Building and Environment 119 (2017) 140e152

Contents lists available at ScienceDirect

Building and Environment

journal homepage: www.elsevier.com/locate/buildenv

A new method to assess infiltration rates in large shopping centers

Yuchen Shi a, Xiaofeng Li a, b, *, Haoru Li a
a
Department of Building Science, School of Architecture, Tsinghua University, Beijing, 100084, China
b
Beijing Key Laboratory of Indoor Air Quality Evaluation and Control, Beijing, 100084, China

a r t i c l e i n f o a b s t r a c t

Article history: Large shopping centers integrate multiple functions within a single building, which typically results in
Received 16 December 2016 high energy consumption. Building energy performance and indoor air quality are significantly depen-
Received in revised form dent upon ventilation, which is influenced by uncontrolled air leakage (i.e. air infiltration) across the
13 March 2017
building exterior. In general, shopping centers in China are in a state of significant negative pressure due
Accepted 15 April 2017
Available online 18 April 2017
to the sub-optimal management of air conditioning operational systems. Therefore, an accurate esti-
mation of the air infiltration rate is necessary to instruct operational management and to improve energy
performance. Common infiltration rate testing methods often fail to continuously measure infiltration
Keywords:
Deposition
rates in large inner spaces for extended time periods. This paper presents a new methodology to assess
Field test the infiltration rate using the mass balance of particulate matters whose aerodynamic diameters are
Infiltration rate smaller than 2.5 mm (PM2.5). The deposition, generation and penetration of indoor PM2.5 are also
Shopping centers considered. A field test is conducted in a large shopping center in Beijing (China), with the new method
PM2.5 applied to assess the infiltration rate. The results are verified by the airflow mass balance throughout the
entire building. Also, the methodology helps to determine the main sources of indoor PM2.5 as well as to
instruct how indoor PM2.5 concentration can be effectively reduced.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction
The large shopping center is an emerging type of large public
building being built in cities across China. It integrates multiple
functions within a single building, such as shopping, dining and
entertainment. Current data shows that the total construction area
for commercial complexes with large shopping centers will
approach 100 million square meters by 2019 [1]. In general, large
shopping centers are characterized by higher energy consumption
for air conditioning than other types of public buildings [1]. The
rapid development of large shopping centers is accelerating energy
consumption for air conditioning systems. Therefore, it is impor-
tant to identify viable ways to improve energy performance.
Uncontrolled air leakage across a building exterior is a primary
reason for the excessive energy consumption of air conditioning
systems [2]. In the United States of America, a simulation study
found that improving airtightness of building envelopes resulted in
a 3e36% energy saving potential [3]. In the cold regions of China,
when the infiltration rate was reduced from 0.3 to 0.03 ach, the
* Corresponding author. Department of Building Science, School of Architecture,
Tsinghua University, Beijing, 100084, China.
E-mail address: xfli@mail.tsinghua.edu.cn (X. Li).
annual heating and cooling load in a public building was reduced by
more than 40% [4]. There are various other negative consequences
of air infiltration in public buildings, including compromised ther-
mal comfort and adverse effects to normal operation of the venti-
lation system [5]. Also, indoor PM2.5 concentrations are highly
correlated with outdoor concentration levels because of air infil-
tration [6]. This means that infiltration directly influences indoor
air quality, especially when the outdoor air is highly polluted.
At present, most shopping centers in China are in a state of
significant negative pressure and have large amounts of air infil-
tration. Fig. 1 shows the field test results of 16 shopping centers
throughout China [7]. Here, 87.5% of the centers are in a state of
negative pressure, with an average infiltration rate of 150
103 m3/
h. The deficiency of make-up air in kitchens is the primary reason
for the huge air infiltration. E-commerce has had a significant
impact on the retail industry, with catering services now becoming
the main way to attract customers into large shopping centers. This
results in increased cooking in open kitchens and requires larger
amounts of exhaust air. Based on China's current standards, make-
up air flow should be 80e90% of the exhaust air flow in a com-
mercial kitchen [8]. However, the majority of restaurant owners in
shopping centers cannot appropriately operate the make-up air
system. As such, open kitchens and related services would cause

