Infiltration of Ambient PM 2 . 5 through Building Envelope in Apartment Housing Units in Korea

Air pollution due to PM2.5 is of public concern in Korea. Ambient PM2.5 can penetrate indoors through the building envelope, affecting the indoor PM2.5 concentrations. Most people stay indoors for approximately 80% of every day, implying that their primary exposure to PM2.5 could be determined by the indoor air. This study aims to investigate the infiltration of ambient PM2.5 through the building envelop in apartment housing units in Korea. The on-site infiltration test method, by using a blower-door depressurization procedure, was suggested in order to maintain an identical indooroutdoor pressure difference among the tests. On-site experiments were conducted in 11 apartment housing units to estimate the PM2.5 infiltration factors. The results showed that the average infiltration factor of all the test housing units was 0.65 ± 0.13 (average ± standard deviation), with a minimum of 0.38 and a maximum of 0.88. Furthermore, the results from the relation of the building airtightness data to the infiltration factors suggests that a leaky housing unit with high ACH50, or a high specific effective leakage area (ELA), would be more significantly influenced by the ambient PM2.5. The study demonstrated that the suggested infiltration test procedure was useful to assess the infiltration factors in conditions of controlled indoor-outdoor pressure differences in real housing units.


INTRODUCTION
Particulate matter (PM) pollution is considered a major threat, as it is associated with detrimental effects on human health and the quality of life (Stanek et al., 2011;Henschel et al., 2012;Van Erp AM et al., 2012;WHO, 2013).Strong evidence points to the negative effects of short-and longterm exposure to PM on the respiratory and cardiovascular health of humans (Beelen et al., 2008;Samoli et al., 2008).In particular, particles with a diameter of less than 2.5 µm (PM 2.5 ) are of high concern because of their adverse effects on the respiratory system (Pope et al., 2009).PM 2.5 has been used commonly as one of indicators to describe PM.It has been shown that long-term exposure to PM 2.5 is a strong risk factor for mortality (Pope et al., 2002;Lepeule et al., 2012;WHO, 2013).Air pollution due to PM 2.5 in Korea is a matter of significant public concern.According to the Organization for Economic Cooperation and Development (OECD), the average annual concentration of the ambient PM 2.5 levels in Korea is 29.1µg m -3 (OECD, 2015), which is the highest among the OECD countries, where the average level is 14.5 µg m -3 , and is much higher than is the annual guideline limit of 10 µg m -3 set by the World Health Organization (WHO) (WHO, 2006).Ambient PM 2.5 is known to be associated with industry, urbanization, and transport.This implies that the majority of the population in Korea is exposed to high levels of PM 2.5 , and the attendant health risks, as a high percentage of people live in urban city areas.Moreover, the Asian dust event, called yellow dust, affects the Korean Peninsula during the winter and spring seasons (Kim and Park, 2001).Studies have indicated that increased PM 2.5 concentrations were observed during such dust events (Kim and Kim, 2001;Stone et al., 2011), exacerbating the already high levels of ambient PM 2.5 .
In view of the growing concerns on ambient PM pollution, the Korea Environment Corporation, an affiliated agency under the Ministry of Environment, has since 2005 been providing real-time PM 2.5 data online, along with the data on other air pollutants (PM 10 , O 3 , CO, SO 2 , No 2 ).The data on the concentrations of PM 2.5 are collected from 317 sites in 97 areas across the country and are displayed hourly on the AirKorea website (KECO, 2016).AirKorea advises sensitive groups to stay indoors when the hourly average PM 2.5 concentration exceeds 90 µg m -3 for over two hours.A warning is issued when the value exceeds 180 µg m -3 for over two hours, with every person prohibited from going outdoors.However, most people stay indoors for about 90% of every day (Jenkins et al., 1992;Robinson and Nelson, 1995;McCurdy et al., 2000;Klepeis et al., 2001;Schweizer et al., 2007), which means that their primary exposure to PM 2.5 could be determined by the indoor air.Moreover, children and the elderly, especially, probably spend most of their time indoors, particularly in a residential setting.Most of the residential units in Korea rely on natural ventilation, without any filtering system; therefore, closing the windows and doors could be the only solution to counteract the high ambient PM 2.5 conditions.Nevertheless, ambient PM 2.5 can penetrate indoors through the building envelope, affecting the indoor PM 2.5 concentrations (Liu and Nazaroff, 2003).
Numerous studies have been conducted to show the effects of ambient PM 2.5 on the indoor environment (Baek and Kim, 1997;Adgate et al., 2002;Long and Sarnat, 2004;Cao et al., 2005;Meng et al., 2005).For quantification of the infiltration process through building leakage, three parameters have been used widely, i.e., the indoor/outdoor (I/O) ratio (Monn et al., 1995;Abt et al., 2000;Long and Sarnat, 2004;Lunden et al., 2008), the infiltration factor (Lee et al., 1997;Landis et al., 2001;Wallace and William, 2005;Lazaridis et al., 2006;Sarnat et al., 2006;Hoek et al., 2008;Meng et al., 2009), and the penetration factor (Long et al., 2001;Chao et al., 2003;Thatcher et al., 2003;Zhu et al., 2005).The I/O ratio represents the relationship between the indoor and outdoor PM concentrations.The infiltration factor represents the equilibrium fraction of ambient PM that penetrates indoors and remains suspended in the indoor air.The penetration factor expresses the fraction of the particles in the infiltration air that passes through the building envelope.Although the penetration factor is the most relevant parameter for penetration through the envelope, the infiltration factor can be a useful parameter to analyze the total indoor PM 2.5 entering from the outdoor environment (Chen and Zhao, 2011).In most previous studies, experiments have been conducted relevant to the infiltration factor in real buildings in natural air exchange conditions.However, differences in wind direction and speed at the time of measurements could cause variations in the indoor-outdoor pressure difference and the air exchange rate, which makes a direct comparison of buildings difficult.Moreover, the urban areas in Asian countries, where high concentrations of PM 2.5 are of concern, have a high density of high-rise residential buildings, with a wide range of indoor-outdoor pressure differences.Thus, it would be effective to derive the infiltration factor under controlled conditions of indooroutdoor pressure difference.
The aim of this study was to investigate the infiltration of ambient PM 2.5 through the building envelop, which affects the indoor air quality in the apartment housing units, by estimating the infiltration factors in controlled indooroutdoor pressure differences.The on-site infiltration test method, using a blower-door depressurization procedure, was suggested to maintain an identical indoor-outdoor pressure difference among the tests in the different apartment housing units.In addition, the building airtightness was measured to determine the relation between the building airtightness and the infiltration factors.

