Measurements and PCA / APCS Analyses of Volatile Organic Compounds in Kaohsiung Municipal Sewer Systems , Southern Taiwan

This study measured 71 volatile organic compounds (VOCs) collected using stainless steel canisters at 15 monitoring sites in two main Kaohsiung municipal sewers, A-Sewer and B-Sewer, during winter and summer periods in 2008 and 2009. The results indicate that the overall average of total VOCs, TVOC, in A-Sewer was 1173.51 ± 187.69 μg/m, which was about 40% higher than that in B-Sewer (689.22 ± 151.64 μg/m). However, TVOC in the winter/dry season was about three (or five) times that in the summer/wet season for A-Sewer (or B-Sewer). For the A-Sewer, chlorinated organics, aromatics, and alkanes were predominant, and altogether contributed to about 90.0% of the TVOC in winter and about 70.6% in summer, with alkenes, ethers, ketones, and sulfur compounds as minor components. For B-sewer, aromatics, alkanes, and chlorinated organics were predominant, and altogether contributed to about 94.1% of the TVOC in winter and about 74.3% in summer, while others were minor ones. The principal component analysis (PCA) and absolute component scores (APCS) models indicate that the percentage source contributions for A-Sewer were solvent usage (31.65 ± 11.27%), oil refineries and storage leaks (28.71 ± 11.52%), auto paintings (19.14 ± 9.74%), asphalt plants (17.05 ± 8.73%), and others (3.45 ± 3.95%). The percentage source contributions for B-Sewer were printing factories (45.35 ± 9.19%), oil refineries and storage leaks (31.78 ± 8.59%), solvent usage (18.64 ± 8.50%), and dry cleaning (4.23 ± 4.70%).


INTRODUCTION
Modern cities operate sewage treatment plants that receive wastewater discharged from various sources, including industry, residential households, public institutions, and commercial facilities.Many studies have shown that industries such as oil refineries, plastic and chemical factories, surface-painting shops, and semiconductor manufacturers represent a major source of volatile organic compounds (VOCs) in the atmosphere (Nikolaou et al., 2002;Khwaja et al., 2008;Nian et al., 2008;Cai et al., 2010;Yuan et al., 2010;Leuchner et al., 2010;Vega et al., 2011).Emissions of VOCs into the atmosphere via sewer collection and treatment systems have also received considerable attention.This is mainly because discharges of wastewater to treatment systems via sewers are related to the transfer of volatile organic compounds from the aqueous to gaseous phase, causing various potential risks, such as worker exposure to toxic chemicals during wastewater collection and treatment, emissions of toxic air contaminants and photochemical precursors to the atmosphere, as well as accumulation of explosive gases in sewer confined environments.While several studies have focused on controlling VOCs in wastewater treatment systems (Atasoy et al., 2004;Hamoda, 2006;Cheng et al., 2008), related works have also shown that VOC emissions from manhole covers in the sewer systems should not be ignored (Corsi et al., 1992;Corsi et al., 1995;Escalas et al., 2003).
Despite numerous measurements of individual VOC species in the aqueous phase (Corsi et al., 1995;Kuo et al., 1997;Cheng et al., 2008;Gasperi et al., 2008;Oskouie et al., 2008), the individual gaseous species in a sewer system have seldom been addressed.Since a sewer is a poorly ventilated and confined space, VOCs may accumulate to high levels and pose a health risk to workers.Moreover, highly concentrated hydrocarbons in confined spaces may explode under unfavorable conditions, thus further threatening the environment and human life (Corsi et al., 1995).Understanding the levels of gaseous species in sewer systems is thus essential for environmental and safety concerns.
Kaohsiung City (22°38'N, 120°17'E), in southern Taiwan, is a heavily industrialized harbor city with around 1.52 million inhabitants and an area of around 153.6 km 2 .The many industrial parks located in and around the city include steel production plants, oil refineries, plastic and chemical factories, metal-making plants, power plants, and municipal waste incinerators.Kaohsiung City has two main sewer systems, A-Sewer and B-Sewer (Fig. 1).A-Sewer is approximately 14.5 km long, and collects about 195,000 tons of household and business wastewater a day.Among the many industries located along the A-Sewer are those that produce mechanical devices, electronic spare parts, chemicals and plastics.B-Sewer is approximately 10.8 km long, and collects about 112,000 ton of household and business wastewater a day.Among the many business located along the B-Sewer include those producing cement, paper, textiles, and mechanical devices.While other wastewater may come from agriculture and animal husbandry, both sewers are eventually discharged into the wastewater treatment plant near the coastal site (Fig. 1).
This study measured 71 VOCs in the gaseous phase in the two sewers in the winter/dry and summer/wet seasons in Kaohsiung City in southern Taiwan from 2008-2009.The characteristics and seasonal variations of VOCs are examined, and their dominant source contributions are assessed using PCA and APCS.The results are important for environmental and safety concerns, and form data base for assessing health risks in the future.