http://dx.doi.org/10.1016/j.buildenv.2017.04.011
0360-1323/© 2017 Elsevier Ltd. All rights reserved.
 Y. Shi et al. / Building and Environment 119 (2017) 140e152
Fig. 1. Field test results of 16 shopping centers throughout China.
serious negative pressure throughout the entire building, resulting
in a significant amount of air infiltration and excessive energy
consumption. Hence, the assessment and monitoring of infiltration
rates in shopping centers are necessary to instruct the operational
management of air conditioning systems.
There are two standard methods to measure the infiltration
rate: fan pressurization and tracer gas. The fan pressurization
method involves the mechanical pressurization or depressurization
of a building, then measuring the resulting air flow rates at given
indoor-outdoor static pressure differences [9]. As a result, it does
not measure air infiltration rates under normal climate conditions
and building operation. The tracer gas method covers techniques
that use tracer gas dilution for determining the air change rate of a
single zone. These techniques include concentration decay, con-
stant injection and constant concentration [10]. The decay method
involves seeding the zone with a tracer gas, ensuring a uniform
concentration within the zone and then analyzing the decay of the
concentration over a given time period. Other methods use the
constant injection of a tracer gas or a constant concentration. Heidt
and Werner [11] concluded that the decay method was more
suitable for measuring air exchange rates of relatively tight spaces,
whereas the constant injection method was more suitable for
‘leaky’ spaces. The tracer gas method has been used in office
buildings [12e14], residential buildings [15e18], student dormi-
tories [19] and motor vehicles [20]. However, for large shopping
centers, the tracer gas method does not seem suitable due to the
huge consumption of tracer gas and also the high implementation
costs. Sulfur hexafluoride (SF6) and carbon dioxide (CO2) are two
kinds of commonly used tracer gases. Taking CO2 as an example,
under the CO2 constant injection method in a large building with a
volume of 4
105 m3, the ASHRAE suggests that the difference of
indoor and outdoor CO2 volume fraction be no less than 700 ppm
[21]. Here, according to Eq. (1) [22]:
FCO2
N¼
ðc1  ce ÞVZ
whereN is the air change rate, h1; FCO2 is the CO2 release rate, m3/
h; c1 is the volume fraction of indoor CO2, ppm; ce is the volume
fraction of outdoor CO2, ppm; andVZ is test area volume, m3. Now,
the consumption of CO2 can be calculated. Assuming the air change
rate is 1.5 h1 in a large building, then 420 m3/h CO2 gas or 750 kg/h
dry ice would be consumed in each experiment. Here, the storage
and release of such amounts of tracer gas is difficult. Furthermore,
141
indoor CO2 concentration is significantly influenced by human
release and SF6 is a very strong greenhouse gas. Hence, these two
tracer gases are not suitable for assessing and monitoring the
infiltration rate in shopping centers.
Another method to assess the infiltration rate is using the indoor
and outdoor airflow mass balance. It's based on the principle where
a building is in steady state, when airflow coming into the building
equals to that going out of the building. Thus, the infiltration rate
can be calculated by knowing the inlet and outlet air flow rates.
However, when this method is used in large shopping centers the
workload of measurement will be tremendous, because there are
over 50 air handling units (AHUs) and over a hundred outlets. The
measurement will therefore require a large number of testers or
take days, and the operating state of the HVAC system may change
during the process. Hence, this method is not applicable in our
study.
This paper proposes a new calculation method to assess the
infiltration rate in large shopping centers based on the mass bal-
ance of particulate matters (PM). Given the ubiquity of PM in the
outdoor environment and being cost-free, this method is able to
avoid the disadvantages of the tracer gas and airflow mass balance
method. However, unlike commonly used tracer gases, the depo-
sition, resuspension and penetration of indoor particulate matters
cannot be neglected. According to previous studies, between the
diameter range of 0.3e10 mm, larger particles are easier to be
deposited and resuspended and more difficult to penetrate
[23e28], which will influence the accuracy of the new method.
Thus, smaller particles are a better choice for our study. Besides,
since people pay more and more attention to the hazy weather, the
measuring instruments and on-line monitoring of PM2.5 are uni-
versal and of mature technology. Hence, PM2.5 are chosen to
establish the mass balance. After the new method is established, a
field test of the infiltration rate is conducted in a large shopping
center in Beijing (China). Then the airflow mass balance is used to
verify the result calculated by the PM2.5 mass balance method. This
(1)
new method makes the assessment of infiltration rates in large
shopping centers more feasible from both an implementation and
costs perspective.
2. Methodology
2.1. PM2.5 mass balance
Taking internal building space as a control body, the airflows
142 Y. Shi et al. / Building and Environment 119 (2017) 140e152
coming into and going out from a shopping center can be catego-
rized as follows: mechanical air supply from the HVAC system,
including the fresh air and the return air; mechanical air exhaust;
air infiltration; and air exfiltration.
There are three ways for PM2.5 to enter a shopping center:
particles in the supply air from the HVAC system; particles trans-
ported through air infiltration; and particles originating indoors.
The HVAC system in a building has multiple AHUs with different
filter efficiencies; as such, the PM2.5 concentration of airflows
supplied by different AHUs varies. PM2.5 concentration of infil-
tration airflow is affected by outdoor PM2.5 levels and the pene-
tration coefficient. When the outdoor-originating PM2.5 in these
airflows go indoors (along with the PM2.5 that originate indoors), a
portion are deposited onto the surfaces in contact with air, and the
remainder exit the building through exhaust air and air exfiltration.
It's notable that PM2.5 remained in the ventilation ducts are not
taken into consideration in this method, because it has been
studied by Wu and Zhao [29] that for particles whose diameter is
smaller than 3 mm, 99.65% will penetrate through the ducts and get
indoors; but for particles whose diameter is larger than 5 mm, the
filtration of the duct cannot be ignored.
The following assumptions are made: inlet and outlet airflows
are constant under a given operating mode for the HVAC system;
PM2.5 concentration of the outdoor air and the supply air is con-
stant for a given time period; indoor temperature and PM2.5 con-
centration are homogeneous. Under these assumptions, the indoor
PM2.5 concentration will eventually come to a steady state. The
steady-state PM2.5 mass balance in the building interior (Fig. 2) is
