Installation of platform screen doors and their impact on indoor air quality: Seoul subway trains

Abstract

In this study, variations of particulate matter (PM) concentrations in subway trains following installation of platform screen doors (PSDs) in the Seoul subway system were investigated. PM samples were collected in the trains on subway lines 1–8 before and after installation of PSDs. It was found that the mean PM₁₀ concentration in the trains after PSDs installation increased significantly by 29.9% compared to that before installation. In particular, the increase of PM₁₀ in line 6 was the highest at 103%. When the relationship between PM₁₀ and PM_2.5 was compared, coefficients of determination (r²) before and after PSDs installations were 0.696 and 0.169, respectively. This suggests that air mixing between the platform and the tunnel after PSDs installation was extremely restricted. In addition, the indoor/outdoor PM₁₀ ratio following PSDs installation increased from 1.32 to 2.97 relative to the period with no installed PSDs. Furthermore, this study revealed that PM levels in subway trains increased significantly after all underground PSDs were put in use. Several potential factors were examined that could result in this PM increase, such as train ventilation systems, operational conditions, passenger volume, subway depth, and the length of underground segments.

Implications

PM₁₀ concentrations inside the subway trains increased after PSDs installation. This indicates that air quality in trains was very seriously impacted by PSDs. PM₁₀ levels were also influenced by the tunnel depth and length of the underground segments. To prevent the adverse effect on human health by PM₁₀ emitted from the tunnel, an applicable ventilation system to reduce PM₁₀ is required inside trains and tunnels.

Introduction

Use of public transportation has increased because of high oil prices and traffic congestion. The subway is the most representative means of public transportation and makes up 36.2% of all means of conveyance in Seoul, Korea (Seoul Metropolitan Government, 2013a). Underground subway stations have different characteristics relative to outdoor conditions, such as confined spaces and the presence of pollution emission sources. Thus, quite a few studies have been conducted to understand the air quality characteristics in subway systems (Aarnio et al., 2005; Adams et al., 2001; Braniš, 2006; Johansson and Johansson, 2003; Karlsson et al., 2006; Kim and Ro, 2010; Kim et al., 2012; Raut et al., 2009; Sohn et al., 2008; Son et al., 2011).

Platform screen doors (PSDs) have been installed between the platform and the tunnel in Seoul subway stations to maintain the safety of passengers and to distinctly improve indoor air quality (Kim et al., 2012; Sohn et al., 2008; Jung et al., 2012). Installing PSDs can prevent injury and loss of human life caused by subway trains passing through the stations. With regard to indoor air quality, some studies have reported that particulate matter (PM) concentrations on the platform decreased after installing PSDs by preventing particles suspended by train-induced wind in the tunnels from flowing into the platforms (Kim et al., 2012; Jung et al., 2010). Kim et al. (2012) reported that the PM₁₀ concentration on the platform of an underground subway station decreased by 16% following PSDs installation. Also, iron (Fe)-containing particles originating from tunnels noticeably decreased on a platform and in the waiting room after installing PSDs (Jung et al., 2010). These studies consistently showed that platform air quality has been improved after PSDs installation, because air flow between the tunnel and the platform was blocked.

However, the PSDs may cause air quality inside the subway trains to deteriorate, possibly due to an increased residence time of particles in the tunnels. The air quality in the trains is particularly crucial because passengers are likely to spend more time in the trains than on the platforms. Epidemiological studies have reported adverse respiratory and cardiovascular health effects associated with PM mass concentrations (Dominici et al., 2006; Gent et al., 2003). Particularly, the composition of PM emitted from tunnels includes mainly iron compounds such as magnetite (Fe₃O₄), hematite (Fe₂O₃), and Fe metal (Jung et al., 2012; Mugica-Álvarez et al., 2012; Salma et al., 2007). With regard to the toxicity of those specific compounds, Karlsson et al. (2005) reported that the subway particles were approximately eight times more genotoxic and four times more likely to cause oxidative stress in lung cells.

