrid system conditioned recirculated air only, as
For the simulated locations, the International Energy Conser- it lacked an outdoor air intake. The system and the ventilation
vation Code (IECC) climate regions map was used to select one airows were only operational during occupied hours, dened as
representative city for different continental U.S. climate regions from 7:00 a.m. to 7:00 p.m. on weekdays (Monday to Friday).
[56,57]. The regions are divided into climate zones 1 through 7,
ranging from very hot to very cold, respectively, with three allow-
able moisture regimes of humid (A), dry (B), and marine (C). Table 1 2.3. Ventilation and inltration ow modeling
summarizes the different cities chosen for the 14 considered
climate zones. For each, weather data collected at the nearest The minimum mechanical ventilation rate was set based on
airport to each location [58] was used for energy modeling, along ASHRAE Standard 62.1e2013, which requires 2.5 L/s per occupant
with hourly outdoor pollutant concentration data from the U.S. plus an additional 0.3 L/s per m2 of occupied oor space [13]. At the
Environmental Protection Agency (EPA). For each location, the default ofce occupant density of 5 occupants per 100 m2, the
mean wind speed and resultant direction were culled from the standard thus calls for a minimum ventilation rate of 8.5 L/s/person
weather data, which are also in Table 1, with the wind direction or ~206 L/s/oor, which equates to a ventilation AER 0.383 h1.
measured clockwise from true north. Baseline inltration was estimated at an AER 0.0418 h1, which
Fig. 1(a) illustrates a rendering of the building model geometry was the value derived using ASHRAE Standard 90.1e2013 and
created in Google SketchUp. In each location-specic simulation, considering an AER 0.167 h1 when the building was not pres-
the building was oriented with its long face normal to the resultant surized by the mechanical system and then reduced by a factor of
windward direction, as shown in Fig. 1(b). The building windows 0.25 when the mechanical system is active to account for pres-
were distributed evenly to attain ~14% fenestration. The default surization impacts of unbalanced outdoor and exhaust airows
envelope constructions provided by the U.S. Department of Ener- [56]. For natural ventilation cases, inltration was ignored during
gys OpenStudio [59] were used, and default building schedules times when ventilation was active (described below), assuming air
were maintained as well, which are shown in Fig. 2. The insulation favors to travel through natural ventilation paths rather than
thickness was varied to achieve desired wall and roof R-values, through envelope leaks, and set to AER 0.167 h1 otherwise.
listed in Table 1, which were chosen to adhere to climate specic Both natural ventilation and inltration ows are affected by the
requirements dictated by ASHRAE Standard 90.1e2013 [56]. The dynamic temperature and pressure differences of the outdoor and
lighting density of 9 W/m2 was also based on ASHRAE Standard the indoor environments. Therefore, modeling these ows required
90.1e2013 [56], and an equipment density of 15 W/m2 was based a dedicated scheme to simulate their hourly transient behavior that
on Wilkins and Hosni [60]. The occupant density was set at 5 would lead to the desired annual average behavior. We modeled
persons per 100 m2, according to default assumptions in ASHRAE zonal natural ventilation and inltration using an EnergyPlus
Standard 62.1e2013 [13]. formulation that incorporates baseline ow rates with model co-
The thermal loads of the building were met using an ideal load, efcients, as shown in Equation (1) [50]:
Table 1
Simulated representative city in each climate zone, along with the building overall thermal resistive values and wind information for each location.
Climate zone City Wall R-value (m2 K/W) Roof R-value (m2 K/W) Mean wind speed (m/s) Resultant wind direction ( )a
1A Miami, FL 2.3 5.3 3.9 105

2A Houston, TX 2.3 5.3 4.6 138
2B Phoenix, AZ 2.3 5.3 3.0 149
3A Atlanta, GA 2.3 5.3 4.0 310
3B El Paso, TX 2.3 5.3 4.0 234
3C Los Angeles, CA 2.3 5.3 3.3 247
4A Philadelphia, PA 2.3 6.7 4.3 275
4B Albuquerque, NM 2.3 6.7 3.9 181
4C Seattle, WA 3.5 6.7 4.1 193
5A Boston, MA 3.5 6.7 5.9 283
5B Salt Lake City, UT 3.5 6.7 4.0 174
6A Milwaukee, WI 3.5 8.6 5.1 255
6B Billings, MT 3.5 8.6 5.5 262
7B Fargo, ND 3.7 8.6 5.8 240
a
Wind direction measured clockwise from true north.
Fig. 1. (a) Model visualization of standard ofce building, and (b) example of building orientation with respect to resultant wind direction.
1 Lighting strategy, we set them equal to each other. We assumed that at

0.9 Occupancy TOA 18.3 C, occupants are expected to start closing windows to
0.8 Equipment some degree. Therefore, for location j during month m, the venti-
lation schedule factor, Fj,m, was expressed by solving for F in
0.7
Fraction of total
Equation (1), so that:

0.6
0.5 8
>
> 1
0.4 >
< TOA;mean;j;m 18:3
1 b18:3 TOA;mean;j;m
0.3 Fj;m (3)
0.2
>
>
>
:
0.1 1 TOA;mean;j;m > 18:3
0
0:00 4:00 8:00 12:00 16:00 20:00 0:00 where TOA,mean,j,m ( C) is the mean monthly outdoor air tempera-
ture in location j during month m, rounded up to the nearest tenth.
Time of day (hour)
This scheme resulted in the Nat_min strategy nearly mimicking
Fig. 2. Internal load schedules during a typical workday used in energy modeling. ventilation rates for Mech_min on a monthly mean basis. Similarly
to the mechanical ventilation, natural ventilation was set to zero
during unoccupied hours, as if windows or other means by which
natural ventilation is delivered were kept closed during unoccupied
Q Q0 Fa bjTz TOA j cvwind (1) times.
For strategy Mech_econ, a temperature-based economizer was
where Q (m3/s) is the airow rate for natural ventilation or inl- applied during cooling cases when the outdoor air temperature was
tration; Q0 (m3/s) is the design airow rate; F is a fraction set to a low enough to provide free-cooling effectively, which was capped
predetermined schedule (described below); Tz ( C) and TOA ( C) are by the maximum supply airow that was automatically sized by
the zonal and outdoor dry bulb air temperatures respectively; vwind EnergyPlus, adding the potential for an additional ventilation
(m/s) is the wind speed; and a, b, and c are constant coefcients, for airow of 3.15 h1, so that the year-long mean ventilation AER could
which we used the Building Load Analysis and System Thermody- attain up to 0.383 3.15 3.53 h1. The actual ventilation delivered
namics (BLAST) default coefcients of 0.6060, 0.03636, and 0.1177, by the economizer strategies was automatically and dynamically
respectively [50]. These coefcients are based on a typical summer employed by EnergyPlus and responded to the real time building
condition where the outdoor temperature is equal to the zone loads. To mimic the economizer airows of Mech_econ for the
temperature and the wind speed is 3.35 m/s. The c coefcients were Nat_econ strategy, additional natural ventilation was added
changed to be location-specic to reect the actual mean summer dynamically to the design ventilation rate, Q0 such that Q0 was
wind speed for each city, according to Equation (2): equal to the ventilation rate under Mech_econ. The ventilation rate
vmean;j under Nat_econ was still subject to dynamic uctuation based on
cj 0:1177 (2) the outdoor temperature variation and wind velocity.
3:35
where cj is the c coefcient for location j; and vmean,j (m/s) is the

mean wind speed in location j when the outdoor air temperature is 2.4. Temperature setpoints
between 18 C and 28 C.
The schedule (F) for the natural ventilation cases was designed We determined indoor temperature setpoints, TSP ( C), as fol-
to emulate a scenario in which windows would be opened or closed lows. For the mechanical ventilation cases, the TSP for heating and
dynamically based on the outdoor air temperature. We therefore cooling were set at 21 C and 24 C, respectively. However, it has
designed a window schedule that allowed a uctuation in natural been demonstrated that people are comfortable in a wider band of
ventilation rates due to temperature changes to achieve a desired setpoints in naturally versus mechanically ventilated buildings. In
yearlong mean value similar to the mechanical ventilation cases. fact, de Dear and Brager [15] offered a functional form that relates
Since we want to maintain Q as close to Q0 as possible so that the the acceptable TSP as a function of the outdoor air temperature (TOA)
natural ventilation strategy mimics the mechanical ventilation in naturally ventilated buildings:
number of occupants, respectively.

TSP 17:8 0:31TOA (4)
For natural ventilation cases, we used this formulation for the 2.5.1. Carbon dioxide and carbon monoxide
heating setpoint within the boundaries of 15 C and 21 C and for In our simulated ofce, CO2 was generated by the respiration of
the cooling setpoint within the boundaries of 24 C and 30 C. human occupants and was also introduced by outdoor-to-indoor
transport. This pollutant is used as an indicator of IAQ and
2.5. IAQ model and pollutant information adequate ventilation [13]. Newer research also suggests it directly
impairs decision-making skills [7,64], though this impact has not
For each pollutant i, the rate of change of the indoor concen- been conrmed outside of laboratory experiments. Within the
tration Ci (ppm, ppb, or mg/m3) is a function of the sources, Si, (ppm/ zone, the per occupant respiratory emission rate of CO2, ECO2, was
h, ppb/h, or mg/m3/h) and the losses, Li (h1), of the pollutant in that 18,530 ppm-m3/h/occ (~33.3 g/h/occ), and the outdoor concentra-
environment, and can be represented as in Equation (5) tion, CCO2,out (ppm), was constant at 450 ppm to represent that of a
[17,18,24,43,46,61]: typical urban environment [24]. Regarding CO, its only source in the
ofce environment was outdoor-to-indoor transport, and the out-
dCi door concentration of CO, CCO,out (ppm), was from the hourly EPA
Si L i C i (5)
dt data. Both CO2 and CO are treated as non-reactive and non-sorbing,
so their losses were owing to ventilation and inltration air ex-
Inputs to the IAQ model, which included outdoor concentrations
change only.
for the 14 locations reported by the U.S. EPA [52] and ow rates and
schedules outputted by the EnergyPlus simulations, were all
sampled hourly over a full calendar year. We assumed that within 2.5.2. Nitrogen dioxide and ozone
each hour the sources and losses of pollutant i remained constant, Both NO2 and O3 exist meaningfully in the ambient urban at-
which allowed for an analytical solution of the transient process at mosphere, and both can be emitted indoors. For example, NO2 is
any hour k with state knowledge of the previous hour k1, as: from combustion processes such as gas-stove use [66], and O3 may
be from emissions of copiers or laser printers [67]. However, we
Si;k assumed that these processes were negligible in our modeled in-
Ci;k Ci;k1 eLi;k tk tk1 1 eLi;k tk tk1 (6) door environment and treat outdoor-to-indoor transport as our
Li;k
only sources of these pollutants. Ozone is a dominant oxidizer in-
where any initial value (Ci,0) was calculated as the steady state doors, though both NO2 and O3 can meaningfully impact indoor
solution of the initial hour of measurements. These initial values chemistry [23,65]. In our model, both NO2 and O3 have losses due to
were necessary to start the annual simulation, as well as to restart surface reactions, which are termed bNO2 (h1) and bO3 (h1),
the simulation process if there was a lapse in the reported hourly respectively, and their values were set at 1.18 h1 and 2.80 h1,
outdoor environmental data. respectively [24,65]. Ozone was also modeled as being reduced by
This study considered six different pollutants: (1) carbon diox- chemical reactions when inltration air passes through the build-
ide (CO2); (2) carbon monoxide (CO); (3) nitrogen dioxide (NO2); ing envelope or when recirculation and ventilation air pass through
(4) ne particles (PM2.5); (5) ozone (O3); and (6) formaldehyde a lter in the mechanical system. These phenomena were quanti-
(HCHO). These pollutants were chosen for various reasons. Notably, ed, respectively, by a penetration factor, pO3 0.78 [68], and by a
PM2.5, O3, NO2, and CO are U.S. EPA criteria pollutants. Human lter removal term hr,O3 or hv,O3 0.080 [69]. We set the ventilation
exposure to them is associated with health impacts, such as pre- efciency term to zero, i.e., hv,O3 0, for the natural ventilation
mature morbidity and mortality and occurrences of asthma cases.
[30e36,38,62]. Additionally, HCHO is a known human carcinogen
[63]. Also, exposure to CO2 and/or VOCs (for which HCHO is a good 2.5.3. Formaldehyde
representative) has been correlated to loss of productivity in Formaldehyde (HCHO) and other VOCs are signicant indoor
simulated task environments [7,64]. All but CO2 may be meaningful pollutants [40]. VOCs are emitted indoors, so their indoor concen-
contributors to indoor chemistry mechanisms [65]. Finally, from trations are often higher than their outdoor concentrations [24,70].
the standpoint of comparing ventilation strategies, PM2.5 is Moreover, some VOCs may be emitted indoors as functions of the
particularly important to include since it is reduced indoors by indoor concentration (which is heavily inuenced by the AER) from
ltration, which by denition does not occur in natural ventilation both instantaneous [71] and time-averaged perspectives [18]. As
strategies except for when hybrid operation is necessary. The such, Rackes and Waring [18] used VOC concentration and AER
pollutant sources and losses are illustrated in Table 2 [24], with data, which were collected in ~100 buildings as part of the EPA
explanations of the model parameters and terms in the following Building Assessment Survey and Evaluation (BASE) study to sta-
subsections. Of the symbols in Table 2 not yet dened, the lv, li, and tistically model whole building, effective emission rates as a func-
lr (h1) denote the ventilation, inltration, and recirculation AERs, tion of the AER [18,54], similarly to in Ref. [71]. Following that
respectively, while Vz (m3) and Pz (occ) are the zonal volume and model, the HCHO emission rate per oor area, Ez,HCHO (ppb-m3/h-
Table 2
Sources and losses of each indoor pollutant. Symbols are dened in text.
Pollutant Sources Losses
Carbon dioxide (CO2) Ep;CO2