EXPERIMENTAL METHOD
On-site experiments in 11 apartment housing units in Korea were conducted from December 2015 to April 2016 in order to investigate the infiltration of ambient PM 2.5 through the building envelope, with the infiltration factors being used as a parameter of the ambient PM 2.5 infiltration.To estimate the infiltration factors, a test method, using a blower-door depressurization technique, with an identical indoor-outdoor pressure difference of 10 Pa was proposed.The 10 Pa pressure difference was determined to reflect the normal pressure differences applicable to a building envelope (Kalamees et al., 2010).Control of the indooroutdoor pressure difference facilitates a direct comparison of the estimated infiltration factors among the test housing units.The building airtightness was also measured to obtain leakage information on the test units and to determine the correlation between their airtightness and infiltration factors.The blower-door airtightness test was conducted in compliance with ISO 9972 standard, estimating the air changes per hour at specified pressure, typically 50 Pa (ACH 50 ), and the effective leakage area (ELA) which are widely used airtightness index.

Description of Apartment Housing Units
Eleven unoccupied apartment housing units, located in three cities in Korea, were chosen as test units, as shown in Table 1.The age of the apartment buildings varied from 1 to 37 years.The types of housing unit ranged from a studio apartment to one-to four-bedroom apartments, and the floor area of the housing units varied from 18 m 2 to 212 m 2 .In general, the north and/or south sides of the housing units were exposed to the outdoor air, and the junctions of the vertical walls and windows were assumed the main particle infiltration passages through the building envelope.
The housing units feature different types of windows and window frames, i.e., single pane and double-strength sheet windows.There was no mechanical ventilation system in the units, except a local ventilation system, such as a kitchen hood and bathroom exhaust fan.