Despite the obvious importance of indoor air quality in the subway trains, previous studies focused on the effects of PSDs installation on air quality only on the subway platforms (Sohn et al., 2008; Cheng et al., 2008; Park and Ha, 2008; Seaton et al., 2005). To the best of our knowledge, no studies have investigated the effect of PSDs installation on the air quality of the interior of trains.

To investigate the potential air quality shift inside subway trains, we measured the PM concentrations in the trains before and after installation of PSDs in this study, which allows us to examine the relationship between PM_2.5 and PM₁₀ concentrations inside the trains for both periods. Finally, variations of relative PM concentrations with respect to outdoor PM concentrations, before and after installing PSDs, were investigated.

Experimental Method

Sampling sites

The Seoul subway system contains nine lines and 358 stations. According to statistics provided by the Seoul Metropolitan Government (2013b), approximately 6.7 million commuters use the Seoul subway system daily. In this study, we collected PM₁₀ and PM_2.5 samples in passenger trains on Seoul subway lines 1–8. To investigate the effect of PSDs on indoor air quality, PM concentrations were measured in passenger trains before (from March 17, 2008, to May 7, 2008) and after (from April 5, 2010, to May 6, 2010) PSDs installation. Each passenger train of the Seoul subway system consisted of 8–10 cabins. All PM samplers were set up in the same places, located in the fourth cabin from the head of each train, for all subway lines. In general, for maintenance purpose, trains are occasionally moved to a tram depot during the daily running schedule. However, upon our request, the trains used in this study did not stop their operations, making it possible to obtain high-quality data. In addition, regarding underground trains’ ventilation systems, it was revealed that the subway trains only had internal ventilation systems. This is because PM concentrations in the tunnels were much higher than those on the platforms and outdoors and thus it was more appropriate at that time to establish the internal ventilation systems with air circulating only inside the trains.

Also, we obtained hourly outdoor PM₁₀ concentrations, measured by the Korean Ministry of Environment (Real-Time Air Quality Data), to directly compare PM levels in indoor and outdoor conditions during the same periods. Outdoor PM₁₀ levels were estimated by calculating averages of all available PM₁₀ measurements from 25 sampling sites (one site in each district) in Seoul (Figure 1). It was considered appropriate to use averages of hourly measurements to compare subway train measurements since each subway train operated across many different districts.

Figure 1. Outdoor PM10 sampling locations and subway service routes in this study.

Measurement method

PM₁₀ and PM_2.5 concentrations were measured using four mini-volume air samplers (Pas-201, Airmetrics, USA) with cellulose filters (Advantec Toyo 5B [110 mm], Toyo Roshi Kaisha, Japan). Two samplers were used for PM₁₀ and the other two for PM_2.5. Each sample was collected for 11 hr (from 7 a.m. to 8 p.m.) at a flow rate of 5 L/min. All samples were collected at 1.5 m above the floor of each cabin. All sampling filters were maintained under controlled temperature (20 ± 2°C) and relative humidity (50 ± 10%) prior to gravimetric analysis. A highly sensitive electronic microbalance (Mettler Toledo UMX2, Switzerland) with 0.1 μg readability was used to measure PM mass. Detailed information regarding the air-sampler can be found in a previous publication (Kim et al., 2012). In addition to the gravimetric PM measurements, we simultaneously used the beta-ray attenuation method (FH 62 c-14, Thermo Scientific, USA) during the same period to compare hourly variations in PM concentrations. The FH 62 c-14 is an automatic air monitor based on beta-ray attenuation, certified by the Korean Ministry of Environment as an effective approach for environmental monitoring. Samples were obtained every half hour. Sampling was conducted at a flow rate of 16.7 L/min and hourly mean PM₁₀ levels were calculated. The comparison between filter-based sampling and beta-ray attenuation method was conducted before actual sampling in subway trains. The correlation coefficient (r) between the two methods was 0.90. In addition, the FH 62 c-14 was calibrated with the zero and span plates during field monitoring. Outdoor PM₁₀ data were acquired from all 25 above-ground monitoring sites as mentioned earlier. Hourly outdoor PM concentrations were determined for each sampling site using the beta-ray attenuation method.