SAMPLING AND ANALYSIS
The samples were collected at ten and five monitoring sites for A-Sewer and B-Sewer, respectively (Fig. 1).Samples for a particular sewer were conducted from one site to the next during the same day.Since the distance and starting sample-time between two neighboring sites were all within 1 km and 10 min, and since the residence time of air is usually much longer than that of the wastewater in the sewer, the changes of VOC concentrations were thus assumed negligible during the sampling period.In southern Taiwan, the average rainfall is only around 18.40 mm during the winter season, yet is high as 430.63 mm during the summer season.Thus, winter and summer are herein referred to as dry and wet seasons, respectively.Sampling periods were therefore chosen on December 16-17, 2008 and on June 2-3, 2009 in order to study the variations between winter/dry and summer/wet seasons.
In compliance with the US-EPA Method TO-15 (USEPA, 1999), air samples were collected using 400 mL stainless steel canisters.At each monitoring site, a sampling tube was inserted approximately 1 m into the sewage through a pick hole on a manhole cover.Samples were then collected for 3 min using an air pump at a fixed flow rate of 100 mL/min.Temperature and pressure at each site were also measured in order to calculate VOC concentrations in μg/m 3 , which can be directly used to evaluate emission amounts of VOCs and possible cancer risks when needed in the future.The air samples were then analyzed with a gas chromatograph (GC, Agilent 6890N) and a mass spectrometer (MS, Agilent 5973N).The GC oven temperature was initially programmed at 40°C, rising to 50°C after 2 min, and then increased at 8 °C/min to 230°C, then held for 10 min.Before sampling, all canisters were cleaned, moisturized, and checked for leaks to ensure a vacuum.Six-point calibrations for each species were conducted, yielding linear regression with the coefficient of determination, R 2 , above 0.995.The method of detection limit (MDL) for each species was determined according to US-EPA Test Methods SW-846 (http://www.epa.gov/sw-846/pdfs/chap1.pdf).A known quantity of each standard substance was measured seven times, and the MDL for each species was three times the standard deviation from the seven tests.The MDLs for various species ranged from 0.26 to 4.38 μg/m 3 , with the precision ranging from 81-105% (Table 1).Laboratory blank samples were prepared and analyzed; all data were corrected with reference to a blank.Recovery efficiencies of 92-105% were achieved.