Envelope Airtightness Testing
Envelope airtightness was measured by using the typical blower-door fan depressurization/pressurization technique (Retrotec 3101 Blower Door System, Retrotec, USA).This technique measures the air flow through the building envelope leaks at indoor-outdoor pressure differences, which can be expressed by a power-law equation of flow through an orifice, as shown in Eq. ( 1).
where, Q is the air flow (m 3 h -1 ), ΔP is the pressure difference (Pa), C is the flow coefficient (m 3 h -1 Pa -n ), n is the pressure exponent, and the subscript f relates to the fan-induced flow or pressure.The constants C and n are obtained by regression of the data derived from the blower-door testing.The aim of the testing is to characterize the airtightness of the test units by estimating the ACH 50 and the ELA (ISO 9972, 2015).The specified pressure ACH 50 facilitates a comparison of the normalized airflow of the test units with a different volume induced by pressure of 50 Pa with the use of a fan.The absolute size of the leakage in the units is quantified by ELA, defined by Eq. ( 2).
where ELA is the effective leakage area (m 2 ), ΔP r is the reference pressure, and ρ is the density of the air (kg m -3 ).The C and n values were determined from the blower-door testing results by fitting Eq. ( 1).The air flow through the building envelope leaks was measured relevant to five-point indoor-outdoor pressure differences (10, 20, 30, 40,

Test Procedure for Infiltration Factors of PM 2.5
Fig. 1 shows the schematic diagram of the test procedure.Maintaining an identical indoor-outdoor pressure difference during the experiments could be achieved by using the blower-door depressurization procedure.Before the main infiltration test procedure, a blower door was installed at the front entrance of each test unit and, simultaneously, two identical PM 2.5 concentration monitors (TSI AM 510, TSI, USA) were installed both indoors and outdoors.Both PM 2.5 monitors were calibrated and certified by the manufacturer right before the experiments for the quality assurance of the measurements.A recent research (Yun et al., 2015) demonstrated that the monitors require correction to measure the mass of particulates and suggested the correction factor of AM510 in Korean situation.In this study, to focus on the investigation of the infiltration factors by measuring ratios between indoor-outdoor particle concentrations, the correction of the monitors was not considered.The indoor PM 2.5 sampling location was a table located in the center of every housing unit.Typically, Korean apartment housing units are equipped with a packaged air conditioner, with an outdoor condenser unit located on the vertical wall outside each unit.The outdoor PM 2.5 monitor was placed on the outdoor condenser unit.Prior to the blower-door depressurization procedure, all the air pathways, including the kitchen hoods, bathroom exhausts, bathroom drains, and wall outlets were sealed.This sealing was done to ensure that only particle infiltration through the building envelope would be considered in the infiltration test.
The main infiltration test started after the preparations had been completed, with the blower door starting the depressurization (ΔP = 10 Pa) and the PM 2.5 monitoring devices starting the data logging.The assumption was that the infiltrated outdoor air would completely replace the stagnant room air after one time constant.The depressurization was therefore continued for more than one time constant in the test housing units.The time constant was determined by the following Eq.(3).
where, τ is the time constant (h), and ACH 10 is the air exchange rate at an indoor-outdoor pressure difference of 10 Pa (h -1 ).The ACH 10 value was obtained by the airtightness test.
After one time constant, the indoor PM 2.5 concentration profile was assumed to have reached steady-state conditions.The blower-door depressurization and the PM 2.5 monitors remained operating for an additional 20 minutes to record the steady-state indoor and outdoor concentrations.Finally, the recorded PM 2.5 concentration data were collected for post data analysis to calculate the PM 2.5 infiltration factors.
The infiltration factors were calculated based on the assumption of the PM 2.5 mass balance in the housing unit.Assuming there was no indoor PM 2.5 generation or resuspension, the mass balance equation can be expressed as follows, by Eq. ( 4).
     where, t is the time (h), C i (t) is the indoor PM 2.5 concentration at time t, P is the penetration coefficient (-), C o is the ambient PM 2.5 concentration, and k is the depositional rate (-).Several assumptions were made for the infiltration calculation; 1) the ambient PM 2.5 concentration is constant during the test period, 2) the indoor air in the housing units is well-mixed, and 3) the indoor PM 2.5 concentration reached equilibrium after one time constant.When the test time is greater than the time constant, it is implied that the indoor PM 2.5 concentration has reached a steady-state condition, then Therefore, Eq. ( 4) can be summarized as follows by Eq. ( 6).

 
10 , , 10 where, C i,ss is the steady-state indoor PM 2.5 concentration.
Finally, the infiltration factor can be obtained from Eq. ( 7).