CCO2 ;out lv li lv li
Vz Pz
Carbon monoxide (CO) CCO;out lv li lv li
Nitrogen dioxide (NO2) CNO2 ;out lv li lv li bNO2
Ozone (O3) CO3 ;out lv 1 hv;O3 CO3 ;out pO3 li lr hr;O3 lv li bO3
Formaldehyde (HCHO) Ez;HCHO

Ep;HCHO
CHCHO;out lv li lv li
Vz Az Vz Pz
Fine particles (PM2.5) CPM2:5 ;out lv 1 hv;PM2:5 CPM2:5 ;out pPM2:5 li lr hr;PM2:5 bPM2:5 lv li
m2), was: strategies over all locations, and they had R2 0.98. Based on the
CDFs and R2 values, we conclude that the natural ventilation
lv li Ceq Soth models emulated their mechanical ventilation analogues to a
Ez;HCHO hz kL (8)
lv li kL reasonable extent to allow comparisons, while still including the
variability of wind and natural ventilation.
where hz (m) is the zone height; Soth (ppb/h) is sources of per- We also assessed ventilation AERs by strategy at each location
person emissions (Ep,HCHOPZ/VZ) and the outdoor concentration individually. For each ventilation strategy, the median mean daily
(CHCHO,out (lvli)); Ceq (ppb) is a conceptual equilibrium concen- ventilation AER is presented with the 5th and 95th percentiles in
tration at the material-air interface; and kL (h1) is a coupling time parentheses in Table 3. We also calculated the root mean squared
constant between the material surface and room air. The outdoor error (RMSE) of the mean daily ventilation AERs of Mech_min
concentration, CHCHO,out (ppb), and per-person emission rate, Ep,H- compared with those of Nat_min, as well as the correlation coef-
3
CHO (ppb-m /h-occ), were constrained at 2.20 ppb and 3.20 ppb- cient (corr) between the mean daily ventilation AER of the two
3
m /h-occ, respectively [18]. From Rackes and Waring [18], the economizing strategies at each location. Errors ranged from
HCHO emission parameters were Ceq 18.2 ppb and kL 1.6 h1. RMSE 0.049 to 0.078 h1 and correlation ranged from corr 0.81
Since the emission model was developed with empirical BASE data, to 0.99 for economizing analogues, so the natural ventilation stra-
which included ~100 buildings under numerous indoor psychro- tegies clearly mimicked the mechanical ventilation strategies in all
metric conditions, the inuence of the variability of the indoor air locations to a large degree with the ow models used, albeit with
temperature and RH on the HCHO emission rate was included slightly larger spans for natural ventilation cases due to the vari-
within its parameter tting but not modeled specically. To ability of outdoor temperature and wind and their impacts on
demonstrate the impact of the AER-dependent emissions model on airow.
HCHO, concentrations were simulated for an additional value of Since economizer operation is enthalpy-driven, we were inter-
kL 0.05 h1, which makes the emission rate modeled by Equation ested in observing how ventilation changed with the outdoor air
(8) AER-independent and near constant, as is most often assumed. conditions. Consequently, we plotted the mean monthly ventilation
For comparison, the Ceq for the AER-independent emission model AER against the mean monthly outdoor air temperature, as in
was adjust to equal 122 ppb, so that Ez,HCHO was the same for both Fig. 3b, pooling the modeling results for all 14 locations. For the two
emission models in Mech_min. economizing strategies, a similar peak in ventilation AER is clearly
observed during times of temperate outdoor air temperatures (e.g.,
2.5.4. Fine particulate matter means of 10e25 C), though the natural ventilation strategy ach-
PM2.5 is the only non-gaseous pollutant considered in this study, ieves slightly higher rates than the mechanical analogue. These
and it exists at many intensities outdoors and has strong impacts on higher ventilation AERs will affect both fan energy consumption for
human health. It exhibits size-resolved behavior, and assumed PM the mechanical case and pollutant indoor concentrations and I/O
size distributions can be combined with size resolved parameters ratios when they occur for both types of economizing strategies.
to estimated integrated parameters for the mass concentration. Results in Fig. 3b can be used to estimate ventilation behavior for
Relevant size resolved losses indoors include deposition to surfaces our modeled building for locations not in our modeling space.
and ltration, and the PM size distribution shape relates one to the
other. Using this relationship, Rackes and Waring [24] t a 3.2. Energy consumption
quadratic expression to estimate the integrated PM2.5 deposition
rate, bPM2.5 (h1), as a function of the nominal PM2.5 lter efciency Simulation results for ofce energy consumption for each
in the recirculation air stream, hr,PM2.5: ventilation strategy in each city of the 14 climate zones are pre-
sented in Table 4, separated into ideal cooling, heating, and fan
bPM2:5 0:171hr;PM2:5 2 0:1378hr;PM2:5 0:0918 (7) energy use. To generalize the energy results beyond the simulated
locations and view trends instead according to climate, the
In our study, three different lter efciencies (hr,PM2.5) were monthly cooling, heating, and fan energy consumption data for all
considered, characterized by the ASHRAE Minimum Efciency 14 locations were pooled and plotted in Fig. 4a against the mean
Reporting Value (MERV) of MERV 8, 11, and 16, which were assigned monthly outdoor air temperature. In addition to the nominal ideal
nominal removal efciencies for PM2.5 of 26.4%, 65.6%, and 96.3%, energy consumption values, Fig. 4def plots the energy savings
respectively [72]. For the mechanical ventilation cases, the venti- relative to the base case Mech_min for each of the three other
lation lter efciency hv,PM2.5 hr,PM2.5. For the natural ventilation ventilation strategies, since this relative comparison allows one to
cases, hv,PM2.5 0. Finally, PM2.5 can be ltered by the cracks in the better observe energy saving trends by strategy for different mean
building envelope when air inltrates into the building, and we outdoor conditions.
modeled this using a penetration factor of pPM2.5 0.975 [24]. Considering Mech_min as the baseline case, cooling energy was
universally reduced under the three other strategies, for reasons as
3. Results and discussion follows. Firstly, the mechanical economizing case, Mech_econ, and
its natural ventilation analogue, Nat_econ, often used free rather
3.1. Ventilation summary than coil-driven cooling during swing and winter seasons to meet
cooling loads. Moreover, the wider setpoint band of natural venti-
We compared the air exchange rate (AER) results over the lation strategies of Nat_min and Nat_econ yielded a load reduction.
ventilation strategies to conrm that the natural ventilation Since Nat_econ takes advantage of both of these opportunities, it
schemes mimicked their mechanical ventilation analogues, using universally exhibited the largest savings in cooling energy for all
the daily mean ventilation AER in each location for comparison, strategies.
therefore employing a large dataset for comparing ventilation Over all locations, heating energy consumption in Mech_econ
strategy analogues at a granular level. A cumulative distribution was identical to heating energy consumption in Mech_min. This
function (CDF) of the mean daily ventilation AERs over all locations equality is expected, since economizing only impacts the amount of
is shown in Fig. 3a. We also calculated the coefcient of determi- cooling and fan energy used. Conversely, the heating energy for the
nation (R2) between the analogue economizing ventilation two natural ventilation cases was typically reduced compared with
Fig. 3. (a) Cumulative distribution function of mean daily ventilation air exchange rate and (b) mean monthly ventilation air exchange rate (AER) as a function of mean monthly
outdoor air temperature under the four ventilation strategies.
Table 3
Median (5th, 95th) percentiles of day-mean ventilation air exchange rates (AER) and root mean square error (RMSE) between analogue minimum strategies and correlations
(corr) between analogue economizing strategies by location.
Climate zone City Minimum AER (h1) Economizing AER (h1)
Mech_min Nat_min RMSE Mech_econ Nat_econ corr
1A Miami, FL 0.39 (0.39, 0.39) 0.33 (0.30, 0.41) 0.064 0.40 (0.39, 2.3) 0.78 (0.34, 2.4) 0.81
2A Houston, TX 0.39 (0.38, 0.40) 0.36 (0.32, 0.53) 0.069 0.50 (0.39, 2.0) 0.80 (0.36, 2.1) 0.83
2B Phoenix, AZ 0.39 (0.39, 0.40) 0.38 (0.34, 0.50) 0.051 1.0 (0.39, 2.2) 0.90 (0.37, 1.9) 0.96
3A Atlanta, GA 0.39 (0.38, 0.39) 0.38 (0.31, 0.53) 0.078 0.62 (0.38, 2.2) 0.90 (0.40, 2.2) 0.81
3B El Paso, TX 0.39 (0.38, 0.39) 0.37 (0.32, 0.51) 0.065 1.0 (0.40, 2.6) 0.78 (0.41, 2.0) 0.89
3C Los Angeles, CA 0.39 (0.39, 0.39) 0.39 (0.32, 0.46) 0.045 1.4 (0.58, 2.7) 1.6 (0.68, 2.6) 0.93
4A Philadelphia, PA 0.39 (0.38, 0.39) 0.36 (0.31, 0.51) 0.066 0.40 (0.38, 2.2) 0.67 (0.34, 2.3) 0.92
4B Albuquerque, NM 0.39 (0.38, 0.39) 0.38 (0.32, 0.48) 0.055 1.1 (0.39, 2.4) 0.89 (0.41, 2.1) 0.97
4C Seattle, WA 0.39 (0.38, 0.39) 0.39 (0.32, 0.48) 0.049 0.46 (0.38, 2.0) 0.49 (0.33, 2.0) 0.99
5A Boston, MA 0.39 (0.38, 0.40) 0.36 (0.30, 0.48) 0.060 0.42 (0.38, 2.4) 0.49 (0.31, 2.3) 0.92
5B Salt Lake City, UT 0.39 (0.38, 0.39) 0.37 (0.32, 0.49) 0.054 0.74 (0.38, 2.4) 0.69 (0.36, 2.0) 0.98
6A Milwaukee, WI 0.39 (0.38, 0.40) 0.37 (0.29, 0.50) 0.067 0.43 (0.38, 2.5) 0.64 (0.31, 2.2) 0.81
6B Billings, MT 0.39 (0.38, 0.40) 0.38 (0.30, 0.51) 0.067 0.51 (0.38, 2.6) 0.58 (0.31, 2.0) 0.99
7B Fargo, ND 0.39 (0.38, 0.40) 0.35 (0.26, 0.50) 0.073 0.42 (0.38, 2.4) 0.51 (0.27, 2.0) 0.97
their mechanical analogues, due to the wider setpoint band. The Table 4 includes the percent of time when economizing was
Nat_econ heating energy use was always larger than Nat_min cases, utilized for the two relevant strategies, so that one may discern the
owing to the higher ventilation AERs. However, both natural differences for the natural over the mechanical ventilation opera-
ventilation strategies occasionally had larger heating energy use as tion. Moreover, we included in Table 4 the percent of time during
compared with their mechanical analogues for some outlying which the hybrid operation was employed by the natural ventila-
cases, as can be seen in Fig. 4 during warmer months that still tion cases (i.e., when natural ventilation alone could not main-
required some heating. These heating trends are a consequence of taining indoor air setpoints and a recirculating mechanical system
the natural ventilation airow model, which mimics mechanical was used to complement it), which is useful to interpret the impact
analogue ow rates but still has variation due to extreme outdoor of the hybrid recirculation-inuenced ltration on PM2.5 for Nat_-
conditions. Thus, the actual transient temperature and wind ve- min and Nat_econ cases. Relatedly, Fig. 5a plots a CDF of the mean
locities in each climate in the TMY datasets dictate the natural daily return air AER over all locations under each strategy, and
ventilation airows (see Equation (1)) that sometimes result in the Fig. 5b plots the mean monthly return AER as a function of the
introduction of larger amounts of colder air indoors during the mean monthly outdoor air temperature. Since the return air AER
heating season. demonstrates the total amount of airow that passes through the
Additionally, fan energy consumption is presented in Table 4 lterdfor mechanical strategies it approximates
and Fig. 4. When the economizer is used in Mech_econ, it supplies ventilation recirculation (i.e., supply) AERs and for natural
additional ventilation air mechanically, causing the fan to work ventilation strategies it equals the recirculation AER owing to the
more. This operation produces increases in fan energy consumption hybrid strategydit is therefore a direct indicator of the efcacy of
during temperate months, as seen in Fig. 5. On the other hand, for the lter on reducing indoor PM2.5 concentrations for each strategy.
Nat_econ, the additional ventilation air is not moved by fan power
and the free-cooling usage causes a thermal load reduction and 3.3. Indoor pollutant absolute concentrations
requires no hybrid fan energy during these temperate months. The
thermal load reduction due to the wider setpoint band also allows We simulated hourly indoor concentrations of the six pollutants
for further reduction in fan energy consumption under both natural over an entire year, but only those during occupied hours were
ventilation strategies. Generally, one can see increased savings in considered in this analysis, since they are the only ones that would
fan energy in the natural ventilation cases as the outdoor air tem- inuence occupant exposure. Individual pollutants exhibited
perature increases. That said, fan energy changes among strategies distinctive trends as functions of ventilation strategy or location,
are one order of magnitude smaller than savings generated by depending on their particular source and loss proles. Both CO2 and
heating or cooling energy reductions among ventilation strategies. HCHO had constant simulated outdoor concentrations across cities
Table 4
Annual energy summary by location and strategy, as well as percent of time for economizing (econ.) and hybrid system operation.
Climate zone City Energy consumption (kWh/m2) % Econ. % Hybrid
Strategy Cooling Heating Fan
1A Miami, FL Mech_min 144 4.85 102 1.55 e e

