Experimental Investigation of Air Quality in a Subway Station with Fully Enclosed Platform Screen Doors

In this study, the indoor air quality (IAQ) was investigated in a subway station with fully enclosed platform screen doors in Beijing, China. Eight indoor air pollutants, including PM2.5, PM10, SO2 (sulfur dioxide), NO2 (nitrogen dioxide), NH3 (ammonia), CO (carbon monoxide), CH2O (formaldehyde) and TVOC (total volatile organic compound), were measured for six consecutive days in October 2019. The results indicated that the IAQ in the subway station was basically stable at good levels for most times during the whole measurement period. All eight indoor air pollutants were far below their corresponding maximum allowable concentrations, except for the PM2.5 concentrations, which occasionally exceeded the concentration limits. The concentrations of indoor air pollutants in the subway station were basically within the corresponding standards. The correlation analyses showed that outdoor air pollutants have important influences on indoor air pollutants. The concentrations of PM10, PM2.5, SO2, NO2 and CO in the subway station were positively correlated with their corresponding outdoor concentrations. PM10 was statistically significantly correlated with the passenger flow and train frequency, but the other air pollutants were less impacted by the passenger flow and train frequency.


Introduction
The subway system is convenient and efficient and plays an important role in relieving the burdens of superficial traffic congestion. Meanwhile, the electric power system has been adopted in the subway and has improved the air quality of the city [1][2][3]. However, the internal environment of a subway station platform is relatively confined, which can easily cause various types of trace air pollutants to accumulate, which will lead to potential health risks [4,5]. Epidemiological and toxicological studies show that the concentration of particulate matter, NO 2 and SO 2 , can affect the cardiovascular, pulmonary functions and respiratory system [6][7][8]. Short-term exposure to PM 2.5 increases the risk for hospital admission for cardiovascular and respiratory diseases [9]. Long term exposure to PM 2.5 increases respiratory disease, chronic lung disease, and mortality [10]. Inhalable CH 2 O can exacerbate asthma symptoms and act as a human carcinogen [11,12]. Long-term exposure to TVOC can easily result in childhood leukemia [13]. CO is an inorganic compound that can bind with hemoglobin and reduce the oxygen carrying capacity of red blood cells. More than that, exposure to CO may result in vision loss and diabetes [14,15]. NH 3 has toxic effects on the central nervous system of the human body, which can lead to behavioral disorders [16]. Consequently, it is of great significance (1) The dry-bulb temperature was 28 ℃ and the range of relative humidity was 40-70% in the station platform for summer rated conditions.
(2) The total ventilation rate was 5.78 × 10 4 m 3 /h and the fresh air rate was 1.08 × 10 4 m 3 /h. The passenger flow and arrival frequency of train were automatically recorded by the subway control centre. The daily outdoor air pollutant data, including PM2.5, PM10, CO, NO2, SO2 and the outdoor atmospheric environment quality index, were retrieved from the website http://beijingair.sinaapp.com/. The data sampling frequency was 1 h.
(a) Measured position on platform ( b) Platform with full-height screen doors

Data Analysis
Statistical analysis was performed using SPSS 25.00 (Armonk, NY, USA: IBM Corp.) Spearman's correlation analyses were used to examine the relationships between indoor air pollutants and their factors, including the corresponding outdoor concentrations, the train frequency, and the passenger flow. Differences were considered significant when p < 0.05 [46].
In addition, an integrated air quality index (AQI) [47] was adopted to evaluate the indoor air level in the subway station, as shown in Equation (1).
where ci is the concentration of the i th air pollutant, cmaxi is the maximum permission concentration of ci, and n is the number of measured air pollutants (here n = 8).
The integrated AQI can be classified into five levels in consideration of the risks to occupant health, as shown in Table 2 [47].

Data Analysis
Statistical analysis was performed using SPSS 25.00 (Armonk, NY, USA: IBM Corp.) Spearman's correlation analyses were used to examine the relationships between indoor air pollutants and their factors, including the corresponding outdoor concentrations, the train frequency, and the passenger flow. Differences were considered significant when p < 0.05 [46].
In addition, an integrated air quality index (AQI) [47] was adopted to evaluate the indoor air level in the subway station, as shown in Equation (1).
where c i is the concentration of the ith air pollutant, c maxi is the maximum permission concentration of c i , and n is the number of measured air pollutants (here n = 8).
The integrated AQI can be classified into five levels in consideration of the risks to occupant health, as shown in Table 2 [47]. Table 2. Classification standard of integrated article air quality index (AQI).