RESULTS AND DISCUSSION
Ambient PM 2.5 Concentration Fig. 2 shows the average and the standard deviation of the ambient PM 2.5 concentrations during the entire test period in each test housing unit.As each test was conducted separately, at different times and locations, there were significant differences in the PM 2.5 concentrations between the test units.The maximum ambient PM 2.5 concentration was 180 µg m -3 in housing unit H, whereas the minimum concentration was 38 µg m -3 in housing unit G.According to the AirKorea website, which provides real-time PM 2.5 concentration data Fig. 2. Average and standard deviation of ambient PM 2.5 concentration on the day of testing for each unit.online, the PM 2.5 concentrations are classified into four categories; good (0-15 µg m -3 ), average (16-50 µg m -3 ), bad (51-100 µg m -3 ), and very bad (more than 100 µg m -3 ).
According to this classification, the measurements indicated that five test housing units had very bad PM 2.5 concentrations (Units A, B, E, H, and K), three had bad conditions (Units C, F, and J), three had average conditions (Units D, G, and I), with no good conditions being found in any of the housing units.
In all the units, the air quality standard of 35 µg m -3 for 24-hour PM 2.5 , established by the United States Environmental Protection Agency (USEPA, 2012), was exceeded.Furthermore, the WHO standard of 10 µg m -3 for 24-hour PM 2.5 was exceeded by a considerable margin (WHO, 2006).Considering that the duration of each test was about five hours, the average ambient PM 2.5 concentrations presented here are approximately five-hour averaged concentration data.Although no precise direct comparison with the above standards could be made, it could be stated that the levels of the ambient PM 2.5 concentrations at the time of measurement were relatively high.Compared with the other PM 2.5 data obtained in Korea, the PM 2.5 concentration measured in Gosan (Stone et al., 2011) during the spring of 2007 showed similar levels, ranging from 28 to 122 µg m -3 , with the 24-h average PM 2.5 concentration being 36 ± 26 µg m -3 .As could be expected, periods with no dust events showed lower levels, while periods with dust events presented higher levels, exceeding 100 µg m -3 (Stone et al., 2011).

Airtightness Test Result
Table 2 shows the results of the airtightness tests on the housing units, including the ACH 50 and ELA.The estimated airtightness test results represent the information on leakage relevant to the exterior vertical walls and the windows of the housing units.The ACH 50 values of the housing units ranged from 1.44 h -1 to 12.39 h -1 , which far exceeded the Passivhaus standard of 0.6 ACH 50 (PHI, 2007).Various units (Units A, B, E, G, and K) met the requirement of 5.0 ACH 50 set by Energy Star 3.0 (USEPA, 2013), while the rest of the units (Units C, D, F, H, I, and J) showed high levels, above 5.0 ACH 50 .Except for Unit I, the buildings in the group of leaky housing units (Units C, D, F, H, and J) were older than 14 years.Generally, the results showed that the older housing units had poor airtightness, resulting in high ACH 50 values.
The ELA of the housing units ranged from 8.44 cm 2 to 435.35 cm 2 , generally showing a proportional relationship to the floor area.An exceptionally small ELA value of 8.44 cm 2 was indicated for Unit G.This is ascribed to Unit G being recently built, i.e., within a year prior to the tests, having an airtight window system, and being a studio apartment, with the smallest floor area of the all the units.
Usually, the flow exponent depends on the flow regime, namely, 0.5 for fully turbulent flow and 1.0 for laminar flow.In practice in many instances, flow exponent values of approximately 0.6 and 0.7 are found.Generally, in this study, the estimated flow exponent values were found to be within reasonable ranges.