mathematically expressed as:
X
n   2.2. PM2.5 deposition, generation and penetration
Qms;i cms;i þ PQinf cout þ F ¼ Qme þ Qexf cin þ D
i¼1
where Qms;i is the flow rate of the mechanical air supply from
different AHUs in the HVAC system, m3/h; cms;i is the corresponding
PM2.5 concentration in the supply airflows, mg/m3; n is the number
of on-going AHUs; P is the fraction of PM2.5 that penetrate the
building envelope (i.e., penetration coefficient), dimensionless; Qinf
is the air infiltration rate, m3/h; cout is the outdoor PM2.5 concen-
tration, mg/m3; F is the indoor PM2.5 generation rate, mg/h; Qme is
the total flow rate of mechanical air exhaust, m3/h; Qexf is the air
exfiltration rate, m3/h;cin is the indoor PM2.5 concentration, mg/m3;
D is the indoor PM2.5 deposition rate, mg/h.
Fig. 2. Airflows in a shopping center.
Given that the density of indoor and outdoor air is different due
to temperature differences, the airflow mass balance in the com-
bined building (Fig. 2) is expressed as:
X
n
Qms;i rms;i þ Qinf rout ¼ Qexf rin þ Qme rin (3)
i¼1
where rms;i is the density of supply air from different AHUs in the
HVAC system, kg/m3; and rout and rin are the outdoor and indoor
air densities, kg/m3, respectively. Air is a kind of ideal gas; hence,
according to the Ideal Gas Law and substituting Eq. (3) into (2)
yields:
P
n  
Tin
Qms;i Tms;i cin  cms;i þ D  F
i¼1
Qinf ¼ (4)
Pcout  TTout
in
cin
where Tin and Tout are the indoor and outdoor air temperatures, K,
respectively, and Tms;i is the temperature of supply air from
different AHUs in the HVAC system, K.
The air flow rates in Eq. (4) are affected by the operation mode of
the HVAC system; they are constant under specific operation
modes and easily measured. Temperature and PM2.5 concentration
are also easily measured at the scene or monitored in the building
automation system (BAS). Then, in order to calculate the infiltration
rate, the deposition, generation and penetration of PM2.5 should be
determined.
(2)
2.2.1. PM2.5 deposition
The indoor PM2.5 deposition rate is determined by deposition
area, deposition velocity and indoor PM2.5 concentration. Deposi-
tion areas include both smooth and rough surface areas, which
accommodate the variations in deposition velocities for different
surfaces. The expansion equation of the indoor PM2.5 deposition
rate in Eq. (4) is expressed as:
D ¼ ðAu vdu þ Av vdv þ Ar vdr Þcin (5)
whereAu and Av are the smooth surface areas for particle deposi-
tion, m2; u and v are the subscripts for horizontally-upward and
 Y. Shi et al. / Building and Environment 119 (2017) 140e152
vertical surfaces, respectively; Ar is the rough surface area for
particle deposition, m2; and vdu , vdv andndr are the corresponding
deposition velocities, m/h. In Eq. (5), horizontally-downward
smooth surfaces are not included because its PM2.5 deposition
velocity is near to 0 (i.e., negligible); this was previously proven by a
study at Berkeley [23].
2.2.1.1. Deposition velocity. The study at Berkeley showed that the
deposition velocity onto a smooth surface was influenced by par-
ticle size and flow conditions near the surface [23]. The researchers
drew curves describing how particle size and air velocity affected
the deposition velocity based on simulation and experiment. The
“deposition velocity-particle diameter” curves of vertical and
horizontally-upward surfaces were different. The current method
uses these results to calculate the deposition velocity of smooth
surfaces.
Another study established a model room for simulations,
obtaining the relationship curve of area-weighted deposition ve-
locity and particle diameter under different surface roughness [24].
The area-weighted deposition velocity is the deposition velocity
weighted by the areas of the vertical, horizontally-upward and
horizontally-downward surfaces. The current method uses these
results to calculate the deposition velocity of rough surfaces.
Assuming that PM2.5 diameter distribution is identical for the
entire inner space, then the deposition velocities of different sur-
faces are calculated by the following formulas:
X
m are obtained from the relevant CAD drawings; the additional area
ndu ¼ vdu;i xi (6)
i¼1
X
m is determined.
ndv ¼ vdv;i xi (7)
i¼1
X
m
ndr ¼ vdr;i xi (8)
i¼1
where vdu;i , vdv;i and vdr;i are the deposition velocities of different
surfaces in variable PM2.5 diameter intervals (fixing air velocity
and surface roughness), m/h; xi is the weight percentage of PM2.5
in the corresponding diameter interval; and m is the number of
PM2.5 diameter intervals. Now, after indoor PM2.5 diameter dis-
tribution is measured, the deposition velocities of different surfaces
can be calculated.
2.2.1.2. Surface area for particle deposition. As mentioned earlier,
deposition areas include both smooth and rough surface areas. As
for shopping centers, smooth surface areas includes floors, walls
and furniture surfaces; rough surface areas are mainly provided by
clothing surfaces in clothing stores and are small enough to be
ignored in other areas. Besides, a large shopping center is divided
into two parts: stores and public areas. As such, the surface areas for
deposition are expressed as:
k 
X 
Au ¼ Af ;i þ lu Af ;i þ Af ;p (9)
i¼1
k 
X 
Av ¼ Aw;i þ lv Af ;i þ Aw;p (10)
i¼1
143
X
k
Ar ¼ lr Af ;i (11)
i¼1
where k is the number of stores;Af ;i and Aw;i are the floor and
vertical wall areas in each store, m2, respectively; Af ;p and Aw;p are
the floor and vertical wall areas in public spaces, m2; lu is the
additional horizontally-upward-surface factor, which describes the
additional horizontally-upward deposition area provided by the
furniture per unit floor area in each store; lv is the additional
vertical-surface factor, which describes the additional vertical
deposition area provided by the partition walls and furniture per
unit floor area in each store; lr is the additional rough-surface
factor, which describes the rough-surface deposition area pro-
vided by the clothing surfaces per unit floor area in each store.
2.2.1.3. Indoor PM2.5 deposition rate. Substituting Eqs. (6)e(11)
into Eq. (5) yields:
(" # "
k 
X  k 
X 
D¼ Af ;i þ lu Af ;i þ Af ;p vdu þ Aw;i þ lv Af ;i
i¼1 i¼1
# )
X
k
þ Aw;p vdv þ lr Af ;i vdr cin
i¼1
(12)
For a given shopping center, the stores and public usage areas
factors are surveyed and measured in typical stores. Now, the sur-
face areas for particle deposition can be calculated. Combining
Section 2.2.1.1 and Section 2.2.1.2, the indoor PM2.5 deposition rate
2.2.2. PM2.5 generation and penetration
The major indoor sources of PM2.5 are smoking, cooking and
particle resuspension induced by light human activities (especially
walking) [25e34]. In most countries, smoking is prohibited in
public places (such as shopping centers). Cooking activities involves
the combustion of fuel and gas, which emits particles. However,
these particles are directly exhausted to outdoors in the kitchens.
Therefore, the indoor PM2.5 generation rate caused by smoking and
cooking in a large shopping center is zero. Then, particle resus-