Integrated AQI Air Level Implication
Air quality is satisfactory. 0.5-1.0 Acceptable Air quality is acceptable. There may be some risks for unusually sensitive groups.
1.0-1.5 Slight One air pollutant exceeds its limit value. There are potential health risks for the susceptive groups. 1.5-2.0 Moderate Two or three air pollutants exceed their limit values. There are health risks.
>2.0 Heavy More than three air pollutants exceed their limit values. There are serious health risks.
According to some indoor air quality standards [48][49][50][51], the maximum permissible concentrations of air pollutants are listed in Table 3.

Passenger Flow and Train Frequency
The passenger flow and train frequency are shown in Figure 2. Day 1 to day 4 represent the weekdays of Tuesday to Friday, and day 5 to day 6 represent the weekend days of Saturday and Sunday. As shown in Figure 2, the train frequency and passenger flow on the weekdays were obviously higher than those on the weekends during the peak hours. The passenger flow peaks in the subway station were at 8:00-9:00 and 18:00-9:00 on weekdays. The average passenger number was 67,126 per hour. The passenger traffic was much busier during the morning peak. There was no clear difference in train frequency and passenger flow during the off-peak hours between weekdays and the weekend.  According to some indoor air quality standards [48][49][50][51], the maximum permissible concentrations of air pollutants are listed in Table 3.

Passenger Flow and Train Frequency
The passenger flow and train frequency are shown in Figure 2. Day 1 to day 4 represent the weekdays of Tuesday to Friday, and day 5 to day 6 represent the weekend days of Saturday and Sunday. As shown in Figure 2, the train frequency and passenger flow on the weekdays were obviously higher than those on the weekends during the peak hours. The passenger flow peaks in the subway station were at 8:00-9:00 and 18:00-9:00 on weekdays. The average passenger number was 67,126 per hour. The passenger traffic was much busier during the morning peak. There was no clear difference in train frequency and passenger flow during the off-peak hours between weekdays and the weekend.    Figure 3 and Table 4 illustrate the variations of indoor air pollutant concentrations in the subway station. The variations of indoor NH 3 concentrations ranged from 0.012 mg/m 3 to 0.014 mg/m 3 , as shown in Figure 3a. The indoor NH 3 concentrations were basically stable at a low level, and did not exceed the maximum permissible concentration of 0.2 mg/m 3 . Figure 3b shows that the concentrations of indoor CH 2 O were from 0.008 mg/m 3 to 0.079 mg/m 3 . Most of the concentrations were below 0.08 mg/m 3 and did not exceed the maximum permissible concentration of 0.12 mg/m 3 . Figure 3c depicts the concentrations of indoor TVOC remaining in the range between 0.374 mg/m 3 and 0.423 mg/m 3 . The TVOC concentrations kept quite consistent during the test period and did not exceed the maximum permissible concentration of 0.6 mg/m 3 .

Air Pollutant Concentrations
The indoor NO 2 concentrations changed notably with time from 0.006 mg/m 3 to 0.127 mg/m 3 , as shown in Figure 3d, but they remained below the maximum permissible concentration of 0.24 mg/m 3 . The indoor NO 2 concentrations increased markedly from 17:00 and reached their peaks at 20:00-21:00, except for on day 3.
In Figure 3e, the indoor SO 2 concentrations fluctuated in the range between 0.001 mg/m 3 to 0.007 mg/m 3 and remained below the maximum permissible concentration of 0.5 mg/m 3 . The indoor SO 2 concentrations rose from 11:00 to their peak values at approximately 16:00, and then decreased. The daily trends were similar throughout the whole test period. Figure 3f shows that the variations of indoor CO concentrations were from 0.046 mg/m 3 to 0.111 mg/m 3 . These were below the maximum permissible concentrations during the test period. From day 1 to day 3, the indoor CO concentrations fluctuated with time. However, the peak values appeared at different times. From day 4 to day 6, the indoor CO concentrations did not obviously fluctuate with time. Hence, the indoor CO concentrations were less impacted by the changes of train frequency and passenger flow.
The concentration ranges of indoor PM 2.5 and PM 10 were from 0.006 mg/m 3 to 0.196 mg/m 3 and from 0.008 mg/m 3 to 0.237 mg/m 3 , respectively, as shown in Figure 3g 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Figure 4 shows variations of indoor AQI during the days of investigation. Most of the days, except for day 2, showed values below 0.5 and remained at a good level. The change range of AQI on day 2 was approximately 0.6-0.7. The AQI level during day 2 was at an acceptable level which was affected by the serious outdoor air pollution.  Figure 4 shows variations of indoor AQI during the days of investigation. Most of the days, except for day 2, showed values below 0.5 and remained at a good level. The change range of AQI on day 2 was approximately 0.6-0.7. The AQI level during day 2 was at an acceptable level which was affected by the serious outdoor air pollution.