Nat_min 117 1.75 102 1.04 e 99%
Mech_econ 137 4.85 102 2.27 17% e
Nat_econ 107 2.48 102 0.843 30% 93%
2A Houston, TX Mech_min 105 1.55 1.08 e e

Nat_min 90.0 1.97 0.669 e 89%
Mech_econ 96.7 1.49 1.89 23% e
Nat_econ 86.0 9.22 0.484 32% 75%
2B Phoenix, AZ Mech_min 111 0.711 1.64 e e

Nat_min 92.2 0.316 1.10 e 94%
Mech_econ 92.6 0.711 3.09 35% e
Nat_econ 75.0 0.580 0.800 35% 84%
3A Atlanta, GA Mech_min 71.1 6.94 0.852 e e

Nat_min 60.4 4.38 0.552 e 85%
Mech_econ 56.9 6.94 1.67 27% e
Nat_econ 44.5 4.99 0.322 35% 68%
3B El Paso, TX Mech_min 84.3 2.16 1.42 e e

Nat_min 73.1 0.744 0.939 e 85%
Mech_econ 65.8 2.16 2.84 35% e
Nat_econ 57.6 0.872 0.640 30% 73%
3C Los Angeles, CA Mech_min 61.1 0.137 0.951 e e

Nat_min 60.5 0.119 0.619 e 88%
Mech_econ 24.3 0.137 3.07 63% e
Nat_econ 25.3 0.184 0.205 68% 45%
4A Philadelphia, PA Mech_min 51.5 18.4 0.658 e e

Nat_min 44.2 10.5 0.423 e 81%
Mech_econ 39.3 18.4 1.51 24% e
Nat_econ 31.0 11.2 0.264 28% 67%
4B Albuquerque, NM Mech_min 56.7 6.22 0.909 e e

Nat_min 51.0 1.87 0.562 e 83%
Mech_econ 35.9 6.22 2.44 40% e
Nat_econ 32.8 2.36 0.291 38% 68%
4C Seattle, WA Mech_min 22.9 13.3 0.384 e e

Nat_min 22.4 7.42 0.232 e 68%
Mech_econ 8.30 13.3 1.07 27% e
Nat_econ 7.60 7.95 7.66 102 29% 45%
5A Boston, MA Mech_min 38.2 21.4 0.640 e e

Nat_min 35.0 10.7 0.300 e 75%
Mech_econ 24.5 21.4 1.29 23% e
Nat_econ 20.8 11.5 0.151 26% 60%
5B Salt Lake City, UT Mech_min 49.6 13.5 0.836 e e

Nat_min 44.2 4.87 0.570 e 77%
Mech_econ 34.2 13.5 1.84 28% e
Nat_econ 30.1 6.37 0.329 27% 64%
6A Milwaukee, WI Mech_min 40.3 28.3 0.587 e e

Nat_min 36.6 14.2 0.338 e 77%
Mech_econ 24.9 28.3 1.42 26% e
Nat_econ 27.6 26.5 0.269 27% 65%
6B Billings, MT Mech_min 42.5 25.7 0.679 e e

Nat_min 39.2 13.8 0.563 e 73%
Mech_econ 26.6 25.7 1.85 27% e
Nat_econ 24.0 14.6 0.282 26% 61%
7B Fargo, ND Mech_min 41.2 47.9 0.631 e e

Nat_min 37.3 26.0 0.434 e 81%
Mech_econ 26.7 47.9 1.44 24% e
Nat_econ 23.0 27.3 0.217 26% 67%
(450 ppm and 2.2 ppb, respectively) and rather stable indoor is, Mech_min and Nat_min ventilation cases resulted in similar sets
emission rates, so factors affecting their indoor concentrations of indoor concentrations for both pollutants, and Mech_econ and
were the ventilation AER and the number of occupants. Owing to Nat_econ cases resulted in other similar sets of concentrations. For
these factors, variations in CO2 and HCHO concentrations were CO2 and HCHO indoor concentration distributions, medians and 5th
signicant when compared across certain ventilation strategies, but and 95th percentiles are presented in Table 5. Additionally, Fig. 6
city-to-city concentration variations were largely insignicant. That includes CDFs of HCHO indoor concentrations for representative
Fig. 4. (aec) Ideal system monthly energy consumption as a function of mean monthly outdoor air temperature under the four ventilation strategies, as well as (def) energy savings
compared to Mech_min.
Fig. 5. (a) Cumulative distribution function of mean daily recirculation air exchange rate and (b) mean monthly return air exchange rate (AER) as a function of mean monthly
outdoor air temperature under the four ventilation strategies.
Table 5 ASHRAE minimum rate or greater [13,54]. For the two minimum
Indoor CO2 and HCHO concentration distributions under four ventilation cases, rate cases, CO2 distributions had a median of ~810 ppm, which was
listing the median (5th and 95th) percentiles.
~175 ppm higher than their economizing counterparts. Though
Case CO2 (ppm) HCHOa (ppb) median CO2 concentrations were different for the minimum versus
kL 1.6 h1 kL 0.05 h1 economizing ventilation cases, all four ventilation strategies
resulted in similar CO2 ranges. On average, they all existed within a
Mech_min 802 (474, 925) 14.8 (14.7, 15.5) 14.9 (14.1, 19.9)
Nat_min 821 (478, 1040) 15.3 (14.5, 15.9) 17.4 (13.5, 22.1) 90% condence interval (CI) of ~465e965 ppm, although the two
Mech_econ 628 (459, 905) 14.5 (7.31, 15.2) 13.8 (3.94, 18.2) natural ventilation CO2 distributions exhibited slightly longer up-
Nat_econ 640 (458, 989) 14.1 (7.65, 15.6) 13.4 (4.11, 19.4) per tails due to low ventilation AERs driven by variability in the
a
HCHO was computed with two emission models, one which assumed emission TMY ambient temperatures and wind velocities over the modeled
rates dependent on the ventilation AER (column labeled kL 1.6 h1) and one which locations.
assumed independent emission rates (column labeled kL 0.05 h1). The AER- The best estimate HCHO concentration distributions are those
dependent model HCHO results comprise our best estimate. See text for details
predicted with the AER-dependent model at kL 1.6 h1 [18], and
about these HCHO emission models, which are from Rackes and Waring [18].
simulated values are similar to those measured in ofces [54]. The
HCHO distributions predicted using the AER-independent model
location 4A: Philadelphia, PA, with separate plots for the two kL (kL 0.05 h1) demonstrate the concentration outcomes (and
values. tendency to error, we believe) when assuming HCHO emissions are
These CO2 distributions reect the magnitude one would expect AER-independent and constant. Indeed, HCHO concentration dis-
for an ofce at default occupant densities that was ventilated at the tributions were more different across the kL values than across
Fig. 6. Cumulative distribution functions of indoor HCHO concentrations under four ventilation strategies, simulated in Philadelphia, PA (climate zone 4A) for (a) the best estimate
AER-dependent emissions model with kL 1.6 h1 and (b) the AER-independent emissions model with kL 0.05 h1. See text for details about HCHO emission models from Rackes
and Waring [18].
ventilation strategies, showing the importance of appropriately affected by the economizing process of Mech_econ and Nat_econ.
modeling the variability of the emission rate with AER when esti- However, being of outdoor origin only, indoor NO2 and O3 con-
mating indoor VOCs. The AER-independent concentrations were centrations also varied strongly by location. Distribution values for
much more variant than AER-dependent ones because the higher the outdoor concentrations, as well as indoor concentrations of NO2
kL 1.6 h1 of the AER-dependent model drives the indoor HCHO and O3 by ventilation strategy, are listed by location in Table 7. At
concentration closer to Ceq and lessens effects of ventilation (See the medians, the economizing strategies resulted in concentrations
Rackes and Waring [18] for more discussion on this phenomenon.) that were a factor of ~1.5 larger than their minimum ventilation
Also, the HCHO distributions had similar median and somewhat counterparts for both NO2 and O3. Finally, because O3 was removed
similar upper tails over the four ventilation cases for each kL, by the HVAC lter at a low efciency, this removal mechanism did
though the economizing analogues differed from the minimum not manifest itself meaningfully in differences between mechanical
ventilation cases by having much longer lower tails, because these and natural ventilation strategy analogues.
lowest concentrations are realized at the highest AERs resulting For PM2.5, the location-specic ambient and indoor concentra-
from economizer operation. tions distributions are also listed in Table 7 and are typical of those
For CO, since its only source is outdoor air and it is treated as measured in U.S. ofces [54]. Additionally, Fig. 7 shows CDFs of the
non-reactive and non-sorptive in the simulations, indoor concen- PM2.5 concentration in representative climate zone 4A: Philadel-
trations were nearly identical to outdoor ones. Any small differ- phia, PA for the three different MERV lters. Notably, PM2.5 was the
ences were only due to lag and storage effects of the outdoor air only pollutant in the simulations strongly inuenced by whether
exchanging with that within the indoor space. Therefore, CO con- ventilation was natural versus mechanical in nature, because
centrations were effectively independent of the ventilation strategy aerosol ltration is made possible by mechanical airow. In fact, the
used. However, CO distributions did vary by location due to ltration impact was so strong that results of Mech_min compared
different ambient concentrations. The location-specic indoor CO with Mech_econ were largely indistinguishable, most especially as
concentration distributions are summarized in Table 6. the ltration efciency increased from MERV 8 to 11 to 16.
Since NO2 and O3 were of outdoor origin only and are reactive, Moreover, though ltration can have an impact on the PM2.5
both had much smaller indoor concentrations than their respective concentration when the natural ventilation is employed owing to
outdoor concentration counterparts. The strength of that concen- the hybrid operation of using recirculated air to meet thermal loads
tration reduction varied over the four ventilation strategies. Simi- when necessary, this effect is still weak enough that Mech_econ
larly to CO2 and HCHO results, the NO2 and O3 were strongly always had lower indoor concentrations than Nat_min, even with
the lowest efciency MERV 8 lter. Expectedly, the Nat_econ case,
which had the highest ventilation AERs and little ltration,
Table 6
exhibited the largest PM2.5 concentrations. Since PM2.5 was the
Indoor CO concentration distribution median (5th and 95th) percentiles in each only pollutant that had meaningful differences in natural versus
climate zone. Outdoor concentrations are not provided since they are nearly iden- mechanical cases, we use it as the primary metric of comparing IAQ
tical to indoor concentrations, since CO was treated as non-reactive and non- differences under the mechanical versus natural ventilation.
sorptive with no indoor emission.
Climate zone City CO (ppm)