pension needs to be considered. To evaluate resuspension quanti-
tatively, Slinn [35] defined the resuspension rate as:
R
r¼ (13)
L
where r is the resuspension rate, h1; R is the resuspension flux,
mg/(m2$h); and L is the floor dust loading, mg/m2 Thus, the indoor
PM2.5 generation rate can be expressed as:
F ¼ rLA (14)
where A is the surface area for human walking and particle resus-
pension in a shopping center, m2.
Previous researches provided the resuspension rates induced by
indoor human activities for particles with different diameters
[27,36,37]. The floor dust loading can be measured using a vacuum
cleaner with HEPA filter to catch smaller particles [38]. The surface
area for particle resuspension is obtained from CAD drawings.
Then, the indoor PM2.5 generation rate is determined. Its magni-
tude will be discussed in detail in Section 3.2.3, and finally it's
neglected compared to the indoor PM2.5 deposition rate.
The air infiltration in a large shopping center occurs through
144 Y. Shi et al. / Building and Environment 119 (2017) 140e152
glass curtain walls and windows with large gaps and frequently
opened doors for people coming and going. Hence, the penetration
coefficient is set as 1, which is also recommended by Thatcher and
Thornburg [36,39].
2.3. Summary of the methodology
From the above, the infiltration rate in large shopping centers
can be calculated by Eq. (4) using the new method. The following is
a summary of how to obtain each parameter mentioned in the new
method as well as some simplifications for the field test.
For a given shopping center, air flow rates are constant under
specific operation modes of the HVAC system, which allows for
measurements after shifting operation modes.
Existing research at Tsinghua University (Beijing, China) showed
that during the spring and summer seasons, the diameter distri-
bution of PM2.5 in Beijing remained relatively stable even though
the general concentration was different [40]. As such, diameter
distribution can be measured ahead of time instead of being
measured at the scene. Then the deposition velocities of the
different surfaces can be obtained in advance, which simplifies the
calculations and field test.
Typically, when the usage area in the shopping center does not
change significantly, the surface area for particle deposition can be
treated as an invariable parameter. In fact, the area won't change by
much in one year's time, because generally the leases for stores are
on an annual basis.
Finally, the infiltration rate can be calculated after the related
temperature and PM2.5 concentrations are measured. Since tem-
perature and PM2.5 concentration sensors are accurate and inex-
pensive, the measurement is simple. The authors note that, for
many shopping centers in China, the temperature and PM2.5 con-
centration are monitored by the building automation system (BAS),
allowing real-time values to be easily obtained. This also allows the
infiltration rate to be monitored in real-time. As such, the new
method ensures routine management for infiltration rate moni-
toring in large shopping centers.
3. Validation
3.1. Validation of the assumptions
Equations (2)e(4) in Section 2.1 are established under three
assumptions.
For assumption 1, the inlet and outlet airflows are constant
during the process of reaching the PM2.5 mass balance because the
air valves are not regularly adjusted under a given operating mode
of the HVAC system.
For assumption 2, the outdoor PM2.5 concentration is some-
what dependent on outdoor climate conditions, especially wind
Fig. 3. Measurement results for outdoor PM2.5 concentration.
speed [41]. Also, the average wind speed of a typical year in Beijing
is 2.4 m/s, with a wind speed lower than 2 m/s for 62% of the time
[42]. Besides, with the rapid development of China's urban con-
struction in recent years, large quantity of high-rise buildings are
built in cities. Wind speed is significantly reduced due to the in-
fluence of tall buildings and friction when air current passes urban
areas, leading to an increase in horizontal static wind [43]. The
horizontal static wind is not conducive to the spread and dilution of
pollutants in the atmosphere. Hence, it's one of the main reasons
for the formation of hazy weather and can cause a steady outdoor
concentration of PM2.5 for a period of time [44]. Fig. 3 shows the
measurement results for outdoor PM2.5 concentration from 05:00
December 4th, 2016 to 05:00 December 5th, 2016 in Beijing, China.
Time periods for steady outdoor concentration are frequently
encountered. During these periods, PM2.5 mass balance can be
achieved. The PM2.5 concentration of the supply air is dependent
on the outdoor concentration and filter efficiency in the AHUs.
Since filter efficiency does not significantly change for short time
periods, the concentration of the supply air is also steady during the
process of reaching the PM2.5 mass balance.
For assumption 3, air change rate and air infiltration through
doors and windows are two factors that influence the homogeneity
of indoor temperature and PM2.5 concentration. A high air change
rate contributes to a healthy indoor environment; as such, large
amounts of supply air are required in shopping centers, which also
contributes to indoor airflow distribution and promotes the ho-
mogeneity of temperature and particle concentration. However, the
opening of doors and windows brings in outside air, which has
different temperatures to indoor air as well as a higher PM2.5
concentration. Hence, it affects the temperature and PM2.5 con-
centration homogeneity of the nearby areas. In order to maintain a
good indoor environment, doors and windows are usually closed in
a large shopping center. Also, temperatures throughout a large
shopping center are centrally monitored and controlled in order to
be close to the set point; this ensures a thermally comfortable
environment. Hence, indoor temperature and PM2.5 concentration
can be regarded as homogenous.
In Eq. (4), temperature is a correction term of the air density and
has little effect on the result; as such, its homogeneity is only
theoretically analyzed. However, in order to verify the homogeneity
of indoor PM2.5 concentration, two field tests were conducted in a
large shopping center in Beijing (China). The shopping center had a
total area of 87, 000 square meters, with three floors aboveground
and one underground floor. The building was multi-functional,
incorporating retailing shops, restaurants, a supermarket and
other businesses. The three aboveground floors were connected by
a tall atrium, but the underground floor had no atrium connection
to the first floor nor were there doors or windows to provide
exposure to outside air. As such, infiltration did not influence the
PM2.5 concentration in the underground floor. Therefore, the three
 Y. Shi et al. / Building and Environment 119 (2017) 140e152
aboveground floors of the shopping center were treated as the
control volume.
The first field test acted as a pre-experiment. It divided each