Infiltration Factors of the Test Housing Units
Fig. 3 shows an example of the ambient and indoor PM 2.5 concentration profile during the infiltration tests to demonstrate the data used to calculate the infiltration factors.The indoor and outdoor concentrations at a steadystate condition were averaged, then listed in Table 3 for all housing units.The infiltration factors were calculated by using the ratios of the steady-state ambient and indoor PM 2.5 concentrations.
Fig. 4 shows the estimated PM 2.5 infiltration factors relevant to 10 Pa of pressure difference.The average infiltration factor in all the test housing units was 0.65 ± 0.13 (average ± standard deviation), with a minimum of 0.38 to a maximum of 0.88.Table 3 lists the infiltration factors, as well as the outdoor air exchange rates and the average steady-state ambient PM 2.5 concentrations of the housing units.Small standard deviation values were indicated for the infiltration factors and average ambient PM 2.5 concentrations, indicating that the infiltration tests were conducted in sufficiently steady-state conditions.
In their research, Chen and Zhao (2011) reviewed 21 previous experimental studies on the relationship between indoor and outdoor particles, suggesting that the average infiltration factors for PM 2.5 measured by the different researchers varied considerably between 0.3 and 0.82.The reported experimental data were obtained from homes in the USA or Europe.Compared with the values from previous research, the average infiltration factor of 0.65 measured in   It is assumed that the 10 Pa indoor-outdoor difference adopted in this study could have affected the level of the infiltration factor.Fig. 5 shows the relationship between the infiltration factors and the airtightness test results.The ACH 50 , effective leakage area and the specific effective leakage area were correlated with the infiltration factors.Power trend lines were drawn to illustrate the relationship to the measured data.Generally, although the R-square values were less than 0.5, the infiltration factors tended to increase with the ACH 50 values.
The results from data on the relationship between the building airtightness and the infiltration factors indicate that a leaky housing unit, with high ACH 50 or high specific ELA, would be influenced more significantly by polluted ambient PM 2.5 .At the present time, air-tight building construction practices are followed in Korea in the effort to reduce energy consumption in the building sector.The aim of this endeavor is to comply with the standards prescribed by international treaties, such as the Kyoto Protocol and the Paris Agreement (2015) for greenhouse gas reduction.This trend implies that the new build housing units are likely less influenced by ambient PM 2.5 than are aged buildings with relatively leaking envelopes.

DISCUSSION
This study demonstrated that the infiltration test procedure employed was useful to assess the infiltration factors in  conditions of controlled indoor-outdoor pressure difference in real housing units.The vertical distribution of pressure on the exterior wall of a high-rise multi-family house, similar to those often constructed in urban areas in Asian countries, would vary, depending on the wind pressure and stack effect.The suggested particle infiltration test method enabled direct comparison of the infiltration factors in all the test housing units, with the same pressure condition.
It should be noted that the infiltration factors were estimated in a pressure difference condition of 10 Pa between the indoor and the outdoor.Generally, it is known that the normal pressure difference applicable to a building envelope is 0 to 5 Pa.The slightly higher pressure difference in this study was intended to reduce the uncertainty of the infiltration tests.In numerous Asian cities, where high-rise residential buildings are abundant, the indoor-outdoor pressure difference would be larger than 5 Pa because of the wind and stack effect.Especially in winter, when the ambient PM 2.5 shows high levels because of industrial/domestic heating and dust events, the stack effect, combined with the outdoor wind velocity, would increase the pressure difference.Therefore, the 10 Pa difference test condition used in this study could be meaningful to indicate the impact of ambient PM 2.5 in indoor environments in such conditions.In real buildings, the direction of air flow and PM 2.5 transport would vary depending on the pressure conditions applied to building envelope resulting in different I/O ratios and infiltration factors.Further investigation of the influence of various pressure differences is being planned in future research.
Additionally, apartment housing units with no mechanical ventilation systems were investigated in this study; however, mechanically ventilated housing units would have different infiltration characteristics.Since 2009 in Korea, new build or remodeled apartment complexes with more than 100 units have been adopting mechanical ventilation systems in an effort to meet the level of 0.5 ACH specified for indoor air quality (IAQ).The effect of ambient PM 2.5 on the indoor environment in mechanically ventilated housing units should be examined in further study.

CONCLUSIONS
The current study investigated the infiltration of ambient PM 2.5 through the building envelop in apartment housing units in Korea.The on-site infiltration test method, by using a blower-door depressurization procedure, was used in order to maintain identical indoor-outdoor pressure differences among all the tests.On-site experiments in 11 apartment housing units were conducted to estimate the PM 2.5 infiltration factors.The results showed that the average infiltration factor of all the test housing units was 0.65 ± 0.13 (average ± standard deviation), with a minimum of 0.38 to a maximum of 0.88.In addition, the results from the relationship between the building airtightness data and the infiltration factors indicated that a leaky housing unit, with high ACH50 or high specific ELA, would be significantly more influenced by ambient PM 2.5 .This study demonstrated that the infiltration test procedure suggested in this study was useful to assess the infiltration factors, in controlled indoor-outdoor pressure differences, in real housing units.

Fig. 3 .
Fig. 3.An example of ambient and indoor PM 2.5 concentration profile during the infiltration test (Unit G).

Table 1 .
Description of test housing units.

Table 2 .
Airtightness test results of housing units.

Table 3 .
Results of infiltration tests in test housing units.