Data analysis

Previous studies have reported that sources of PM that affect the indoor air quality inside subways could be associated with complicated conditions, such as train frequency and efficiencies of ventilation systems (Sohn et al., 2008; Son et al., 2013). Therefore, high variability is expected in the data from different locations. R software (version 2.15.3) was used with the collected data to determine the required statistical measures. Simple and multivariate linear regression analyses were conducted to determine whether the means of two PM concentration groups (before and after PSDs installations) were statistically different.

Limitations and uncertainties

Potential limitations and uncertainties in this study included the relatively small sample size and the consequent potential lack of representativeness of the samples, various operating conditions, deterioration of tunnel ventilation systems, diverse kinds of trains and their running conditions, variations of ambient PM₁₀ concentrations, and heterogeneity between measurement instruments. To address these issues, we used diversified statistical methods.

Results and Discussion

Variations in PM concentrations before and after PSDs installation

Table 1 shows the PM₁₀ and PM_2.5 concentrations measured inside the subway trains as well as indoor/outdoor PM₁₀ and indoor PM_2.5/indoor PM₁₀ ratios before and after installing PSDs. Mean indoor PM₁₀ concentrations across the subway lines before and after PSDs installation were 80.5 μg/m³ (range, 43.7–146.4 μg/m³) and 104.6 μg/m³ (range, 53.6–173.8 μg/m³), respectively, indicating that the PM₁₀ level inside the trains rose by a statistically significant (P = 0.374) 29.9% relative to the earlier period. The increase in indoor PM₁₀ levels after PSDs installation was further evidenced by the increased ratio of indoor PM₁₀ to outdoor PM₁₀. Since outdoor PM₁₀ concentrations typically affect indoor PM₁₀ levels, their direct comparison with indoor PM₁₀ concentrations before and after PSDs installation could be confounded by the increase in outdoor PM₁₀ levels between the study periods in 2008 and 2010. To avoid this potential confounding we compared indoor/outdoor PM₁₀ ratios before and after the PSDs installation, across all eight subway lines. This analysis showed that the ratio in 2010 (2.97) was significantly higher than that in 2008 (1.32) (P = 0.0004). This pooled model analysis seems more appropriate than subway line-specific comparisons due to our small sample size in 2010.

Table 1. Particulate matter concentrations inside passenger trains before and after installing platform screen door