floor into 45 to 55 districts (based on area size), then arranged a
measure point at each district. The PM2.5 concentration was
measured by a Portable Explosion-proof Dust Meter (TSI, Model
AM510) at each measure point. The height of the measure points
was 1.5 m which corresponded to the breathing zone when people
were standing. Fig. 4 shows the arrangement plans and results. The
pentagram stands for the measure point, with the number near it
representing the serial number of each point. The average con-
centration of the three floors was 62 mg/m3, which is marked
beneath each image (Fig. 4). The green pentagrams represent
measure points whose concentration was between 57 and 67 mg/m3
(average concentration ± 5 mg/m3). Blue pentagrams represent
measure points whose concentration was within 52e57 mg/m3 and
67e72 mg/m3 (average concentration ± 5 mg/m3 ~ average
concentration ± 10 mg/m3). Red pentagrams represent measure
points whose concentration was within 47e52 mg/m3 and
72e77 mg/m3 (average concentration ± 10 mg/m3 ~ average
concentration ± 15 mg/m3). The different colors reflect the PM2.5
concentration homogeneity throughout the entire building.
Fig. 4(a) and (b) shows that the PM2.5 concentration around the
atrium was over 70 mg/m3, with measure points marked as blue or
red. This is because there were three sensor doors near the atrium
on the 1st floor; when people passed by, the doors automatically
opened. For areas near the atrium on the 1st and 2nd floors, the
concentration was influenced by the outdoor concentration, which
was 110 mg/m3 on the measuring day. However, there was no
exhaust outlets near the atrium and only smooth surfaces for
deposition; therefore, according to Eq. (5), the concentration non-
uniformity of this area did not significantly affect the calculation
result of infiltration.
For measure points on the 3rd floor (Fig. 4(c)), the PM2.5 con-
centration of the first three points was much higher than the others
(over 85 mg/m3). This is because there was a balcony (Fig. 4(c), black
dotted line) with open doors near those measure points; hence, the
concentration in this area was also affected by the outdoor PM2.5
concentration. However, because the influenced area was small,
these three points were omitted when calculating the average
concentration and, therefore, also excluded from the subsequent
analyses.
In summary, Fig. 4 shows that the proportion of area where the
PM2.5 concentration unevenness was over 15% was less than 5%; as
such, it had little effect on the infiltration calculation. Therefore, the
PM2.5 concentration inside the test building was treated as
homogeneous.
To further explore the homogeneity of PM2.5 concentration and
validation of the measurement, one additional field test of con-
centration homogeneity was conducted and the results are shown
in Fig. 5. This time each floor was divided into 74 to 78 districts and
more measure points were added. The average concentration of the
three floors was 57 mg/m3 and the outdoor PM2.5 concentration
was 95 mg/m3. Obviously, the results demonstrates a better uni-
formity of indoor PM2.5 concentration.
To verify the concentration homogeneity in the vertical direc-
tion, two measure points were selected in each floor to conduct the
tests. The story height was 3.65 m in the test building and PM2.5
concentrations were measured at 0 m, 0.5 m, 1 m … 3.0 m, 3.5 m at
each measure point. The results are presented in Fig. 6 and they
show no significant difference of PM2.5 concentration in the ver-
tical direction.
Hence, assumption 3 of Eq. (2)e(4) is valid.
145
3.2. Methodology validation
In order to verify the viability of the new method, another field
test was conducted at the same test building.
3.2.1. Deposition velocity
During the field test, the particle diameter distribution of the
indoor PM2.5 was measured by an Optical Particle Size Spectrom-
eter (TSI, Model 3330). Table 1 lists the measurement results of the
PM2.5 diameter distribution and the data of deposition velocities
under different particle size, which were obtained from Alvin C. K
[23]. and Zhao [24]. Equations (6) and (7) calculated the deposition
velocities of the horizontally-upward and vertical smooth surfaces,
which were 0.1319 and 0.0006m=h, respectively. The area-weighted
deposition velocity of the rough surfaces calculated by Eq. (8) was
0.2145m=h. Table 2 lists the calculated results.
3.2.2. Surface area for particle deposition
Retail stores in the tested shopping center were classified into
three categories: restaurants, clothing stores and others. The three
additional area factors, lu , lv and lr , were different for each cate-
gory. Fourteen stores in the tested shopping center were selected to
measure the additional area factors, including four restaurants, five
clothing stores and five other stores. Table 3 lists the average values
of each additional area factor. The areas of the vertical walls and
floors were measured from the CAD drawings. Equations (9)e(11)
calculated the different surface areas for particle deposition.
Table 4 lists the results.
3.2.3. Indoor PM2.5 generation rate
According to previous researches, for particles whose diameter
was smaller than 5.0 mm, the resuspension rate induced by indoor
human activities was no more than 1.8
103 h1 [27,36,37]. As
such, the resuspension rate of PM2.5 is obviously lower than that.
Besides, the maximum value of resuspension rate in the above re-
searches was obtained by Qian [27] after 30 min of human activities
in a chamber with floor area of 17.9 m2. While, the occupant density
is very small in a large shopping center. Fig. 7 shows the hourly
occupant density of the tested shopping center in the National Day
holiday (Oct. 1st e7th) of 2014. Overall, the occupant density was
lower than 0.070 m-2 and the average occupant density was
0.033 m-2. In other words, averagely there was only one person per
30 m2 in the building. Thus, the resuspension rate would be even
smaller.
The floor dust loading was measured in the test building using a
vacuum cleaner with HEPA filter before a routine clean. The mass
discrepancy of the HEPA filter before and after dust collection was
measured and the calculated dust loading was 0.0003 g/m2. The
dust loading was quite small because shopping centers were swept
frequently by cleaners.
The surface area for particle resuspension was obtained from
CAD drawings, whose value was 60789 m2. Hence, the indoor
PM2.5 generation rate calculated by Eq. (14) was no more than
3.28
104 mg/h, which was only 1.7% of the indoor PM2.5 deposi-
tion rate. In fact, the ratio was much lower than 1.7% according to
the above analysis.
Based on the above reasons, PM2.5 resuspension is neglected
and the indoor PM2.5 generation rate is set as zero in the proposed
method.
3.2.4. Infiltration rate
After particle deposition and generation were determined, the
infiltration rate was calculated using the PM2.5 mass balance
equation. Table 5 lists the input parameters in the calculation
process.
146 Y. Shi et al. / Building and Environment 119 (2017) 140e152