Variations of Indoor Air Pollutants
The NH 3 was mostly generated indoors, such as from the toilets on the platform [52]. The indoor NH 3 has been well diluted by the HVAC system to maintain a low level far below the concentration limit. The indoor CH 2 O mainly accumulated from the emissions of building materials, furniture and various adhesive coatings [53]. The change of indoor CH 2 O concentrations could be related to indoor temperature. Higher indoor temperature can be helpful for the release of more CH 2 O from the building finishing materials [54]. This might explain the increase of CH 2 O concentrations which occurred at the morning or evening peaks. The TVOC concentrations remained stable during the test period, because the TVOC mostly came from the building material emissions [55]. In sum, the concentrations of NH 3 and TVOC (including CH 2 O) were mostly generated indoors and kept relatively stable during the test period by the ventilation of the HVAC system. Figure 5 shows the variations of outdoor air pollutant concentrations during the measurement. It can be seen that the daily variations of indoor NO 2 , SO 2 , CO, PM 2.5 and PM 10 concentrations were quite consistent with the corresponding variations of outdoor concentrations. The indoor NO 2 , SO 2 and CO mainly came from the exhaust of motor vehicles introduced through the HVAC system and subway entrances [56]. Similarly, a large portion of indoor PM 10 and PM 2.5 came from the road re-suspension dust and vehicular emissions [57], which were also brought in by the ventilation of HVAC system or directly through the entrances. Meanwhile, most of vehicle exhausts were found to be composed of fine aerosol lower than 2.5 µm. Thus, the daily change trends of indoor PM 2.5 and PM 10 were well correlated (Figure 3g,h), which was consistent with the findings of Park et al. [58] Consequently, the indoor NO 2 , SO 2 , CO, PM 2.5 , PM 10 concentrations basically fluctuated with their corresponding outdoor concentrations. Meanwhile, their indoor concentrations were basically lower than the outdoor concentrations due to the filtration and dilution by the ventilation of HVAC system.
In general, the peaks of indoor concentrations of these five pollutants mainly occurred during the morning or evening rush hours. Therefore, highly congested traffic situations during the peak hours may exacerbate the IAQ of subway station under the ground vehicle road. There were bus stops located next to the subway station entrance so that passengers connect conveniently, which could also contribute to the variations of the pollutants. Table 5 shows the indoor air pollutant concentrations from other references. As shown in Table 5, the studies used for the comparison were mostly conducted in the summer and transitional season, with HVAC systems in operation. In our study, the measurement campaign was performed in late October (transitional season), when the weather in Beijing was mild, but the HVAC system of the subway station was still operating in cooling mode due to the high passenger flow. The average passenger numbers given in the few studies were also comparable to the average passenger flow of the subway station investigated in our study. Most of the previous studies shown in Table 5 have investigated multiple subway stations, but the stations size and ventilation system parameters could not be compared, due to a lack of relevant information in these studies.