						Before PSDs (from March 17, 2008 to May 7, 2008)											After PSDs (from April 5, 2010 to May 6, 2010)
			Indoor PM2.5 (μg/m^3) ^a^)			Indoor PM10 (μg/m^3) ^a^)			Outdoor PM10 (μg/m^3) ^b^)				Indoor PM2.5 (μg/m^3) ^a^)			Indoor PM10 (μg/m^3) ^a^)			Outdoor PM10 (μg/m^3) ^b^)
Line Number	TD ^c^) (m)	TL ^d^) (km)	SN ^e^) (n)	Mean (SD)	Min– max	SN ^e^) (n)	Mean (SD)	Min– max	SN ^e^) (n)	Mean (SD)	Min– max	Indoor PM2.5/ indoor PM10	Indoor/ outdoor PM10	SN ^e^) (n)	Mean (SD)	Min– max	SN ^e^) (n)	Mean (SD)	Min– max	SN ^e^) (n)	Mean (SD)	Min– max	Indoor PM2.5/ indoor PM10	Indoor/ outdoor PM10
1	10.9	7.8	12	57.2 (20.4)	31.3–91.5	12	96.4 (36.5)	60.5–146.4	66	74.6 (36.4)	35.3–127.9	0.61	1.36	4	44.6 (4.2)	41.6–47.6	4	126.0 (67.7)	78.1–173.8	22	47.0 (2.0)	45.6–48.5	0.40	2.71
2	14.4	54.2	14	54.0 (9.0)	38.7–66.0	14	88.1 (12.8)	75.3–108.9	77	52.4 (16.6)	33.2–84.2	0.62	1.80	4	44.6 (7.5)	39.3–49.9	4	63.8 (14.4)	53.6–74.0	22	56.6 (11.2)	48.7–64.5	0.70	1.12
3	17.7	38.2	12	64.5 (16.1)	44.4–88.5	12	95.2 (27.3)	67.9–137.1	66	60.2 (24.6)	31.2–102.5	0.68	1.71	4	49.2 (1.9)	47.8–50.5	4	106.0 (24.8)	88.4–123.5	22	36.8 (6.7)	32.0–41.5	0.48	2.87
4	16.8	38.2	12	51.0 (19.2)	33.2–75.7	12	67.9 (23.3)	43.7–95.5	66	83.8 (30.9)	45.3–123.9	0.74	0.83	4	54.7 (19.9)	40.6–68.7	4	110.1 (41.8)	80.5–139.6	22	58.5 (26.5)	39.7–77.2	0.50	1.92
5	23.5	52.3	12	52.0 (12.6)	34.7–64.9	12	91.6 (12.2)	74.7–107.9	66	121.4 (159.4)	33.8–445.4	0.56	1.45	4	49.3 (9.8)	42.3–56.2	4	81.4 (7.2)	76.3–86.5	22	60.2 (0.9)	59.6–60.8	0.60	1.35
6	22.5	35.1	10	42.4 (13.9)	27.0–60.2	10	73.3 (8.7)	58.0–79.0	55	70.4 (9.3)	57.9–82.1	0.57	1.04	4	48.4 (12.2)	39.7–57.0	4	148.8 (8.5)	142.8–154.8	22	17.1 (7.7)	11.6–22.5	0.32	9.58
7	22.6	57.1	10	33.4 (7.2)	27.7–45.4	10	63.6 (8.4)	50.2–70.9	55	50.2 (17.4)	25.4–69.8	0.53	1.39	4	45.8 (13.9)	35.9–55.6	4	108.4 (10.0)	101.3–115.5	22	45.3 (8.9)	39.0–51.6	0.42	2.42
8	20.0	17.7	10	42.5 (8.3)	32.2–54.3	10	68.3 (10.3)	50.4–76.4	55	72.2 (18.6)	53.8–96.2	0.62	1.00	4	37.1 (1.5)	36.0–38.1	4	92.1 (8.0)	86.4–97.7	22	54.6 (14.9)	44.1–65.2	0.40	1.77
Total			92	49.6 (9.8)	27.0–91.5	92	80.5 (13.6)	43.7–146.4	506	73.2 (22.6)	25.4–445.5	0.62	1.32	32	46.7 (5.1)	35.9–68.7	32	104.6 (26.2)	53.6–173.8	176	47.0 (14.4)	11.6–77.2	0.48	2.97

Notes: ^aGravimetric method: Each sample was collected for 11 hr.

^bβ-ray attenuation method: Each sample was measured every 1 hr.

^cTunnel depth.

^dTotal length.

^eSample number.

The rate of increase of the PM₁₀ concentration in the trains on line 6 (103%) was the highest among all subway lines in Seoul. The reason may be that all subway rails on line 6 are located underground and the tunnel is as deep as 22.5 m, thus potentially having more limited fresh air and ventilation rate. We found a similar increase pattern in line 7 (70.4%), with an average tunnel depth of 22.6 m. Additionally, we found that PM₁₀ concentrations after PSDs installation increased on all subway lines except for lines 2 and 5. The PM₁₀ concentration on line 2 (–27.6%) decreased, counter to the trend of the other lines. Line 2 is composed of 13 above-ground stations out of a total of 50 train stations, and trains passing through these stations were exposed to much more fresh ambient air. For line 5 the PM₁₀ concentration after PSDs installation also decreased compared to that before PSDs installation (–11.1%), and it seems that this result was due to a remarkably high ambient PM₁₀ concentration (121.4 ㎍/m³) for this line during the test period, which is 1.4–2.4 times higher than ambient PM₁₀ concentrations (50.2–83.8 μ/m³) for the other lines in 2008 (each tested during a different time period).