3.4. Pollutant indoor/outdoor (I/O) ratios
1A Miami, FL 0.23 (0.037, 0.41)
2A Houston, TX 0.25 (0.15, 0.54) Beyond absolute indoor concentrations, the impacts of a venti-
2B Phoenix, AZ 0.27 (0.10, 0.83)
3A Atlanta, GA 0.41 (0.20, 0.73)
lation strategy on indoor pollution are more obvious according to
3B El Paso, TX 0.20 (0.11, 0.57) its inuence on the indoor/outdoor (I/O) ratio, since that metric is
3C Los Angeles, CA 0.28 (0.17, 0.74) more independent of the variability in outdoor pollutant concen-
4A Philadelphia, PA 0.24 (0.067, 0.52) trations. Therefore, the mean monthly I/O ratios for all pollutants in
4B Albuquerque, NM 0.19 (0.11, 0.43)
this study, as well as the PM2.5 I/O ratios for each MERV lter, were
4C Seattle, WA 0.21 (0.13, 0.43)
5A Boston, MA 0.25 (0.17, 0.47) plotted against mean monthly ventilation AER in Fig. 8. Corre-
5B Salt Lake City, UT 0.25 (0.14, 0.88) spondingly, Table 8 lists the median (5th and 95th) percentiles of
6A Milwaukee, WI 0.25 (0.25, 0.27) the I/O ratios for each pollutant under Mech_min and Nat_min
6B Billings, MT 0.16 (0.00, 0.75) strategies with a parameter, ki, equal to the square of the ratio of
7B Fargo, ND 0.15 (0.097, 0.22)
95th and 5th percentiles, as well as the linear ts, R2, and root mean
Table 7
Ambient and indoor concentration distribution median (5th and 95th) percentiles for NO2, O3, and PM2.5 under four ventilation cases in each simulated city.
Climate zone City Case NO2 (ppb) O3 (ppb) PM2.5 (mg/m3)
MERV8 MERV11 MERV16
1A Miami, FL Ambient 1.0 (0.00, 15) 30 (14, 51) 9.1 (4.5, 19) 9.1 (4.5, 19) 9.1 (4.5, 19)
Mech_min 0.31 (0.00, 4.0) 3.6 (1.7, 6.2) 4.1 (2.2, 8.4) 1.4 (0.64, 3.6) 0.33 (0.14, 1.3)
Nat_min 0.26 (0.00, 3.5) 2.9 (1.3, 5.2) 4.7 (2.6, 9.6) 2.9 (1.4, 6.6) 2.2 (1.0, 5.2)
Mech_econ 0.35 (0.00, 5.9) 3.9 (1.8, 18) 4.4 (2.2, 9.8) 1.5 (0.65, 4.5) 0.34 (0.14, 1.2)
Nat_econ 0.41 (0.00, 7.2) 3.7 (1.5, 20) 5.8 (2.7, 13) 3.7 (1.5, 12) 2.8 (1.1, 11)
2A Houston, TX Ambient 8.8 (2.7, 32) 22 (0.95, 53) 9.4 (3.7, 21) 9.4 (3.7, 21) 9.4 (3.7, 21)
Mech_min 2.4 (0.88, 7.6) 2.5 (0.14, 6.2) 4.5 (1.9, 10) 1.8 (0.63, 5.4) 0.47 (0.15, 2.5)
Nat_min 2.1 (0.74, 7.6) 2.3 (0.13, 5.8) 5.3 (2.2, 12) 3.6 (1.2, 10) 2.8 (0.89, 8.8)
Mech_econ 3.0 (0.90, 11) 3.1 (0.18, 14) 5.0 (1.9, 12) 2.1 (0.62, 6) 0.45 (0.14, 2.4)
Nat_econ 3.3 (0.79, 13) 3.1 (0.20, 16) 6.4 (2.4, 15) 4.8 (1.4, 13) 4.0 (1.1, 12)
2B Phoenix, AZ Ambient 16 (4.0, 46) 26 (1.0, 60) 9.5 (3.0, 20) 9.5 (3.0, 20) 9.5 (3.0, 20)
Mech_min 3.2 (1.0, 11) 4.1 (0.24, 7.4) 3.5 (1.5, 9.7) 1.2 (0.36, 4.9) 0.31 (0.071, 2.1)
Nat_min 2.9 (0.89, 11) 3.6 (0.27, 7.1) 4.4 (1.9, 12) 2.8 (0.78, 11) 2.1 (0.48, 9.3)
Mech_econ 4.6 (1.1, 21) 5.6 (0.35, 20) 3.9 (1.4, 13) 1.5 (0.36, 6.7) 0.33 (0.071, 2.1)
Nat_econ 4.0 (0.94, 20) 5.0 (0.35, 18) 5.4 (1.9, 17) 3.7 (0.97, 16) 3.0 (0.59, 15)
3A Atlanta, GA Ambient 6.5 (1.5, 26) 21 (2.0, 61) 8.0 (1.2, 19) 8.0 (1.2, 19) 8.0 (1.2, 19)
Mech_min 1.6 (0.39, 6.0) 3.3 (0.24, 7.5) 3.7 (0.92, 9.2) 1.5 (0.34, 5.2) 0.39 (0.079, 2.5)
Nat_min 1.4 (0.34, 5.9) 3.1 (0.21, 6.6) 4.4 (1.2, 11) 3.0 (0.72, 9.7) 2.4 (0.53, 8.7)
Mech_econ 2.0 (0.50, 8.2) 4.1 (0.25, 19) 4.2 (0.96, 11) 1.8 (0.38, 5.8) 0.42 (0.084, 2.4)
Nat_econ 2.2 (0.47, 8.8) 4.1 (0.28, 20) 5.4 (1.3, 14) 4.5 (0.93, 14) 3.8 (0.72, 13)
3B El Paso, TX Ambient 10 (3.1, 35) 32 (1.2, 58) 6.4 (1.9, 18) 6.4 (1.9, 18) 6.4 (1.9, 18)
Mech_min 2.7 (0.85, 8.9) 4.2 (0.30, 7.0) 2.7 (0.82, 8.6) 0.99 (0.23, 4.9) 0.25 (0.052, 2.2)
Nat_min 2.4 (0.75, 8.8) 3.7 (0.32, 6.4) 3.3 (1.1, 11) 2.1 (0.57, 9.9) 1.5 (0.41, 8.9)
Mech_econ 3.7 (0.99, 16) 5.6 (0.41, 24) 3.3 (0.87, 11) 1.5 (0.26, 6.0) 0.29 (0.056, 2.2)
Nat_econ 3.2 (0.86, 14) 4.7 (0.40, 21) 4.1 (1.2, 13) 3.2 (0.69, 12) 2.7 (0.50, 12)
3C Los Angeles, CA Ambient 12 (2.7, 40) 27 (0.62, 52) 12 (0.010, 37) 12 (0.010, 37) 12 (0.010, 37)
Mech_min 2.6 (0.69, 10) 3.9 (0.21, 6.6) 6.0 (0.00, 19) 2.3 (0.00, 9.4) 0.56 (0.00, 3.8)
Nat_min 2.4 (0.62, 10) 3.5 (0.20, 6.1) 6.9 (0.010, 22) 4.4 (0.00, 18) 3.3 (0.00, 15)
Mech_econ 4.8 (1.2, 17) 8.1 (0.26, 24) 8.4 (0.010, 25) 3.9 (0.00, 12) 0.69 (0.00, 3.5)
Nat_econ 5.0 (1.3, 18) 9.0 (0.29, 24) 11 (0.010, 33) 10 (0.010, 31) 9.4 (0.010, 30)
4A Philadelphia, PA Ambient 14 (5.0, 36) 18 (1.2, 40) 15 (6.8, 31) 15 (6.8, 31) 15 (6.8, 31)
Mech_min 3.6 (1.4, 9.4) 2.3 (0.19, 5.0) 8.3 (3.9, 18) 4.0 (1.3, 10) 1.3 (0.31, 4.8)
Nat_min 3.2 (1.3, 8.6) 2.0 (0.16, 4.8) 9.9 (4.7, 21) 7.6 (2.9, 19) 6.3 (2.1, 17)
Mech_econ 4.5 (1.7, 12) 2.8 (0.20, 13) 9.1 (4.3, 20) 4.6 (1.6, 11) 1.3 (0.34, 4.8)
Nat_econ 4.4 (1.6, 12) 2.6 (0.19, 13) 12 (5.6, 26) 10 (3.9, 24) 9.0 (2.9, 23)
4B Albuquerque, NM Ambient 8.7 (3.7, 31) 35 (3.0, 60) 4.2 (0.73, 15) 4.2 (0.73, 15) 4.2 (0.73, 15)
Mech_min 2.3 (1.1, 8.2) 4.6 (0.49, 7.2) 2.0 (0.47, 7.0) 0.84 (0.14, 3.8) 0.23 (0.031, 1.8)
Nat_min 2.1 (0.90, 7.9) 4.0 (0.52, 6.4) 2.5 (0.57, 8.5) 1.7 (0.30, 7.5) 1.3 (0.21, 6.6)
Mech_econ 3.8 (1.3, 11) 6.4 (0.60, 26) 2.5 (0.51, 8.2) 1.2 (0.19, 4.5) 0.27 (0.036, 1.8)
Nat_econ 3.6 (1.0, 11) 5.5 (0.59, 23) 3.2 (0.71, 10) 2.7 (0.48, 9.6) 2.4 (0.36, 8.8)
4C Seattle, WA Ambient 17 (5.0, 36) 17 (1.0, 38) 5.0 (1.7, 14) 5.0 (1.7, 14) 5.0 (1.7, 14)
Mech_min 4.5 (1.4, 9.8) 2.1 (0.12, 4.8) 2.8 (1.1, 7.6) 1.4 (0.47, 4.5) 0.48 (0.13, 2.1)
Nat_min 4.1 (1.3, 8.8) 1.9 (0.11, 4.6) 3.4 (1.3, 9.2) 2.8 (0.96, 8.5) 2.4 (0.75, 7.7)
Mech_econ 5.6 (1.7, 16) 2.8 (0.12, 12) 3.3 (1.2, 8.2) 1.7 (0.62, 4.6) 0.50 (0.15, 2.1)
Nat_econ 5.3 (1.6, 15) 2.7 (0.11, 12) 4.1 (1.5, 10) 3.9 (1.4, 9.6) 3.5 (1.2, 8.9)
5A Boston, MA Ambient 16 (6.0, 35) 22 (3.0, 42) 8.5 (2.6, 23) 8.5 (2.6, 23) 8.5 (2.6, 23)
Mech_min 4.4 (1.8, 9.2) 2.9 (0.48, 5.4) 4.8 (1.6, 13) 2.4 (0.58, 7.7) 0.86 (0.14, 3.4)
Nat_min 3.7 (1.6, 8.0) 2.5 (0.41, 5.0) 5.6 (1.8, 15) 4.7 (1.1, 14) 3.9 (0.84, 12)
Mech_econ 5.3 (2.1, 13) 3.4 (0.50, 15) 5.3 (1.8, 14) 2.8 (0.75, 7.8) 0.86 (0.16, 3.4)
Nat_econ 4.9 (1.9, 13) 3.1 (0.46, 14) 6.5 (2.3, 17) 5.9 (1.8, 15) 5.2 (1.4, 14)
5B Salt Lake City, UT Ambient 13 (2.0, 48) 30 (7.0, 54) 5.5 (0.80, 47) 5.5 (0.80, 47) 5.5 (0.80, 47)
Mech_min 3.2 (0.66, 13) 4.2 (1.0, 6.6) 2.6 (0.8, 31) 1.1 (0.23, 18) 0.34 (0.052, 7.2)
Nat_min 2.9 (0.60, 11) 3.6 (0.95, 6.2) 3.2 (0.98, 37) 2.4 (0.51, 33) 1.9 (0.35, 29)
Mech_econ 4.4 (0.88, 14) 5.2 (1.1, 20) 3.3 (0.83, 31) 1.6 (0.27, 18) 0.38 (0.056, 7.2)
Nat_econ 4.0 (0.79, 13) 4.6 (1.1, 18) 4.2 (1.1, 37) 3.7 (0.70, 33) 3.4 (0.52, 29)
6A Milwaukee, WI Ambient 7.7 (1.6, 26) 28 (3.9, 53) 9.0 (3.0, 23) 9.0 (3.0, 23) 9.0 (3.0, 23)
Mech_min 2.0 (0.46, 6.4) 3.6 (0.82, 6.5) 4.9 (1.7, 14) 2.3 (0.64, 7.8) 0.81 (0.15, 3.6)
Nat_min 1.8 (0.42, 5.7) 3.2 (0.74, 6.1) 5.8 (2.1, 16) 4.4 (1.3, 14) 3.6 (0.96, 13)
Mech_econ 2.6 (0.55, 9.3) 4.4 (0.95, 20) 5.6 (1.9, 15) 2.9 (0.76, 8.2) 0.84 (0.17, 3.5)
Nat_econ 2.3 (0.51, 9.5) 4.1 (0.88, 18) 6.6 (2.2, 18) 5.6 (1.5, 17) 4.9 (1.2, 16)
6B Billings, MT Ambient 2.0 (0.00, 13) 30 (7.0, 54) 4.0 (0.00, 17) 4.0 (0.00, 17) 4.0 (0.00, 17)
Mech_min 0.61 (0.079, 4.0) 4.3 (1.0, 6.6) 2.3 (0.11, 8.1) 0.99 (0.037, 4.4) 0.30 (0.010, 1.9)
Nat_min 0.57 (0.12, 3.6) 3.7 (0.93, 6.3) 2.8 (0.14, 9.8) 2.0 (0.070, 8.4) 1.6 (0.036, 7.6)
Mech_econ 0.77 (0.00, 5.3) 5.1 (1.1, 23) 2.7 (0.068, 11) 1.3 (0.030, 5.4) 0.35 (0.010, 1.9)
Nat_econ 0.72 (0.00, 5.0) 4.5 (1.0, 19) 3.4 (0.10, 13) 2.8 (0.065, 12) 2.3 (0.037, 11)
7B Fargo, ND Ambient 2.8 (0.00, 16) 28 (7.0, 49) 3.4 (0.00, 12) 3.4 (0.00, 12) 3.4 (0.00, 12)
Mech_min 0.64 (0.00, 4.1) 3.7 (0.99, 6.2) 1.9 (0.23, 5.9) 0.89 (0.070, 3.2) 0.32 (0.016, 1.4)
Nat_min 0.56 (0.00, 3.3) 3.0 (0.86, 5.8) 2.1 (0.28, 6.8) 1.6 (0.15, 5.8) 1.3 (0.09, 5.1)
Mech_econ 0.83 (0.00, 4.6) 4.4 (1.2, 18) 2.1 (0.18, 7.0) 1.1 (0.067, 3.8) 0.34 (0.014, 1.3)
Nat_econ 0.72 (0.00, 3.9) 3.8 (1.0, 16) 2.5 (0.24, 9.1) 2.1 (0.16, 8.6) 1.7 (0.11, 8.0)
Fig. 7. Cumulative distribution functions of ambient PM2.5 and indoor PM2.5 concentrations under four ventilation strategies simulated in Philadelphia, PA (climate region 4A) using
lters with efciencies of (a) MERV 8, (b) MERV 11, and (c) MERV 16.
Fig. 8. Mean monthly I/O ratios as a function of mean monthly ventilation AER, for (a) CO2; HCHO with emissions model set at (b) best estimate of kL 1.6 h1 and (c) kL 0.05 h1;
(d) CO; (e) O3; (f) NO2; and PM2.5 with lters of efciencies (g) MERV 8, (h) MERV 11, and (i) MERV 16. For (aec, e, f) data points for minimum strategies are behind data points for
economizing strategies on the plots. See text for details about HCHO emission models from Rackes and Waring [18].
squared error (RMSE) for the I/O ratio as a function of the range of mostly due to their I/O ratios having little change with the venti-
mean monthly ventilation AER when economizing. The I/O ratios lation AER (i.e., a at slope). However, all ts have low RMSE. For
have good t parameters, reinforcing the strong linear relationships PM2.5 for example, the linear ts well predict the I/O ratio as a
demonstrated in Fig. 8. We note that certain ts have low R2 values, function of ventilation AER with an accuracy of <11% of the outdoor
Table 8
For minimum ventilation cases, I/O ratio distribution median (5th and 95th) percentiles for pollutants and k, which is the squared ratio of the 95th and 5th percentiles. For
economizing cases, ts of I/O ratios as a function of ventilation AER, as well as R2 and normalized root mean squared error (NRMSE) for the ts.
Pollutant Mechanical ventilation
I/O at Mech_min ki I/O at Mech_econ R2 RMSE
CO2 1.64 (1.62, 1.66) 1.05 I/O 0.248lv 1.69 0.76 0.071
CO 1.03 (0.99, 1.14) 1.31 I/O 0.0240lv 1.05 0.090 0.038
NO2 0.270 (0.270, 0.290) 1.15 I/O 0.179lv 0.217 0.97 0.015
O3 0.121 (0.116, 0.127) 1.19 I/O 0.138lv 0.074 0.98 0.009
HCHO kL 1.6 h1 6.79 (6.70, 6.82) 1.04 I/O 1.37lv 7.20 0.99 0.079
kL 0.05 h1 7.11 (6.64, 7.40) 1.24 I/O 2.22lv 7.53 0.95 0.268
PM2.5 TOA > Tbal MERV8 0.445 (0.371, 0.548) 2.18 I/O 0.141lv 0.374 0.72 0.044
MERV11 0.166 (0.121, 0.221) 3.32 I/O 0.101lv 0.105 0.73 0.030
MERV16 0.0492 (0.0319, 0.0761) 5.69 I/O 0.019lv 0.030 0.37 0.012
PM2.5 TOA < Tbal MERV8 0.679 (0.532, 0.810) 2.32 I/O 0.009lv 0.750 0.00 0.102
MERV11 0.370 (0.227, 0.438) 3.74 I/O 0.036lv 0.439 0.06 0.067
MERV16 0.155 (0.0859, 0.189) 4.83 I/O 0.049lv 0.194 0.35 0.033
Pollutant Natural ventilation
I/O Nat_min ki I/O at Nat_econ R2 RMSE
CO2 1.75 (1.64, 1.83) 1.25 I/O 0.337lv 1.83 0.73 0.101
CO 1.03 (0.992, 1.18) 1.41 I/O 0.034lv 1.06 0.13 0.043
NO2 0.243 (0.218, 0.285) 1.70 I/O 0.210lv 0.176 0.97 0.019
O3 0.107 (0.0926, 0.128) 1.90 I/O 0.151lv 0.055 0.98 0.009
HCHO kL 1.6 h1 6.96 (6.73, 7.11) 1.11 I/O 1.51lv 7.38 0.97 0.135
kL 0.05 h1 8.02 (6.88, 8.99) 1.71 I/O 3.02lv 8.48 0.92 0.457
PM2.5 TOA > Tbal MERV8 0.521 (0.453, 0.675) 2.23 I/O 0.201lv 0.480 0.75 0.054
MERV11 0.311 (0.241, 0.415) 2.96 I/O 0.300lv 0.201 0.90 0.048
MERV16 0.228 (0.173, 0.325) 3.53 I/O 0.317lv 0.105 0.91 0.046
PM2.5 TOA < Tbal MERV8 0.811 (0.643, 1.02) 2.50 I/O 0.029lv 0.903 0.01 0.169
MERV11 0.716 (0.447, 0.903) 4.07 I/O 0.056lv 0.803 0.23 0.158
MERV16 0.620 (0.363, 0.788) 4.72 I/O 0.095lv 0.679 0.11 0.130
concentration for Mech_econ and <17% of the outdoor concentra- of I/O ratios as functions of ventilation AERs are instructive. By
tion for Nat_econ strategies. These ts are useful for estimating I/O substituting the baseline minimum AER 0.383 h1 into lv for the
ratios at different AERs under various ventilation regimes for use in ts in Table 8, one recovers the central tendency of the I/O ratio at
concentration and exposure modeling and risk assessment. the minimum condition for any pollutant. These values are slightly
Between the minimum Mech_min and Nat_min strategies, the I/ lower than minimum strategy medians for the I/O ratio distribu-
O ratios changed little for gaseous pollutants of CO2, HCHO, CO, tions of pollutants of outdoor origin, and vice versa for pollutants of
NO2, and O3, due to their nearly equivalent ventilation AERs in the indoor origin. All pollutants were linear functions of the ventilation
modeling and the fact that these analogues had no differences in AER except for CO, which changed little with ventilation since it
indoor losses. For these gases, the I/O ratios also had small vari- was non reactive and without indoor sources. The ts for CO2,
ability within each of the minimum strategies. As such, the ki pa- HCHO, NO2, and O3 were basically independent of whether the