Fig. 4. Measure points layout of each floor in the first field test.
 Y. Shi et al. / Building and Environment 119 (2017) 140e152 147

Fig. 5. Measure points layout of each floor in the second field test.
148 Y. Shi et al. / Building and Environment 119 (2017) 140e152
Fig. 6. PM2.5 concentration homogeneity in vertical direction.
The mechanical air supply from the HVAC system included fresh
air and return air. The field test was conducted in the transition
season (March 27, 2015), when the building was not heated or
cooled. As such, the fresh air temperature was equal to the outdoor
air temperature and the return air temperature was equal to the
indoor air temperature. The air flow rates were measured by a
thermal balloon anemometry (Beijing Detector Instrument Co., Ltd.
Model D30JS).
The PM2.5 concentration of the mechanical air supply was
measured in the AHUs after the filters by a Portable Explosion-
proof Dust Meter (TSI, Model AM510). The values were between
94 and 150mg=m3 because filter efficiency was different in different
AHUs.
According to the parameters above, the infiltration rate calcu-
lated by the PM2.5 mass balance equation was209
103 m3 =h,
indicating that the building was under significant negative
pressure.
In order to verify the accuracy of the above result, the airflow
mass balance was used to calculate the air infiltration rate. The flow
rates of all exhaust outlets in the test building were measured to
obtain the total flow rate of the mechanical air exhaust, which
yielded a value of403
103 m3 =h. The exfiltration rate was close to
0 because the building was under negative pressure, which was
verified by testing the air flow direction at the openings. Table 5
lists the total flow rate of the mechanical air supply and the tem-
perature of the airflows. Now, according to the airflow mass balance
in Eq. (3), the outdoor airflow infiltrating rate was 187
103 m3 =h.
Table 6 lists the infiltration rates calculated by the airflow and
PM2.5 mass balances. The relative error between the results was
12%, which was within the acceptable error range. This means that
the proposed method would be feasible in practical applications.
3.2.5. Sensitivity analysis
In Eq. (4), all the parameters can be measured with acceptable
value of accuracy except the deposition rate. So sensitivity analysis
is performed for the determining factors of deposition rate: surface
areas for particle deposition and different deposition velocities.
Surface areas for particle deposition are determined by the usage
area in shopping centers, which, as mentioned earlier, won't
change much in one year's time. In the proposed method the area
values were obtained from CAD drawings and the additional area
factors were measured in typical stores. Overall, its variation range
was ±5%. Studied and reviewed by Lai [23,45], the variation range of
deposition velocity on horizontally-upward smooth surfaces was
0.0542e0.2125m=h; and the variation range of deposition velocity
on vertical smooth surfaces was 0.0002e0.0019 m=h depending on
the friction velocity. Researched by Zhao [24], the variation range of
area-weighted deposition velocity on rough surfaces was
0.0565e0.7200 m=h depending on the surface roughness.
Based on the variation ranges above, the output of sensitivity
analysis is listed in Table 7. It's obvious that infiltration rate is the
most sensitive to the deposition velocity on horizontally-upward
smooth surfaces, because the area of horizontally-upward smooth
surfaces for particle deposition is the biggest among all deposition
surfaces; and the result is the second most sensitive to the area-
weighted deposition velocity on rough surfaces due to its wide
variation range. Nevertheless, the relative error is no more than 25%
when parameters vary in their ranges. Thus, the variations of
determining factors for deposition rate are all considered in the
sensitivity analysis.
4. Discussion
The traditional tracer gas method is not suitable to assess the
infiltration rate of shopping centers with large inner space due to
significant consumption of tracer gas and high implementation
costs. The airflow mass balance method needs large amount of
measurement workload when used in large buildings. Based on the
PM2.5 mass balance, the proposed method makes it feasible to
assess the infiltration rate in large shopping centers. Although the
deposition, generation and penetration of PM2.5 are treated in
detail, the measurement error is still approximately 15e20%. As
such, further research is required to enhance measurement accu-
racy. However, considering that the measuring error for airflow in
ducts or outlets is about 5e10% [46,47], it could be said that the new
method makes the infiltration rate assessment in large shopping
 Y. Shi et al. / Building and Environment 119 (2017) 140e152 149