To evaluate the influence of tunnel depth, we categorized the entire set of subway lines into two groups, lines 1–4 and 5–8. This is because lines 1–4 were constructed with shallower depths (10.9–17.7m) than lines 5–8 (20.0–23.5 m) and it seems more appropriate to compare small homogeneous groups, considering the limited number of samples included in our analysis. The results suggested that deeper tunnels result in a greater increase in PM following PSDs installation, since the mean PM₁₀ increase (49.3%) on lines 5–8 was larger than that (19.2%) on lines 1–4. To examine the data further, we analyzed the relationship between indoor/outdoor PM₁₀ ratios and tunnel depth. We found that this ratio for lines 1–4 (1.61) was smaller than that for lines 5–8 (1.93), but the difference between the groups was not statistically significant (P = 0.4478). This result suggests that air in the tunnel can be more easily vented as a tunnel depth gets shallower, with the assumption that the tunnel is ventilated under the same fan operation conditions. However, the insignificant difference may indicate that other environmental factors, such as the pressure difference, wind speed, height of the ventilation opening, and efficiency of the ventilation system (Son et al., 2013; Song et al., 2008), also need to be considered to better understand the overall ventilation mechanism.

Previous studies reported that train operation was a major source of particulate matter in tunnels (Birenzvige et al., 2003; Kim et al., 2012; Kang et al., 2008; Son et al., 2013). In terms of the trains’ operational conditions, we compared the frequencies of operated trains between 2008 and 2010. According to the statistics of the Seoul Metropolitan Government (http://stat.seoul.go.kr), the frequencies of operated trains in 2008 and 2010 were 3,332 and 3,225 per day (on weekdays), respectively. Though the frequency of train runs in 2010 decreased compared with that in 2008, PM concentration inside the trains increased from 2008 to 2010.

To figure out the effect of passengers on PM concentrations, the number of passengers was taken into account in this work. The annual numbers of passengers in 2008 and 2010 were approximately 2,293,848,000 and 2,349,374,000 persons, respectively. This could imply that the average numbers of passengers per train (which consisted of 8–10 cabins) for the years 2008 and 2010 were 1,886 and 1,995 persons, respectively, which indicated a 5.8% increase during the period. Therefore, it is presumed that PM generated due to the increased number of passengers is not likely to have significantly deteriorated indoor air quality inside subway passenger cabins.

As shown in Table 1, mean indoor PM_2.5 concentrations on all eight subway lines before and after PSDs installations were 49.6 μg/m³ (range, 33.4–64.5 μg/m³) and 46.7 μg/m³ (range, 37.1–54.7 μg/m³), respectively, indicating that the PM_2.5 concentration in all trains after PSDs installations was decreased by 5.8%, on average, which was not statistically significant (P = 0.4679). The direction of PM_2.5 concentration variation was opposite to that of the PM₁₀. The indoor PM_2.5 levels decreased after PSDs installation in all subway lines, except for lines 4, 6, and 7. The PM_2.5 levels on lines 4, 6, and 7 increased by approximately 7.3%, 14.2%, and 37.1%, respectively, but an average the PM_2.5 level on the other lines decreased by approximately 14.5%. These results indicate that PM concentrations in the tunnels and trains might be critically affected by the tunnel depth of the subway system. On the other hand, it should be noted that there were no PM removal devices in the tunnel.

Table 1 shows that the mean PM_2.5/PM₁₀ ratio in trains decreased from 0.62 to 0.48 after PSDs installation on all subway lines (i.e., the decrease was statistically significant; P = 0.0001). This indicates that PM₁₀ levels in trains following installation of PSDs increased relative to PM_2.5 levels, which decreased or stabilized. This may be because separating the tunnel from the platform by installing PSDs prevents ventilation of relatively larger particles (>2.5 μm) from the inside of the train to the outside.

PM_2.5/PM₁₀ ratios in trains before PSDs installation were 0.53–0.74, which was very similar to results reported by Kim et al. (2012), who obtained an average ratio of 0.56 (range, 0.44–0.80). PM_2.5/PM₁₀ ratios from three platforms with PSDs were 0.51–0.60 (Sohn et al., 2008).