rameters of the gaseous I/O ratios are all within 30% of unity for the economizing was driven by mechanical or natural ventilation,
Mech_min and are only slightly higher for Nat_min, indicating a owing to there being no or little (for O3 only) removal by the HVAC
narrow range of change across cities or during different times of lter for these pollutants. Their linear t slope intensities depend
year. For PM2.5, while its ki parameters for the I/O ratios under on pollutant source strengths, and the sign of slope on whether
Mech_min and Nat_min are relatively larger, the nominal spreads source proles were outdoor or indoor dominated. Therefore,
seen in the condence intervals are low and the high ki appear due slopes were negative for indoor generated CO2 and HCHO, and
to the small magnitudes of these PM2.5 I/O ratios. there were larger I/O ratios and a more negative slope for HCHO
Also, for PM2.5, there were differences between the minimum predicted with the AER-independent versus AER-dependent
strategies, and the PM2.5 I/O ratios for Nat_min were higher than models. Conversely, slopes were positive for O3 and NO2, and
Mech_min since the natural ventilation strategy used active ltra- they were slightly higher for NO2 since it is less reactive indoors
tion only when air was recirculated by the hybrid system during than O3 under the surface reactivity assumptions used in this
times of extreme outdoor temperatures. Table 4 and Fig. 4 sum- model.
marize the hybrid operational times, and to further aid in inter- Opposite than the gases, PM2.5 had clear differences between
pretation, Table 8 separately lists I/O ratios for PM2.5 when the the Mech_econ and Nat_econ strategies regarding its linear ts of I/
outdoor air temperature (TOA) is above or below the simulated O ratios with ventilation AER. For the same reason discussed earlier,
buildings balance temperature (Tbal 17.5 C). Because low the PM2.5 I/O ratios and linear t parameters exhibited two distinct
ambient temperatures can sometimes offset cooling needs but trends within each strategy, one when outdoor temperatures (TOA)
outdoor temperatures higher than the balance temperature always were lower and one when TOA were higher than the balance tem-
require cooling (and associated ltration), PM2.5 I/O ratios are lower perature (Tbal 17.5 C). For both economizing strategies, during
when TOA > Tbal. Furthermore, the lter MERV had a large impact on the warmer months (TOA > Tbal), the slopes are steeper due to the
I/O ratios for PM2.5, so the differences in PM2.5 I/O ratios due to the use of recirculation to meet loads, while during the cooler months
impact of ltration between the Mech_min and Nat_min strategies (TOA < Tbal), the slopes are atter and change less with ventilation.
increased as the lter MERV improved. Also, for Mech_econ, the slopes decreased as the lter MERV
For economizing cases of Mech_econ and Nat_econ, the linear ts increased, no matter the TOA. For Nat_econ, however, the trends are
Fig. 9. Mean monthly I/O ratios as a function of mean monthly outdoor air temperature, for (a) CO2; HCHO with emissions model set at (b) best estimate of kL 1.6 h1 and (c)
kL 0.05 h1; (d) CO; (e) O3; (f) NO2; and PM2.5 with lters of efciencies (g) MERV 8, (h) MERV 11, and (i) MERV 16. See text for details about HCHO emission models from Rackes
and Waring [18].
reversed and the slopes increase with MERV lter efciency, trends of I/O ratios for these gases are nearly identical when
meaning higher efciency ltration leads to the PM2.5 I/O ratio comparing the mechanical ventilation strategies to their corre-
increasing more per associated ventilation AER unit change. For this sponding natural ventilation analogues.
case, a higher ventilation rate indicates that a smaller thermal load In contrast with the gases, the PM2.5 I/O ratios strongly varied
is required to condition the building, as inferred by the fact that the between ventilation analogues at different mean monthly tem-
system is economizing, and therefore we observe a decrease in peratures. I/O ratios for mechanical strategies were always lower
hybrid operation and return air and hence less efcacious ltration than natural ventilation strategies, particularly for economizing
(see Fig. 4b). times. Indeed, for Mech_Econ, the PM2.5 I/O ratios were little
Since the ventilation rate is dynamic and enthalpy-driven in the affected by higher economizing ventilation AERsdespecially as the
economizing strategies, we also plotted the mean monthly I/O ra- MERV lter efciency increaseddeven though the ventilation rate
tios of each pollutant against the mean monthly outdoor air tem- increased sometimes by an order of magnitude over minimum
perature in Fig. 9. This plot allows one to estimate the impacts of rates to take advantage of free cooling. In fact, the impact of higher
ventilation strategies across a range of climates and seasons. Fig. 9 efciency ltration was so strong that PM2.5 I/O ratios were almost
demonstrates that for CO2 and HCHO, both of which have strong identical under Mech_min and Mech_econ strategies for each MERV
indoor sources, there is a signicant decrease in I/O ratios when lter, and were independent of whether the system was econo-
economizing is employed during temperate months by the mizing. Conversely, ltration had a minimal effect on PM2.5 I/O
Mech_econ and Nat_econ strategies. For NO2 and O3, which have ratios for Nat_min and Nat_econ strategies, and use of these stra-
indoor concentrations entirely driven by outdoor sources in this tegies would lead to higher PM2.5 exposures, especially during
model, a reverse trend occurs. For CO, which would have a steady temperate months. One can observe a small decrease of PM2.5 I/O
state concentration equal to its outdoor concentration, we observe ratios in warm climates for naturally ventilated ofces due to the
an I/O ratio of near unity throughout all possible U.S. climates, hybrid operation that allows the use of a recirculation air lter (see
seasons, and ventilation strategies. Again, due to similar losses, the Table 4 and Fig. 4 for hybrid information).
3.5. Assumptions and limitations reduction since it uses economizing and allows a wider tem-
perature setpoint band.
This simulation study compared the impacts of natural versus Heating energy was often reduced by natural ventilation stra-
mechanical ventilation strategies on energy consumption and IAQ tegies, due to the wider setpoint band allowed with natural
parameters in an ofce. It compared absolute differences and ventilation strategies.
quantied trends of how I/O ratios change as a function of venti- Fan energy changes were an order of magnitude smaller than
lation AER, for analogous natural and mechanical ventilation stra- heating or cooling energy saving possibilities among strategies,
tegies. When interpreting the results, there are a number of and it should not be considered as a driver of decision-making.
assumptions and limitations that must be kept in mind. The rst Regarding gases, CO2 and HCHO were most affected by whether
and most important is that this study was meant as an ideal ventilation was at a constant minimum or allowed to econo-
investigation of the limits of pollutant concentrations and I/O ratios mize, and they were little affected by location. However, O3 and
at the natural and mechanical ventilation boundaries of strict NO3 were also affected by location since their outdoor variability
minimums and economizing at full thermodynamic potential. Since is greater. These four pollutants were generally independent of
it was ideal, we neglected many HVAC system implementation is- whether ventilation was natural or mechanical. Finally, CO was
sues, such as difculties in supplying natural ventilation effectively little affected by any of the variables altered in this work except
to all building zones; the inability of multi-zone VAV systems to the outdoor concentration.
supply ventilation well to all zones in general; and the inability of Regarding particles, PM2.5 was affected by all changed variables,
building openings to be controlled in coordination with transient including whether ventilation was at a constant minimum or
wind magnitudes and directions such that ow rates mimic me- economizing, whether it was natural or mechanical, and by
chanically driven ows. location. Moreover, higher lter efciency ltration greatly
Regarding simulation-specic issues, the HVAC system was ideal reduced indoor PM2.5 concentrations.
in that particular equipment components with their various ef- For economizing operation, since higher ventilation rates
ciencies were not specied. Therefore, the energy consumption elevate outdoor originating pollutants and reduce indoor orig-
values should not be taken as representative at their nominal inating pollutants, the outdoor originating pollutants have
values, but reasonable average efciencies for different equipment higher I/O ratios during temperate months as compared to more
could be applied if desired. However, changes in energy use be- extreme hot or cold months, and vice versa for indoor-
tween strategies can be readily compared to impacts of changes in originating pollutants. However, for PM2.5, the use of high ef-
IAQ behavior for ventilation strategies during different ambient ciency ltration leads to little difference between mechanical
conditions (both are presented as a function of the mean monthly minimum and economizing ventilation strategies.
outdoor air temperature). Moreover, the IAQ model comprised In our modeled climates, there was often a need to use a hybrid
many assumptions, such as that the indoor air was well mixed, recirculating system in conjunction with natural ventilation
which is likely a less good assumption for natural ventilation; that strategies to maintain comfort conditions, but the associated
no indoor particle formation occurred due to indoor chemistry ltration impact was rarely strong enough to match that of
[73e77] and that particles could be treated as nonvolatile [78]; that mechanical ventilation strategy ltration.
outdoor and emission conditions were constant over each hour;
and that the losses were constant over an annual or daily schedule. Therefore, natural ventilation may be desirable to reduce heat-
Finally, a streamlined set of indoor air pollutants was analyzed, and ing and cooling loads. As long as natural ventilation air delivery
one discrete ofce building with one HVAC system-type was could be controlled well, the IAQ difference may be minimal for all
simulated. pollutants except PM2.5, which cannot be ltered effectively when
outdoor air is delivered through building openings rather than by
an AHU. Conversely, ltration was highly mitigating for both me-
3.6. Summary and conclusions chanical ventilation strategies. Thus, for PM2.5, understanding the
climate, ventilation strategy, and ltration efciency is necessary to
Though this simulation study was ideal and made assumptions, predict I/O ratio trends. Though PM2.5 was the only pollutant
the trends understood from both the energy and IAQ analyses are affected by the use of natural versus mechanical ventilation, this
instructive of potential manifestations of impacts in real buildings. PM2.5 concentration change is meaningful owing to the extensive
The energy consumption data can be coupled with the IAQ pre- chronic health impacts that may result from exposure to PM2.5
dictions, and for a given mean monthly outdoor air temperature concentrations [34,40,79,80]. Therefore, we suggest that the use of
and ventilation strategy, the energy consumption from heating, natural ventilation needs to be further assessed within a health and
cooling and fan energy as well as the I/O ratios of the modeled monetary framework as a tradeoff between energy savings and
pollutants can be contextualized. Additionally, the I/O ratios of the elevation of indoor PM to determine its appropriateness and
modeled pollutants can be predicted as a function of average favorability over mechanical ventilation.
ventilation AER, using ts in Table 8. For each AER, the energy
consumption (or savings) due to ventilation can be computed, and Acknowledgements
since the AERs dictate the I/O ratios of indoor pollutants in a pre-