THE PCA/APCS RECEPTOR MODELS
In this study, principal component analysis (PCA) was applied to identify the influences of potential sources.PCA is frequently used in data reduction to identify a small number of factors that explain most of the variance observed in a larger number of variables.It was performed by utilizing the orthogonal transformation method with varimax rotation to determine the eigenvalues of variance matrix of original variables; usually, factors with eigenvalues > 1 were chosen.(Amaya et al., 2009;Wang et al., 2010).Once a factor is determined by PCA, it consists of patterns of variation of the factor loadings of input parameters.The factor loadings indicate the correlation of each species with each component and a species was said to load on a given component if the factor loading was 0.5 or greater for that component.The characteristics of a factor can then be inferred from the dominant pollutants, such as VOCs, PAHs and carbonyl compounds (Ho et al., 2002;Chang et al., 2009;Guo et al., 2009;Wang et al., 2010).
The variances of individual factors in PCA indicate the relative magnitudes among dominant potential sources.Their absolute contributions can then be determined by absolute principal component scores (APCS) using multiple regression of VOC mass concentrations (Ho et al., 2006;Chen et al., 2008).Like all other multivariate receptor models, the PCA and APCS models need adequate source profiles and degrees of freedom to perform the regressions.Furthermore, the two receptor models may not be able to separate sources that are similar.The problem of collinearity can be solved by combining similar sources into single category (Ho et al., 2002;Chen et al., 2008;Wang et al., 2010).
The wintertime mean TVOC was 1766.98 ± 51.24 μg/m 3 and 1178.19 ± 56.80 μg/m 3 for the A-Sewer and B-Sewer, respectively; while the summertime mean TVOC was 580.04 ± 17.18 μg/m 3 and 200.26 ± 6.55 μg/m 3 for the A-Sewer and B-Sewer, respectively.That is, TVOC in the winter/dry season was about three (or five) times that during the summer/wet season for the A-Sewer (or B-Sewer).Most VOC species in the two sewer systems are thus more diluted during the summer/wet period than during winter/dry period, except for ketones.For instance, the acetone concentration was higher during the summer/wet season (90.74 μg/m 3 for A-sewer; 40.04 μg/m 3 for B-Sewer) than in the winter/dry season (49.20 μg/m 3 for A-Sewer; 21.71 μg/m 3 for B-Sewer).As is well known, acetone is an important solvent in the chemical and semiconductor industries (Cheng et al., 2008;Lu et al., 2009).Given its high polarity and easy dissolution in water, acetone may enter a sewer system, with rains causing higher concentrations during the summer/wet period than those during the winter/dry period.Wu et al. (2006) observed similar results, in which acetone was easily found in the sewer system in Shin-Chu Industrial Park in northern Taiwan owing to its high volatility.
In summary, chlorinated organics, aromatics, and alkanes were the three predominant compounds in the two sewers, cumulatively contributing to about 85.1% of the TVOC for the A-Sewer, and 91.2% for the B-Sewer.Chlorinated organics dominated the A-Sewer, since they are largely used for organic solvents and oil emulsion cleaners in industrial parks (Wilkie et al., 1996;Orchard et al., 2000;Ndon et al., 2000;Escalas et al., 2003;Rule et al., 2006).However, aromatics dominated the B-Sewer, which are largely used in the plastic, painting and resin industries (Cheng et al., 2008).

CONCLUSIONS
This study identified 71 volatile organic compounds in seven hydrocarbon groups from Kaohsiung municipal sewers from 2008-2009.Measurement results indicate that the concentrations of most VOC species during the winter/ dry season were higher than those during the summer/wet season.The three most abundant hydrocarbon groups in the two sewers were chlorinated organics, aromatics, and alkanes.
Five principal factors were determined for A-Sewer using PCA, accounting for 91.38% of total variance.The APCS results indicate that the pre-dominant source contributors for A-Sewer were solvent usages (31.65 ± 11.27%), oil refineries and storage leaks (28.71 ± 11.52%), auto paintings (19.14 ± 9.74%), asphalt plants (17.05 ± 8.73%), and others (3.45 ± 3.95%).Four principal factors were determined for B-Sewer using PCA, accounting for 91.40% of total variance.The APCS results indicate that the pre-dominant source contributors for B-Sewer were printing factories (45.

Fig. 1 .
Fig. 1.Location of the sampling sites for A-and B-Sewers in Kaohsiung City.

Fig. 2 .
Fig. 2. Average concentrations of seven hydrocarbon groups for the A-and B-Sewers during winter/dry and summer/wet seasons.

Fig. 3 .
Fig. 3. Average composition (%) of seven hydrocarbon groups for the A-and B-Sewers during winter/dry and summer/wet seasons.

Fig. 4 .
Fig. 4. Average percentage source contributions (% of calculated mass ± SD) to VOCs of APCS results for A-and B-Sewer.

Table 1 .
Method of detection limit (MDL) and precision of 71 VOCs.

Table 2 .
Overall mean concentrations of VOCs for the two sewers (Unit: μg/m 3 ).

Table 3 .
Factor loadings of PCA results for A-Sewer.

Table 4 .
Factor loadings of PCA results for B-Sewer.

Table 4 .
(continued).Only factor loadings > 0.5 are listed and > 0.7 appear in bold.9.19%), oil refineries and storage leaks (31.78 ± 8.59%), solvent usages (18.64 ± 8.50%), and dry cleanings (4.23 ± 4.70%).Since the species and concentrations of VOCs in a sewer may be changed at different time or different day, the results in this work give a preliminary understanding of the behaviors of VOCs in the two sewer systems.