Previous studies that targeted the inside of trains and the platform before PSDs installation revealed that the PM_2.5 /PM₁₀ ratios ranged from 0.50 to 0.85 (Park and Ha, 2008; Chan et al., 2002), which is consistent with results of the present study. However, in this study, the average PM_2.5/PM₁₀ ratio after PSDs installations was 0.48 (range, 0.32–0.70), while the PM ratios in lines 1, 3, 6, 7, and 8 were less than 0.50, indicating that the relatively larger particle concentrations inside the train increased after installing PSDs compared with those before PSDs installation.

To examine variations in the relationship between PM₁₀ and PM_2.5 concentrations in subway trains following PSDs installation, the correlation between PM₁₀ and PM_2.5 in lines 1–8, before and after installing PSDs, was compared. The coefficient of determination (r²) before installing PSDs was relatively high (0.696), but that after PSDs installation was relatively low (0.169). While the ratio of PM_2.5 to PM₁₀ before PSDs installation on the overall subway lines was reasonably correlated, it seems that the PM_2.5/PM₁₀ ratio was influenced by the depth of the station, the operating conditions of the train fan, the number of passengers, and the proportion of above-ground stations on subway lines after PSDs installation. A recent subway study (Kim et al., 2012) reported that there was a greater decrease in PM₁₀ on the platform compared to PM_2.5, possibly due to the PSDs installation. This indicates that an increase of coarse particles (PM₁₀–PM_2.5) might have occurred in the tunnels. Furthermore, it can be concluded that the correlation coefficients after PSDs installation would be lower than before since air mixing between the platform and the tunnel became extremely restricted. A previous study (Kim et al., 2009) also observed the decrease of coefficients of correlation on the platform after PSDs installation.

Evaluation of PM concentration between rush hour and non-rush hour

The beta-ray attenuation method was used to measure PM₁₀ concentrations inside the trains on all eight subway lines during rush hours (8–10 a.m. and 6–8 p.m.) and non-rush hours (10 a.m.–6 p.m.), both before and after PSDs installation (Table 2). We found that the PM₁₀ concentration after PSDs installation increased from 88.6 μg/m³ to 105.8 μg/m³ (P = 0.0005) during non-rush hours, while it decreased from 108.6 μg/m³ to 101.3 μg/m³ (P = 0.2833) during rush hours. Since the yearly and diurnal variability in outdoor PM₁₀ concentrations might affect indoor PM₁₀ concentrations, we compared the indoor/outdoor PM₁₀ ratios (both hourly PM₁₀ measurements) before and after PSDs installation separately for rush and non-rush hours. During the rush hours, the ratio increased from 1.76 (2008) to 3.33 (2010), and the increase was statistically significant (P < 0.0001). During the non-rush hours, the ratio also significantly increased from 1.57 (2008) to 3.41 (2010) (P < 0.0001).

Table 2. Variations of PM₁₀ concentrations in passenger trains during rush and non-rush hours following PSDs installation

		Sample number (n)	Mean (SD) (μg/m^3)	Min–max (μg/m^3)	Rush hour/ non-rush hour	Indoor/ outdoor PM10
Before PSDs (2008)	Rush hour	230	108.6 (50.6)	18.6–299.1	1.23	1.76
	Non-rush hour	211	88.6 (39.1)	29.3–264.9		1.57
After PSDs (2010)	Rush hour	65	101.3 (38.6)	21.3–204.8	0.96	3.33
	Non-rush hour	97	105.8 (41.3)	42.3–236.0		3.41

Notes: All values were obtained from the β-ray attenuation method. The PM₁₀ concentrations using this instrument were measured every half hour.

The rush/non-rush hour ratio of PM₁₀ concentrations in train decreased from 1.23 to 0.96. This ratio indicates that the variation of PM₁₀ concentration due to train frequencies decreased after PSDs installation. This suggests that the indoor PM₁₀ concentration is likely affected by mechanical factors such as fan operating conditions in tunnels and trains, rather than the number of trains.

Effects of outdoor PM concentrations

PM concentrations inside both subway stations and trains are associated with those of ambient PM (Aarnio et al., 2005; Braniš, 2006; Kim et al., Jung et al., 2010). PM levels in ambient air and inside trains were measured to evaluate how the relation between these two was affected before and after installing PSDs (Table 1).