Table 1
Parameters to calculate deposition velocity.

Serial number Particle diameter Weight percentage Smooth surface Rough surface

(mm) xi Horizontally upward Vertical Area-weighted

(%)
vdu;i vdv;i vdr;i
(m=h) (m=h) (m=h)

1 0.300e0.374 21.68 0.0202 0.0011 0.1980

2 0.374e0.465 17.52 0.0288 0.0008 0.1980
3 0.465e0.579 12.45 0.0360 0.0006 0.1980
4 0.579e0.721 7.00 0.0576 0.0006 0.1980
5 0.721e0.897 5.18 0.0720 0.0005 0.1980
6 0.897e1.117 7.26 0.1008 0.0004 0.2304
7 1.117e1.391 3.12 0.1548 0.0004 0.2336
8 1.391e1.732 6.01 0.2160 0.0003 0.2369
9 1.732e2.156 13.33 0.3600 0.0002 0.2466
10 2.156e2.500 6.45 0.5760 0.0002 0.2628

Table 2 centers practicable within an acceptable range of error.

Deposition velocities of different surfaces. Under the proposed method, the infiltration rate of shopping
Deposition velocity (m=h) Values centers could be supervised in real-time by monitoring indoor
vdu 0.1319
temperature and PM2.5 concentration, which avoids excess energy
vdv 0.0006 consumption caused by excess air infiltration. This method could
vdr 0.2145 function as a routine means for shopping centers to instruct oper-
ational management and improve air conditioning energy perfor-
mance. The proposed method also informs the reduction of indoor
Table 3 PM2.5 concentration for large shopping centers. According to Eqs.
Three additional area factors. (4) and (5), indoor PM2.5 concentration is expressed as:
Category lu lv lr
P
n
Restaurants 0.44 0.42 0 Qinf Pcout þ Qms$i cms;i þ F
Clothing stores 0.03 0.29 0.25 i¼1
cin ¼ (15)
Others 0.25 0.24 0 P
n
Qms;i TTms;i
in
þ Qinf TTout
in
þ ðAr vdr þ Av vdv þ Au vdu Þ
i¼1

Table 4 In order to make the result dimensionless, the equation above is

Surface areas for particle deposition. transformed into:

Surface area (m2 ) Values

P
n
c F
Au 79074 Qinf P þ Qms;i cms;i þ cout
cin i¼1
out
Av 18982 ¼ n (16)
Ar 3273 cout P
Qms;i TTms;i
in
þ Qinf TTout
in
þ ðAr vdr þ Av vdv þ Au vdu Þ
i¼1

This equation shows that there are two ways to reduce indoor

Fig. 7. Hourly occupant density in the test building.

150 Y. Shi et al. / Building and Environment 119 (2017) 140e152

Table 5
Input parameters in the infiltration rate calculation process.

Parameters Symbols Values

Penetration coefficient P 1
Indoor PM2.5 generation rate (mg=h) F 0
Indoor PM2.5 deposition rate (mg=h) D 1.97
106

Indoor air temperature ( C) Tin 20.5

Outdoor air temperature ( C) Tout 19.1
Temperature of the mechanical air supply ( C) Tms;i 19.1 (Fresh air)
20.5 (Return air)

Indoor PM2.5 concentration (mg=m3 ) cin 177

Outdoor PM2.5 concentration (mg=m3 ) cout 230
PM2.5 concentration of the mechanical air supply (mg=m3 ) cms;i 94e150
Pn
Total flow rate of the mechanical air supply (m3 =h) i¼1 Qms;i
215
103

Table 6 PM2.5 concentration: increase filter efficiency (i.e., decrease the

Comparison of infiltration rates calculated by the two methods. c
value of cms;i
out
) and decrease infiltration rate (because the outdoor
Methods Infiltration rate (
103 m3 =h) Relative error PM2.5 concentration is usually higher than the indoor concentra-
Airflow mass balance 187 12% tion). Fig. 8 shows the effect of the air infiltration rate and filter
PM2.5 mass balance 209 efficiency on indoor PM2.5 concentration in the tested shopping
center.