Comparison with Previous Studies
The average indoor NH 3 concentration of 0.012 mg/m 3 in our study was relatively low, compared with the NH 3 concentration given in the references [59]. The average indoor CH 2 O and TVOC concentrations were 0.035 mg/m 3 and 0.405 mg/m 3 , which were much higher than the concentrations on the Seoul subway platforms [21] and the Taipei subway platform [60]. They also indicated that the indoor TVOC (including CH 2 O) concentrations had no correlation with the number of passengers, but had a weak correlation with the depth of the platform. This support our findings that the indoor TVOC (including CH 2 O) concentrations could be primarily attributed to the emissions of interior building materials. The higher TVOC concentrations measured in our study were probably caused by the emissions of detrimental decoration materials. The average NO 2 concentration in our study was slightly lower than the average concentration on the Seoul subway platforms [18]. The I/O ratios of NO 2 in our study were also quite similar to the I/O ratios of 0.59-0.74, as indicated in the reference [18]. The higher outdoor concentrations of NO 2 could be attributed to the diesel exhaust fumes from motor vehicles on the roads in urban areas. Consequently, the indoor NO2, SO2, CO, PM2.5, PM10 concentrations basically fluctuated with their corresponding outdoor concentrations. Meanwhile, their indoor concentrations were basically lower than the outdoor concentrations due to the filtration and dilution by the ventilation of HVAC system. In general, the peaks of indoor concentrations of these five pollutants mainly occurred during the morning or evening rush hours. Therefore, highly congested traffic situations during the peak hours may exacerbate the IAQ of subway station under the ground vehicle road. There were bus stops located next to the subway station entrance so that passengers connect conveniently, which could also contribute to the variations of the pollutants.  Consequently, the indoor NO2, SO2, CO, PM2.5, PM10 concentrations basically fluctuated with their corresponding outdoor concentrations. Meanwhile, their indoor concentrations were basically lower than the outdoor concentrations due to the filtration and dilution by the ventilation of HVAC system. In general, the peaks of indoor concentrations of these five pollutants mainly occurred during the morning or evening rush hours. Therefore, highly congested traffic situations during the peak hours may exacerbate the IAQ of subway station under the ground vehicle road. There were bus stops located next to the subway station entrance so that passengers connect conveniently, which could also contribute to the variations of the pollutants.  The average SO 2 concentration in our study was 0.003 mg/m 3 , which was much lower than the concentrations reported in the Guangzhou subway stations [61]. The average indoor CO concentration of 0.059 mg/m 3 was much lower the average concentration reported in the Taipei subway stations [60], but quite comparable with the average concentration in the Nanjing subway stations [62]. There was no indoor source for CO and SO 2 in the subway station, therefore the indoor CO and SO 2 basically came from the contaminated ambient air being brought down from street level. The relatively low indoor CO and SO 2 concentrations in the Beijing subway station indicated a good ventilation performance by the HVAC system.

Comparison with Previous Studies
The average PM 10 concentration of 0.061 mg/m 3 was lower than the concentrations reported in the subway stations in Taipei [60], Nanjing [62] and Seoul [63]. The average PM 2.5 concentration was 0.048 mg/m 3 , which was also lower than the concentrations reported in the references [60] and [58]. The lower PM concentrations observed in our study could be attributed to both the platform screen doors and the good ventilation performance of the HVAC system. There is a certain amount of PM generated from the train operation [26]. Several researchers have indicated that the fully enclosed platform screen doors could help prevent the PM generated by the train operation from entering the platform [26,41]. In addition, the screen doors could also prevent a portion of outdoor air pollutants from entering the platform through the piston wind in the tunnel [42]. Nevertheless, the indoor space of the station would be decreased by installing the fully enclosed screen doors, which might result in a slight increase of other indoor air pollutant concentrations.
It is worth noting that the majority of indoor PM was still introduced from outdoors through the HVAC system and station entrances, which could not be prevented by screen doors. As shown in Table 5, high PM 10 concentrations were observed in the Nanjing subway stations [62], which could be attributed to the ventilation method they used in the transitional season. During the time of sampling, they used natural ventilation systems instead of HVAC systems, which no doubt fully reduced both the ventilation rates and filtration efficiency. Similarly, the high PM 10 concentrations reported in the Seoul subway stations were also caused by insufficient air circulation and improper ventilation [56]. In their study, the PM 10 concentrations on platforms were even obviously higher than those outdoors, because the ventilation was insufficient to remove the accumulated particles brought in from outdoors. Therefore, the proper operation of the HVAC system was also crucial to control the concentrations of indoor PM and other pollutants to maintain them at acceptable levels.   Figure 6a,b. It was reported that some particles would be generated in the subway, due to the friction between the track and the wheel [26]. In addition, when the passenger flows were large, the airborne particulate matter from the floor would be re-suspended, due to the passenger movement around the subway platform [58]. Hence, increased passenger flow may cause an increase in the particle concentration in the subway platform. The I/O ratios of PM 2.5 and PM 10 were within the ranges of 0.77-2.34 and 0.57-1.58, respectively. During most of that time, the indoor PM concentrations were smaller than the outdoor concentrations, which indicated that the fully enclosed platform screen doors could prevent the generation of pollutants from the train running [64]. Thus, the PM in the subway station mainly came from the outdoor environment through the HVAC system and the entrances. On days 3 and 4, the indoor PM concentrations were higher than the outdoor concentrations, which might have been affected by the concentrations of the previous day. The air conditioning system was switched off after the last train every day, possibly resulting in the accumulation of indoor air pollutants on the platform. Therefore, the average indoor air pollutants' concentrations could be affected by the high concentration in the previous day, such as the day 2 in this study.   SO 2 is the combustion product of coal or oil, and is mainly associated with industrial sources [65]. There was no SO 2 production source in the subway station. Indoor SO 2 was mainly affected by the outdoor SO 2 through the ventilation. As shown in Figure 6c, the indoor SO 2 concentrations were mainly consistent with the outdoor SO 2 concentration. The I/O ratios of SO 2 were in the range between 0.44 to 2.15. Similar to the indoor PM concentrations, the indoor SO 2 concentrations were also higher than the outdoor concentrations on days 3 and 4.