These measurements revealed that while the ambient PM₁₀ concentration decreased from 73.2 μg/m³ to 47.0 μg/m³ before and after installing PSDs, the PM₁₀ concentration inside the trains increased. Generally, indoor PM₁₀ concentrations inside trains before installing PSDs were similar to or lower than those of ambient PM₁₀ (lines 4–8). However, the results showed that PM₁₀ concentrations inside trains on all eight lines were higher than ambient PM₁₀ concentrations after installing PSDs. Particularly, the PM₁₀ concentration of the line 6 after installing PSDs was 148.8 μg/m³ although the ambient PM₁₀ concentration was only 17.1 μg/m³. This may have occurred due to polluted air in the tunnel that blew into the trains. Indoor/outdoor PM₁₀ ratios were compared before and after installing PSDs, and the mean ratios were 1.32 (range 0.83–1.71) and 2.97 (range 1.12–9.58), respectively. This indicated that increased indoor PM concentrations inside the trains were mostly due to PSDs. Previous studies also reported that outdoor PM concentrations affected indoor PM concentrations in subway trains and platforms (Kim et al., 2012; Cheng et al., 2008). Therefore, relatively higher indoor/outdoor PM₁₀ ratios after the PSDs installation indicate that the effect of PM₁₀ generated in the tunnels on that in the subway trains became larger.

Out of all eight lines, the PM₁₀ concentration after PSDs installation of line 2, which is comprised of 26% above-ground stations, was most similar to that of outdoor air (ratio of 1.12). In contrast, lines 3 and 6, which comprise the lowest proportion of aboveground stations, had the highest ratios, 2.87 and 9.58, respectively.

PSDs were installed to assure passenger security and to improve indoor air quality inside subway stations, and have worked well for these roles. However, PSDs installations result in increasing PM₁₀ concentrations in the tunnels and inside the trains. To solve these problems, high-efficiency filtering systems to treat polluted inflowing air and suitable train ventilation systems are needed. Additionally, forced ventilation to control accumulated PM in the tunnel is required. A study on PM_2.5 management and control methods is needed to effectively reduce PM concentrations inside trains.

Suggestions on PM control in tunnels

Various control techniques such as ventilation, filtration, and tunnel-cleaning trains have been used to remove the PM in underground subway tunnels. However, these methods have limitations in terms of high operating cost and PM removal efficiency. Therefore, we suggest more reasonable control approaches for PM reduction for all Seoul subway tunnels. These methods could also be used to lower PM levels in subway systems elsewhere in the world.

Most of the ventilation holes for PM removal in tunnels are equipped with forced ventilation systems. In the case of the Seoul subway systems, they have been set up and operated in tunnels of lines 2–8. However, these systems have several problems (e.g., public grievance due to noise, deterioration over time, unsuitable design for ventilation, and high energy costs). To solve these issues, present forced ventilation systems should be maintained and operated well using a more suitable methodology. In our previous study, although airflow rates used were different, depending on various tunnel environments, we confirmed that the PM₁₀ concentration in a tunnel was below 150 μg/m³ (which is the Korean standard value for indoor air quality) when an adequate air ventilation rate was supplied in a tunnel (Son et al., 2013). Additionally, according to the results reported by some researchers, PM₁₀ levels in tunnels fluctuated extensively, depending on the train frequency (Birenzvige et al., 2003; Johansson and Johansson, 2003; Son et al., 2013). These results suggest that the generation of PM₁₀ in tunnels is higher during certain times, such as rush hour. On the basis of this phenomenon, the ventilation operating conditions should be adjusted either by the train time schedule or by using artificial intelligence methods (Le et al., 2013; Liu et al., 2013; Kim et al., 2012). The basic concept of the time-schedule methods is to either modulate operation of fans depending on specific run events (e.g., rush hour and non-rush hour), or to adjust the fan frequency using an artificial intelligence method, depending on measured and/or predicted PM₁₀ levels in tunnels. Using these methods could significantly reduce both the energy cost and the public grievance.