Table 7
Output of sensitivity analysis.

Input Typical value of the input Range of the input Infiltration rate calculated by PM2.5 mass Relative error to the infiltration rate calculated by
parameter parameter parameter balance (
103 m3 =h) airflow mass balance

Au , Av , Ar See Table 4 ±5% 206.8e210.6 10.9e12.9%

vdu 0.1319m=h 0.0542e0.2125m=h 188.0e230.2 0.8e23.4%
vdv 0.0006m=h 0.0002e0.0019m=h 208.7e208.8 11.9e12.0%
vdr 0.2145m=h 0.0565e0.7200m=h 207.0e214.3 11.0e14.9%

Fig. 8. Effect of air infiltration rate and filter efficiency on indoor PM2.5 concentration.
Pn
Qms;i cms;i
h ¼ 1  Pi¼1n
Qms;i cout
i¼1
 Y. Shi et al. / Building and Environment 119 (2017) 140e152
Fig. 8 shows that under given air infiltration rate, the ratio of
indoor and outdoor PM2.5 concentration reduced when the filter
efficiency increased, and that under a given filter efficiency, the
ratio dropped when the air infiltration rate decreased. These results
indicate two specific ways to improve indoor air quality: periodi-
cally clean and replace the filters in the AHUs, and increase the flow
rate of make-up air in kitchens to relieve the significant negative
pressure (in the building) and reduce air infiltration.
5. Conclusion
The assessment of air infiltration rates is important for large
shopping centers that aim to reduce energy consumption and
improve indoor air quality. However, to date, there has been a lack
of effective testing methods. This paper proposes a way to calculate
the infiltration rate in a large shopping center using the PM2.5 mass
balance principle. PM2.5 deposition, generation and penetration
are carefully considered and a field test is conducted to verify the
method. The results confirm that the new method has a viable
degree of accuracy and is suitable for calculating the infiltration
rate, especially in a large shopping center. Furthermore, the method
could be used to monitor the infiltration rate in real-time as well as
instruct the operational management of the HVAC system.
Acknowledgments
This work was supported by the National Natural Science
Foundation of China (No. 51521005).
References
[1] Annual Report on China Building Energy Efficiency [R], Tsinghua University,
Building Energy Research Center, 2014.
[2] F.R.A. Alfano, M. Dell’Isola, G. Ficco, et al., Experimental analysis of air tight-
ness in Mediterranean buildings using the fan pressurization method [J],
Build. Environ. 2012 (53) (2012) 16e25.
[3] S.J. Emmerich, T.P. McDowell, W. Anis, Investigation of the Impact of Com-
mercial Building Envelope Airtightness on HVAC Energy Use [R], US Depart-
ment of Commerce, Technology Administration, National Institute of
Standards and Technology, 2005.
[4] J. Lin, Application Strategy Research of Natural Ventilation in the Three Types
of Climatic Conditions in Shaanxi Province [D], Master Thesis, Xi’an University
of Architecture and Technology, China, 2011, 116 pages. (in Chinese).
[5] S.J. Emmerich, A.K. Persily, T.P. McDowell, Impact of infiltration on heating
and cooling loads in US office buildings [C], in: Proceedings of the 26th IEA
Conference of the Air Infiltration and Ventilation Center, 2005.
[6] Q. Meng, B. Turpin, J. Lee, et al., How does infiltration behavior modify the
composition of ambient PM2.5 in indoor Spaces? An analysis of RIOPA data [J],
Environ. Sci. Technol. 2007 (41) (2007) 7315e7321.
[7] H. Li, X. Li, Field testing of ventilation in large shopping centers in China [C],
in: Proceedings of the 14th International Conference of Indoor Air Quality and
Climate, Ghent, Belgium, July 3-8, 2016, 2016.
[8] Y. Lu, Code for Design HVAC in Practical Use [M], China Buiding Materials
Press, 2008, 2685 Pages. (in Chinese).
[9] ASTM, Standard Test Method for Determining Air Leakage Rate by Fan Pres-
surization [S]. ASTM Standard E779, American Society of Testing and Mate-
rials, Philadelphia, 2010, 2010.
[10] ASTM, Standard Test Method for Determining Air Change in a Single Zone by
Means of a Tracer Gas Dilution [S]. ASTM Standard E741, American Society of
Testing and Materials, Philadelphia, 2010, 2010.
[11] F.D. Heidt, H. Werner, Microcomputer-aided measurement of air change rates
[J], Energy Bldgs 9 (1986) 313e320, 1986.
[12] C. Buratti, R. Mariani, E. Moretti, Mean age of air in a naturally ventilated
office: experimental data and simulations [J], Energy Build. 43 (8) (2011)
2021e2027, 2011.
[13] J.T. Reardon, M.R. Atif, C.Y. Shaw, Tracer gas measurements for ventilation, air
movement and air infiltration in a four-sided atrium office building [J], Int. J.
Vent. 1 (1) (2002) 13e22, 06/2002.
[14] C.A. Roulet, F. Foradini, Simple and cheap air change rate measurement using
CO2 concentration decays [J], Int. J. Vent. 1 (1) (2002) 39e44.
[15] A. Sfakianaki, K. Pavlou, M. Santamouris, et al., Air tightness measurements of
residential houses in Athens, Greece [J], Build. Environ. 43 (4) (2008)
151
398e405, 2008.
[16] T. Johnson, J. Myers, T. Kelly, et al., A pilot study using scripted ventilation
conditions to identify key factors affecting indoor pollutant concentration and
air exchange rate in a residence [J], J. Expo. Anal. Environ. Epidemiol. 2004
(14) (2004) 1e22.
[17] N. Russell, A. Edgar, Air infiltration measurements in a home using a conve-
nient perfluorocarbon tracer technique [J], Environ. Int. 8 (1e6) (1982)
419e433, 1982.
[18] Go Iwashita, Hiroshi Akasaka, The effects of human behavior on natural
ventilation rate and indoor air environment in summer d a field study in
southern Japan [J], Energy Build. 25 (3) (1997) 195e205, 1997.
[19] H. Li, X. Li, M. Qi, Field testing of natural ventilation in college student dor-
mitories (Beijing, China) [J], Build. Environ. 2014 (78) (2014) 36e43.
[20] Wayne Ott, Neil Klepeis, Paul Switzer, Air change rates of motor vehicles and
in-vehicle pollutant concentrations from secondhand smoke [J], J. Expo. Sci.
Environ. Epidemiol. 2008 (18) (2008) 312e325.
[21] ASHRAE, ASHRAE Standard 62.1 Ventilation for Acceptable Indoor Air Quality
[S], American Society of Heating, Refrigerating and Air-Conditioning Engi-
neers, Inc, Atlanta, 2010.
[22] B. Zheng, X. Li, F. Xue, Approach for measuring air changes with dry ice
releasing tracer gas [J], 2013, J. HV&AC 43 (12) (2013) 39e45 (in Chinese).
[23] C. K. Lai Alvin, W. Nazaroff William, Modeling indoor particle deposition from
turbulent flow onto smooth surfaces [J], J. Aerosol. Sci. 31 (4) (2000) 463e476,
2000.
[24] B. Zhao, J. Wu, Particle deposition in indoor environments: analysis of influ-
encing factors [J], J. Hazard. Mater. 2007 (147) (2007) 439e448.
[25] Y. Tian, K. Sul, J. Qian, et al., A comparative study of walking-induced dust
resuspension using a consistent test mechanism [J], Indoor Air 2014 (24)
(2014) 592e603.
[26] Norbert Serfozo, Sofia Eirini Chatoutsidou, Mihalis Lazaridis, The effect of
particle resuspension during walking activity to PM10 mass and number
concentrations in an indoor microenvironment [J], Build. Environ. 82 (2014)
(2014) 180e189.
[27] J. Qian, A. Ferro, Resuspension of dust particles in a chamber and associated
environmental factors [J], Aerosol. Sci. Technol. 42 (2008) 566e578, 2008.
[28] Chun Chen, Bin Zhao, Wanting Zhou, et al., A methodology for predicting
particle penetration factor through cracks of windows and doors for actual
engineering application [J], Build. Environ. 47 (2012) (2012) 339e348.
[29] Jun Wu, Bin Zhao, Effect of ventilation duct as a particle filter [J], Build. En-
viron. 42 (2007) (2007) 2523e2529.
[30] N.C. Jones, C.A. Thornton, D. Mark, et al., Indoor/outdoor relationships of
particulate matter in domestic homes with roadside, urban and rural locations
[J], Atmos. Environ. 34 (2000) (2000) 2603e2612.
[31] Q.Y. Meng, B.J. Turpin, L. Korn, et al., Influence of ambient (outdoor) sources
on residential indoor and personal PM2.5 concentrations: analyses of RIOPA
data [J], J. Expo. Analysis Environ. Epidemiol. 15 (1) (2005) 17e28. PMID:
15138449.
[32] G. Buonanno, L. Morawska, L. Stabile, Particle emission factors during cooking
activities [J], Atmos. Environ. 43 (2009) (2009) 3235e3242.
[33] Marianna Luoma, Stuart A. Batterman, Characterization of particulate emis-
sions from occupant activities in offices [J], Indoor Air 2001 (11) (2001)
35e48.
[34] Iman Goldasteh, Yilin Tian, Goodarz Ahmadi, et al., Human induced flow field
and resultant particle resuspension and transport during gait cycle [J], Build.
Environ. 77 (2014) (2014) 101e109.
[35] W.G.N. Slinn, Parameterizations for resuspension and for wet and dry depo-
sition of particles and gases for use in radiation dose calculations [J], Nucl. Saf.
19 (1978) 205e219.
[36] T.L. Thatcher, D.W. Layton, Deposition, resuspension, and penetration of
particles within a residence [J], Atmos. Environ. 29 (13) (1995) 1487e1497,
1995.
[37] C. Gomes, J. Freihaut, W. Bahnfleth, Resuspension of allergen-containing
particles under mechanical and aerodynamic disturbances from human
walking [J], Atmos. Environ. 41 (25) (2007) 5257e5270.
[38] Jing Qian, Jordan Peccia, Andrea R. Ferro, Walking-induced particle resus-
pension in indoor environments [J], Atmos. Environ. 89 (2014) (2014)
464e481.
[39] J. Thornburg, D.S. Ensor, C.E. Rodes, et al., Penetration of particles into
buildings and associated physical factors. Part I: model development and
computer simulations [J], Aerosol. Sci. Technol. 34 (3) (2001) 284e296, 2001.
[40] Y. Tang, X. Li, Statistics and analysis of PM2.5 concentration categorical par-
ticle diameter measurement [C], in: Proceedings of the 10th International
Green Energy Conference, May 24-27, 2015, Taichung, Taiwan, 2015.
[41] Y. Wang, Y. Yang, Y.W. Xing, Statistical Analysis of the Impact of Meteoro-
logical Factors on Hazy Weather in the Winter of Beijing, North China Uni-
versity of Technology, 2013 (in Chinese), http://sescn.org.cn/.
[42] F. Song, Q. Zhu, R. Wu, et al., Meteorological data set for building thermal
environment analysis of China, in: Proceedings of the 10th International
Building Performance Simulation Association Conference and Exhibition,
Beijing, China, 2007, pp. 9e16.
[43] J.K. Zhou, Analysis of causes and hazards of China's frequent hazy weather,
152 Y. Shi et al. / Building and Environment 119 (2017) 140e152

Open Cybern. Syst. J. 2015 (9) (2015) 1311e1314.
[44] Y.Z. Chen, J.M. Xiong, Analysis on the Causes of Hazy Weather in Nanchang
and the Control Countermeasures, 2014. The Science Education Article Col-
lects, Total.274, April 2014 (A). (in Chinese).
[45] C. K. Lai Alvin, Particle deposition indoors: a review [J], Indoor Air 2002 (12)
(2002) 211e214.
[46] Xiumu Fang, Testing and Measuring Technique of Building Environment (In
Chinese) [M], China Architecture & Building Press, 2009, 2009.
[47] D. Schwenk, Air flow measurement accuracy, in: Conference Proceedings.
Washington D.C: HQUSACE; June 1998, Us Army Corps Of Engineers, Con-
struction Engineering Research Laboratories, 1998.