Influencing Factors
The indoor NO 2 concentrations were lower than the outdoor NO 2 concentrations, and the I/O ratios were from 0.45 to 0.81, as shown in Figure 6d. It is generally believed that the NO 2 is mainly caused by the emission of outdoor road diesel vehicles [66]. The indoor NO 2 concentrations were influenced by the outdoor NO 2 concentrations.
The indoor CO concentrations were much lower than the outdoor CO concentrations, and the I/O ratios were from 0.06 to 0.12, as shown in Figure 6e. CO is produced by incomplete combustion. The indoor CO concentrations are at relatively low levels, because there is no chemical combustion or smoking in the subway station. Hence, the indoor CO might come from the traffic-contaminated air from outdoors [29]. Table 6 lists the correlation analysis between indoor air pollutants and their influencing factors, including the corresponding outdoor concentrations, train frequency and passenger flow. Furthermore, the correlations between the indoor AQI and the outdoor atmospheric environment quality index, train frequency and passenger flow were also analyzed.

Influencing Factors
The results showed that the indoor PM 10 concentrations were statistically significantly correlated with the outdoor PM 10 concentration (r = 0.858, p< 0.01), the passenger flow (r = 0.201, p < 0.05) and the train frequency (r = 0.209, p < 0.05). Other air pollutant concentrations were strongly correlated with their corresponding outdoor concentrations, but less impacted by the passenger flows and train frequency. The AQI also had a significant correlation with the outdoor atmospheric environment quality index (r = 0.649, p < 0.01). Hence, the outdoor air pollutants had significant contributions to the indoor concentrations through the HVAC system. The variations of indoor concentrations of SO 2 , CO, NO 2 , PM 10 and PM 2.5 were most likely related to their corresponding outdoor concentrations.
According to the correlation analysis, the indoor PM 2.5 concentrations and gaseous pollutants were only correlated to the outdoor environment. In contrast, the indoor PM 10 concentrations were not only affected by the outdoor environment, but also related to the passenger flow and the train frequency. Martins et al. [26] indicated that the PM 2.5 concentrations in subway platforms with screen doors were lower than those in open subway stations. Therefore, the fully enclosed platform screen doors can better prevent the fine particles produced by the trains from moving to the platform.

Conclusions
In this study, eight airborne pollutants in a subway station with fully enclosed screen doors were consecutively measured for six days in Beijing, China. The IAQ performance of the station has been evaluated comprehensively, and compared with previous studies. The potential influencing factors of IAQ were also discussed. Future studies were recommended to investigate more subway stations with different station sizes, passenger flows, platform types and ventilation systems, meanwhile covering more outdoor climate conditions. The main conclusions of this study are summarized as follows: (1) The concentrations of indoor air pollutants on the subway platform were basically within the corresponding standards. The AQI were at good and acceptable levels during the whole measurement.
(2) The concentrations of NH 3 and TVOC (including CH 2 O) were kept relatively stable during the test period, because they were mostly generated from indoor emission sources and were well diluted by the ventilation of HVAC system.
(3) The concentrations of indoor PM 10 , PM 2.5 , SO 2 , NO 2 and CO were positively correlated with their corresponding outdoor concentrations. The daily variations of these indoor air pollutant concentrations were also influenced by the corresponding variations of outdoor concentrations to a large extent. The indoor concentrations were generally lower than the outdoor concentrations, due to the filtration and dilution by the HVAC system.
(4) Except for the indoor PM 10 , the other indoor pollutants and the overall air quality had no statistically significant correlation with the passenger flow and the train frequency. Therefore, the fully enclosed platform screen doors can effectively prevent the fine particles produced by the train operation from moving into the platform area. However, it is worth noting that the indoor pollutants were still mostly introduced from outdoors through the HVAC system and subway entrances, as indicated by the correlation analyses, which could not be prevented by screen doors. The proper operation of HVAC system was also crucial to control the indoor pollutant concentrations at acceptable levels.
Author Contributions: L.P. led the data analysis and drafted the manuscript. C.Y., X.C., Q.T. and B.L. participated in the study design and field study. C.Y. and X.C. participated in interpretation of the data and assisted in the drafting of the manuscript. They conceived the study and contributed a lot to improve the manuscript quality. All authors contributed to, read and approved the final manuscript. All authors have read and agreed to the published version of the manuscript.