dictable manner, tradeoffs between the two can be estimated in This work was partially supported by the U.S. National Science
order to explore ventilation strategy impacts on ofce IAQ and Foundation under Grant No. 1511151. We thank Taylor Castonguay
energy consumption. for help with data collection and organization.
This study showed that using natural instead of mechanical
ventilation can result in changes in both energy use and indoor air References
concentrations. Specic conclusions this study reached were:
rez-Lombard, J. Ortiz, C. Pout, A review on buildings energy consumption
[1] L. Pe
Cooling energy was reduced under all ventilation strategies information, Energy Build. 40 (2008) 394e398, http://dx.doi.org/10.1016/
j.enbuild.2007.03.007.
compared to the mechanical minimum strategy, with the nat- [2] U.S. EIA, Emissions of Greenhouse Gases in the United States 2009, U.S. Energy
ural ventilation economizing strategy showing the largest Information Administration, 2011.
[3] U.S. DOE, Buildings Energy Databook, Energy Efciency & Renewable Energy annurev.pu.15.050194.000543.
Department, 2012. [31] D.W. Dockery, J. Schwartz, J.D. Spengler, Air pollution and daily mortality:
[4] M. Orme, Estimates of the energy impact of ventilation and associated associations with particulates and acid aerosols, Environ. Res. 59 (1992)
nancial expenditures, Energy Build. 33 (2001) 199e205, http://dx.doi.org/ 362e373, http://dx.doi.org/10.1016/S0013-9351(05)80042-8.
10.1016/S0378-7788(00)00082-7. [32] F. Dominici, R.D. Peng, M.L. Bell, et al., FIne particulate air pollution and
[5] A. Rackes, M.S. Waring, Using multiobjective optimizations to discover dy- hospital admission for cardiovascular and respiratory diseases, JAMA 295
namic building ventilation strategies that can improve indoor air quality and (2006) 1127e1134, http://dx.doi.org/10.1001/jama.295.10.1127.
reduce energy use, Energy Build. 75 (2014) 272e280, http://dx.doi.org/ [33] Pope, Burnett, Thun, et al., Lung cancer, cardiopulmonary mortality, and long-
10.1016/j.enbuild.2014.02.024. term exposure to ne particulate air pollution, JAMA 287 (2002) 1132e1141,
[6] O. Sepp anen, W.J. Fisk, Q.H. Lei, Ventilation and performance in ofce work, http://dx.doi.org/10.1001/jama.287.9.1132.
Indoor Air 16 (2006) 28e36, http://dx.doi.org/10.1111/j.1600- [34] C.A. Pope, D.W. Dockery, J. Schwartz, Review of epidemiological evidence of
0668.2005.00394.x. health effects of particulate air pollution, Inhal. Toxicol. 7 (1995) 1e18, http://
[7] U. Satish, M.J. Mendell, K. Shekhar, T. Hotchi, D. Sullivan, S. Streufert, et al., Is dx.doi.org/10.3109/08958379509014267.
CO2 an indoor pollutant? Direct effects of low-to-moderate CO2 concentra- [35] C.A. Pope, R.T. Burnett, D. Krewski, M. Jerrett, Y. Shi, E.E. Calle, et al., Cardio-
tions on human decision-making performance, Environ. Health Perspect. 120 vascular mortality and exposure to airborne ne particulate matter and
(2012) 1671e1677, http://dx.doi.org/10.1289/ehp.1104789. cigarette smoke shape of the exposure-response relationship, Circulation 120
[8] M.G. Apte, W.J. Fisk, J.M. Daisey, Associations between indoor CO2 concen- (2009) 941e948, http://dx.doi.org/10.1161/CIRCULATIONAHA.109.857888.
trations and sick building syndrome symptoms in U.S. Ofce buildings: an [36] J. Schwartz, D.W. Dockery, L.M. Neas, Is daily mortality associated specically
analysis of the 1994e1996 base study data, Indoor Air 10 (2000) 246e257, with ne particles? J. Air Waste Manag. Assoc. 46 (1996) 927e939, http://
http://dx.doi.org/10.1034/j.1600-0668.2000.010004246.x. dx.doi.org/10.1080/10473289.1996.10467528.
[9] W.J. Fisk, A.G. Mirer, M.J. Mendell, Quantication of the Association of [37] M.L. Bell, R.D. Peng, F. Dominici, The exposure-response curve for ozone and
Ventilation Rates with Sick Building Syndrome Symptoms, Lawrence Berkeley risk of mortality and the adequacy of current ozone regulations, Environ.
Natl. Lab., 2009. http://escholarship.org/uc/item/2d10c901 (accessed Health Perspect. 114 (2006) 532e536.
04.02.16). [38] M.L. Bell, A. M, S.L. Zeger, J.M. Samet, F. Dominici, Ozone and short-term
[10] J. Sundell, H. Levin, W.W. Nazaroff, W.S. Cain, W.J. Fisk, D.T. Grimsrud, et al., mortality in 95 us urban communities, 1987-2000, JAMA 292 (2004)
Ventilation rates and health: multidisciplinary review of the scientic litera- 2372e2378, http://dx.doi.org/10.1001/jama.292.19.2372.
ture, Indoor Air 21 (2011) 191e204, http://dx.doi.org/10.1111/j.1600- [39] G. Bylin, T. Lindvall, T. Rehn, B. Sundin, Effects of short-term exposure to
0668.2010.00703.x. ambient nitrogen dioxide concentrations on human bronchial reactivity and
[11] D.K. Milton, P.M. Glencross, M.D. Walters, Risk of sick leave associated with lung function, Eur. J. Respir. Dis. 66 (1985) 205e217.
outdoor air supply rate, humidication, and occupant complaints, Indoor Air [40] J.M. Logue, P.N. Price, M.H. Sherman, B.C. Singer, A method to estimate the
10 (2000) 212e221. chronic health impact of air pollutants in U.S. Residences, Environ. Health
[12] M.J. Mendell, E.A. Eliseeva, M.M. Davies, M. Spears, A. Lobscheid, W.J. Fisk, et Perspect. 120 (2012) 216e222, http://dx.doi.org/10.1289/ehp.1104035.
al., Association of classroom ventilation with reduced illness absence: a pro- [41] H. Guo, S.C. Lee, L.Y. Chan, W.M. Li, Risk assessment of exposure to volatile
spective study in California elementary schools, Indoor Air 23 (2013) organic compounds in different indoor environments, Environ. Res. 94 (2004)
515e528, http://dx.doi.org/10.1111/ina.12042. 57e66, http://dx.doi.org/10.1016/S0013-9351(03)00035-5.
[13] ASHRAE, Standard 62.1-2013 Ventilation for Acceptable Indoor Air Quality, [42] G. Beko , G. Clausen, C.J. Weschler, Is the use of particle air ltration justied?
2013. Costs and benets of ltration with regard to health effects, building cleaning
[14] M. Santamouris, F. Allard, Natural Ventilation in Buildings: a Design Hand- and occupant productivity, Build. Environ. 43 (2008) 1647e1657, http://
book, Earthscan, 1998. dx.doi.org/10.1016/j.buildenv.2007.10.006.
[15] R.J. de Dear, G.S. Brager, Thermal comfort in naturally ventilated buildings: [43] Z. El Orch, B. Stephens, M.S. Waring, Predictions and determinants of size-
revisions to ASHRAE Standard 55, Energy Build. 34 (2002) 549e561, http:// resolved particle inltration factors in single-family homes in the U.S. Build.
dx.doi.org/10.1016/S0378-7788(02)00005-1. Environ. 74 (2014) 106e118, http://dx.doi.org/10.1016/
[16] E. Shaviv, A. Yezioro, I.G. Capeluto, Thermal mass and night ventilation as j.buildenv.2014.01.006.
passive cooling design strategy, Renew. Energy 24 (2001) 445e452, http:// [44] W.J. Fisk, D. Faulkner, J. Palonen, O. Seppanen, Performance and costs of
dx.doi.org/10.1016/S0960-1481(01)00027-1. particle air ltration technologies, Indoor Air 12 (2002) 223e234, http://
[17] W.W. Nazaroff, G.R. Cass, Mathematical modeling of chemically reactive dx.doi.org/10.1034/j.1600-0668.2002.01136.x.
pollutants in indoor air, Environ. Sci. Technol. 20 (1986) 924e934, http:// [45] O.O. Ha nninen, J. Palonen, J.T. Tuomisto, T. Yli-Tuomi, O. Seppa nen,
dx.doi.org/10.1021/es00151a012. M.J. Jantunen, Reduction potential of urban PM2.5 mortality risk using mod-
[18] A. Rackes, M.S. Waring, Do time-averaged, whole-building, effective volatile ern ventilation systems in buildings, Indoor Air 15 (2005) 246e256, http://
organic compound (VOC) emissions depend on the air exchange rate? A dx.doi.org/10.1111/j.1600-0668.2005.00365.x.
statistical analysis of trends for 46 VOCs in U.S. ofces, Indoor Air (2015), [46] W.J. Riley, T.E. McKone, A.C.K. Lai, W.W. Nazaroff, Indoor particulate matter of
http://dx.doi.org/10.1111/ina.12224. outdoor origin: importance of size-dependent removal mechanisms, Environ.
[19] C.J. Weschler, H.C. Shields, The inuence of ventilation on reactions among Sci. Technol. 36 (2002) 200e207, http://dx.doi.org/10.1021/es010723y.
indoor pollutants: modeling and experimental observations, Indoor Air 10 [47] M.S. Waring, J.A. Siegel, Particle loading rates for HVAC lters, heat ex-
(2000) 92e100, http://dx.doi.org/10.1034/j.1600-0668.2000.010002092.x. changers, and ducts, Indoor Air 18 (2008) 209e224, http://dx.doi.org/
[20] L.E. Ekberg, Concentrations of NO2 and other trafc related contaminants in 10.1111/j.1600-0668.2008.00518.x.
ofce buildings located in urban environments, Build. Environ. 30 (1995) [48] M.S. Waring, J.A. Siegel, The inuence of HVAC systems on indoor secondary
293e298, http://dx.doi.org/10.1016/0360-1323(94)00040-Y. organic aerosol formation, ASHRAE Trans. 116 (2010) 556e571.
[21] L. Morawska, A. Afshari, G.N. Bae, G. Buonanno, C.Y.H. Chao, O. Ha nninen, et [49] Z. Tong, Y. Chen, A. Malkawi, G. Adamkiewicz, J.D. Spengler, Quantifying the
al., Indoor aerosols: from personal exposure to risk assessment, Indoor Air 23 impact of trafc-related air pollution on the indoor air quality of a naturally
(2013) 462e487, http://dx.doi.org/10.1111/ina.12044. ventilated building, Environ. Int. 89e90 (2016) 138e146, http://dx.doi.org/
[22] T.N. Quang, C. He, L. Morawska, L.D. Knibbs, Inuence of ventilation and 10.1016/j.envint.2016.01.016.
ltration on indoor particle concentrations in urban ofce buildings, Atmos. [50] U.S. DOE, EnergyPlusEngineering Reference: the Reference to EnergyPlus
Environ. 79 (2013) 41e52, http://dx.doi.org/10.1016/j.atmosenv.2013.06.009. Calculations (In Case You Want or Need to Know), 2013.
[23] C.J. Weschler, Ozone in indoor environments: concentration and chemistry, [51] U.S. DOE, Input Output Reference: an Encyclopedic Reference to EnergyPlus
Indoor Air 10 (2000) 269e288, http://dx.doi.org/10.1034/j.1600- Inputs and Outputs, U.S. Department of Energy, 2013.
0668.2000.010004269.x. [52] U.S. EPA, AirData Website File Download Page, (n.d.). http://aqsdr1.epa.gov/
[24] A. Rackes, M.S. Waring, Modeling impacts of dynamic ventilation strategies on aqsweb/aqstmp/airdata/download_les.html (accessed 24.04.15).
indoor air quality of ofces in six US cities, Build. Environ. 60 (2013) 243e253, [53] U.S. EIA, Commercial Buildings Energy Consumption Survey, U.S. Energy In-
http://dx.doi.org/10.1016/j.buildenv.2012.10.013. formation Administration, 2015.
[25] N.E. Klepeis, W.C. Nelson, W.R. Ott, J.P. Robinson, A.M. Tsang, P. Switzer, et al., [54] U.S. EPA, Building Assessment Survey and Evaluation Study, E.S. Environ-
The National Human Activity Pattern Survey (NHAPS): a resource for assess- mental Protection Agency, n.d. http://www.epa.gov/indoor-air-quality-iaq/
ing exposure to environmental pollutants, J. Expo. Anal. Environ. Epidemiol. building-assessment-survey-and-evaluation-study (accessed 04.02.16).
11 (2001) 231e252, http://dx.doi.org/10.1038/sj.jea.7500165. [55] M.V. Swami, S. Chandra, Procedures for Calculating Natural Ventilation
[26] L. Wallace, A decade of studies of human exposure: what have we learned? Airow Rates in Buildings, ASHRAE, 1987.
Risk Anal. Off. Publ. Soc. Risk Anal. 13 (1993) 135e139. [56] ASHRAE, NSI/ASHRAE/IES standard 90.1-2013 e energy standard for buildings
[27] L.A. Wallace, Comparison of risks from outdoor and indoor exposure to toxic except low-rise residential buildings, Am. Soc. Heat. Refrig. Air-Cond. Eng. Inc.
chemicals, Environ. Health Perspect. 95 (1991) 7e13. (2013).
[28] L. Wallace, Indoor particles: a review, J. Air Waste Manag. Assoc. 46 (1996) [57] U.S. DOE, Guide to Determining Climate Regions by County, 7.3, 2015.
98e126, http://dx.doi.org/10.1080/10473289.1996.10467451. [58] U.S. DOE, EnergyPlus: Weather Data, (n.d.). https://energyplus.net/weather
[29] C.J. Weschler, Ozones impact on public health: contributions from indoor (accessed 08.04.15).
exposures to ozone and products of ozone-initiated chemistry, Environ. [59] U.S. DOE, EnergyPlus Energy Simulation Software: Add-Ons., (n.d.). http://
Health Perspect. 114 (2006) 1489e1496. apps1.eere.energy.gov/buildings/energyplus/energyplus_addons.cfm/
[30] D.W. Dockery, C.A. Pope, Acute respiratory effects of particulate air pollution, (accessed 06.04.15).
Annu. Rev. Public Health 15 (1994) 107e132, http://dx.doi.org/10.1146/ [60] C.K. Wilkins, M.H. Hosni, Plug load design factors, ASHRAE J. 53 (2011) 30e32,
34. discussion 109e27.