On the other hand, a part of the design of ventilation holes is to use natural ventilation, and all the ventilation holes of line 1 make use of this design. The natural ventilation system for line 1 has very narrow ducts, with no fans. Using this system, the air circulation depends entirely on the train-induced wind, and thus air exchange rates by the natural ventilation system are considerably lower than those by the forced ventilation system. Recently, to resolve this problem, an approach using small jet fans was introduced to reduce PM levels in tunnels with natural ventilation systems, and these possibilities were validated through actual pilot tests (Lee et al., 2011; Son et al., 2011).

In terms of the structure of a ventilation hole, the height of inflow ventilation hole should be elevated to supply cleaner air. Moreover, most of the subway ventilation holes in Seoul are currently located near main roads, causing PM₁₀ concentrations introduced into these holes to be higher than those in air supplied at other places such as public parks and residential districts. Therefore, if there is a realistic possibility for a relocation of ventilation holes, the inflow ventilation holes should be relocated to cleaner areas.

Results of previous studies reported that the majority of particulate matter generated in tunnels consisted of Fe-containing particles with magnetic properties (Eom et al., 2013; Jung et al., 2012; 2010). Jung et al. (2012) reported that the fraction of magnetic particles in the floor dust of subway systems accounted for 98–100% of particles <25 μm in size (Jung et al., 2012). They also reported that iron particles (2.5–10 μm in size) made up 77.3–86.9% of tunnel dust samples (Jung et al., 2012). Moreover, the nonmagnetic hematite comprised significant portions (30%) of tunnel samples. Nevertheless, most of the particles were attracted to magnets (Eom et al., 2013). On the basis of these results, the magnetic filter has been applied as a method to reduce PM levels in a tunnel, and its efficiency has been recently evaluated in an actual field by our research group (Son et al., 2014).

Conclusions

Air quality in underground subway stations and passenger safety are remarkably improved by installing PSDs in Seoul, Korea. However, our results showed that PM₁₀ concentrations inside the subway trains increased after PSDs installation. Although ambient PM₁₀ concentrations decreased from 2008 (before PSDs installation) to 2010 (after PSDs installation), PM₁₀ levels in the subway trains increased. This suggested that the tunnel was isolated by the PSDs, which caused PM₁₀ to accumulate in the tunnel. Furthermore, it was found that the increase in PM₁₀ levels inside the trains was influenced by tunnel depth and the length of the underground segments. This study implies that installation and operation of appropriate ventilation systems inside the subway trains and tunnels are crucial to reduce passengers’ exposure to PM and to protect the public health.

Acknowledgment

The authors thank Prof. Petros Koutrakis and Dr. Jack M. Wolfson for their insightful contributions.

Funding

This study was supported by the Seoul R&D Program (CS070160). This research was also supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science, and Technology (2012R1A6A3A03039668).

Additional information

Notes on contributors

Youn-Suk Son

Youn-Suk Son and Hyung Joo Lee are postdoctoral research fellows in the Department of Environmental Health at Harvard School of Public Health, Boston, MA, USA.

Jae-Sik Jeon

Jae-Sik Jeon is a doctor of Chemical Engineering and a director of Water Environment Research Department and In-Cheol Ryu is a doctor of Environmental Engineering and an Environmental Researcher of Atmospheric Environment Research Department at Seoul Metropolitan Government Research Institute of Public Health and Environment, Seoul, Korea.

Hyung Joo Lee

Youn-Suk Son and Hyung Joo Lee are postdoctoral research fellows in the Department of Environmental Health at Harvard School of Public Health, Boston, MA, USA.

In-Cheol Ryu

Jae-Sik Jeon is a doctor of Chemical Engineering and a director of Water Environment Research Department and In-Cheol Ryu is a doctor of Environmental Engineering and an Environmental Researcher of Atmospheric Environment Research Department at Seoul Metropolitan Government Research Institute of Public Health and Environment, Seoul, Korea.

Jo-Chun Kim

Jo-Chun Kim is a professor in the Department of Environmental Engineering at Konkuk University, Seoul, Korea.