[61] W.W. Nazaroff, Indoor particle dynamics, Indoor Air (14 Suppl. 7) (2004) [71] E.L. Hult, H. Willem, P.N. Price, T. Hotchi, M.L. Russell, B.C. Singer, Formalde-
175e183, http://dx.doi.org/10.1111/j.1600-0668.2004.00286.x. hyde and acetaldehyde exposure mitigation in US residences: in-home
[62] K. Belanger, J.F. Gent, E.W. Triche, M.B. Bracken, B.P. Leaderer, Association of measurements of ventilation control and source control, Indoor Air 25
indoor nitrogen dioxide exposure with respiratory symptoms in children with (2015) 523e535, http://dx.doi.org/10.1111/ina.12160.
asthma, Am. J. Respir. Crit. Care Med. 173 (2006) 297e303, http://dx.doi.org/ [72] P. Azimi, D. Zhao, B. Stephens, Estimates of HVAC ltration efciency for ne
10.1164/rccm.200408-1123OC. and ultrane particles of outdoor origin, Atmos. Environ. 98 (2014) 337e346,
[63] V.J. Cogliano, Y. Grosse, R.A. Baan, K. Straif, M.B. Secretan, F. El Ghissassi, et al., http://dx.doi.org/10.1016/j.atmosenv.2014.09.007.
Meeting report: summary of IARC monographs on formaldehyde, 2- [73] S. Yousse, M.S. Waring, Transient secondary organic aerosol formation from
butoxyethanol, and 1-tert-butoxy-2-propanol, Environ. Health Perspect. 113 limonene ozonolysis in indoor environments: impacts of air exchange rates
(2005) 1205e1208. and initial concentration ratios, Environ. Sci. Technol. 48 (2014) 7899e7908,
[64] J.G. Allen, P. MacNaughton, U. Satish, S. Santanam, J. Vallarino, J.D. Spengler, http://dx.doi.org/10.1021/es5009906.
Associations of cognitive function scores with carbon dioxide, ventilation, and [74] S. Yousse, M.S. Waring, Indoor transient SOA formation from ozone a-
volatile organic compound exposures in ofce workers: a controlled exposure pinene reactions: impacts of air exchange and initial product concentrations,
study of green and conventional ofce environments, Environ. Health Per- and comparison to limonene ozonolysis, Atmos. Environ. 112 (2015)
spect. (2015), http://dx.doi.org/10.1289/ehp.1510037. 106e115, http://dx.doi.org/10.1016/j.atmosenv.2015.04.001.
[65] M.S. Waring, J.R. Wells, Volatile organic compound conversion by ozone, [75] M.S. Waring, Secondary organic aerosol in residences: predicting its fraction
hydroxyl radicals, and nitrate radicals in residential indoor air: magnitudes of ne particle mass and determinants of formation strength, Indoor Air 24
and impacts of oxidant sources, Atmos. Environ. 106 (2015) 382e391, http:// (2014) 376e389, http://dx.doi.org/10.1111/ina.12092.
dx.doi.org/10.1016/j.atmosenv.2014.06.062. [76] M.S. Waring, J.R. Wells, J.A. Siegel, Secondary organic aerosol formation from
[66] J.R. Girman, M.G. Apte, G.W. Traynor, J.R. Allen, C.D. Hollowell, Pollutant ozone reactions with single terpenoids and terpenoid mixtures, Atmos. En-
emission rates from indoor combustion appliances and sidestream cigarette viron. 45 (2011) 4235e4242, http://dx.doi.org/10.1016/
smoke, Environ. Int. 8 (1982) 213e221, http://dx.doi.org/10.1016/0160- j.atmosenv.2011.05.001.
4120(82)90030-7. [77] Y. Yang, M.S. Waring, Secondary organic aerosol formation initiated by a-
[67] S.C. Lee, S. Lam, H. Kin Fai, Characterization of VOCs, ozone, and PM10 terpineol ozonolysis in indoor air, Indoor Air (2016), http://dx.doi.org/
emissions from ofce equipment in an environmental chamber, Build. Envi- 10.1111/ina.12271 n/aen/a.
ron. 36 (2001) 837e842, http://dx.doi.org/10.1016/S0360-1323(01)00009-9. [78] A.M. Johnson, M.S. Waring, P.F. DeCarlo, Real-time transformation of outdoor
[68] B. Stephens, E.T. Gall, J.A. Siegel, Measuring the penetration of ambient ozone aerosol components upon transport indoors measured with aerosol mass
into residential buildings, Environ. Sci. Technol. 46 (2012) 929e936, http:// spectrometry, Indoor Air (2016), http://dx.doi.org/10.1111/ina.12299.
dx.doi.org/10.1021/es2028795. [79] S.M. Dutton, D. Banks, S.L. Brunswick, W.J. Fisk, Health and economic impli-
[69] G. Beko , G. Clausen, C.J. Weschler, Further studies of oxidation processes on cations of natural ventilation in California ofces, Build. Environ. 67 (2013)
lter surfaces: evidence for oxidation products and the inuence of time in 34e45, http://dx.doi.org/10.1016/j.buildenv.2013.05.002.
service, Atmos. Environ. 41 (2007) 5202e5212, http://dx.doi.org/10.1016/ [80] R.T. Burnett, M. Smith-doiron, D. Stieb, S. Cakmak, J.R. Brook, Effects of par-
j.atmosenv.2006.07.063. ticulate and gaseous air pollution on cardiorespiratory hospitalizations, Arch.
[70] C.P. Weisel, J. Zhang, B.J. Turpin, M.T. Morandi, S. Colome, T.H. Stock, et al., Environ. Health Int. J. 54 (1999) 130e139, http://dx.doi.org/10.1080/
Relationships of indoor, outdoor, and personal air (RIOPA). Part I. Collection 00039899909602248.
methods and descriptive analyses, Res. Rep. Health Eff. Inst. (2005) 1e107

Building and Environment: Tom Ben-David, Michael S. Waring

Uploaded by

Copyright:

Available Formats

Building and Environment: Tom Ben-David, Michael S. Waring

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Building and Environment: Tom Ben-David, Michael S. Waring

Uploaded by

Copyright:

Available Formats

Building and Environment 104 (2016) 320e336

Contents lists available at ScienceDirect

Building and Environment

Impact of natural versus mechanical ventilation on simulated indoor

1. Introduction maintain acceptable indoor air quality [13].

1A Miami, FL 2.3 5.3 3.9 105

1 Lighting strategy, we set them equal to each other. We assumed that at

Equation (1), so that:

where cj is the c coefcient for location j; and vmean,j (m/s) is the

number of occupants, respectively.

Pollutant Sources Losses

Carbon dioxide (CO2) Ep;CO2

Climate zone City Minimum AER (h1) Economizing AER (h1)

Mech_min Nat_min RMSE Mech_econ Nat_econ corr

Climate zone City Energy consumption (kWh/m2) % Econ. % Hybrid

Strategy Cooling Heating Fan

1A Miami, FL Mech_min 144 4.85 102 1.55 e e

2A Houston, TX Mech_min 105 1.55 1.08 e e

2B Phoenix, AZ Mech_min 111 0.711 1.64 e e

3A Atlanta, GA Mech_min 71.1 6.94 0.852 e e

3B El Paso, TX Mech_min 84.3 2.16 1.42 e e

3C Los Angeles, CA Mech_min 61.1 0.137 0.951 e e

4A Philadelphia, PA Mech_min 51.5 18.4 0.658 e e

4B Albuquerque, NM Mech_min 56.7 6.22 0.909 e e

4C Seattle, WA Mech_min 22.9 13.3 0.384 e e

5A Boston, MA Mech_min 38.2 21.4 0.640 e e

5B Salt Lake City, UT Mech_min 49.6 13.5 0.836 e e

6A Milwaukee, WI Mech_min 40.3 28.3 0.587 e e

6B Billings, MT Mech_min 42.5 25.7 0.679 e e

7B Fargo, ND Mech_min 41.2 47.9 0.631 e e

Climate zone City CO (ppm)

Climate zone City Case NO2 (ppb) O3 (ppb) PM2.5 (mg/m3)

MERV8 MERV11 MERV16

Pollutant Mechanical ventilation

I/O at Mech_min ki I/O at Mech_econ R2 RMSE

Pollutant Natural ventilation

I/O Nat_min ki I/O at Nat_econ R2 RMSE

34. discussion 109e27.

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.