- Email: [email protected]
Time and space patterns of volatile organic compounds in a sewage treatment plant
ARTICLE IN PRESS
Water Research 37 (2003) 3913–3920
Time and space patterns of volatile organic compounds in a sewage treatment plant A. Escalasa, J.M. Guadayola, M. Cortinab, J. Riverab, J. Caixachb,* b
a Department of Chemical Engineering, Universitat Polit"ecnica de Catalunya, Colom 1, 08222 Terrassa, Spain Mass Spectrometry Laboratory, Department of Ecotechnologies, IIQAB-CSIC, Jordi Girona Salgado 18, 08034 Barcelona, Spain
Received 11 February 2003; received in revised form 28 May 2003; accepted 3 June 2003
Abstract 47 regulated and non-regulated volatile organic compounds (VOCs) were characterised by closed-loop stripping analysis (CLSA) and high resolution gas chromatography coupled to mass spectrometry (HRGC/MS) in 28 aqueous samples from 4 sampling points along a sewage treatment plant in Manresa, Catalonia, Spain. A 4 22 factorial design (16 samples) was first prepared for the sampling, and reinforced with 12 additional samples at the plant influent. The total analyte weighted mean concentration was 232 mg l1 at the plant influent, with a mass flow of 2231 kg yr1. Petroleum solvents and terpenic compounds accounted for 79% of the influent analyte concentration. VOC concentration in influent was clearly higher for most VOCs from 12 to 22 h (high organic load hours), and lower from 24 to 10 h (lower organic load). Differences between time bands were confirmed through t tests. Differences between weekdays and the weekend were not so clear, and could not be confirmed through t tests. VOC concentrations along the plant are discussed. Overall analyte removal in the plant was 89%. r 2003 Elsevier Ltd. All rights reserved. Keywords: VOCs; CLSA; HRGC/MS; Wastewater; Experimental design
1. Introduction VOCs are present in municipal wastewater due to discharges from domestic, commercial and industrial sources [1]. A reduced number of VOCs were included in the European Union list of priority pollutants for water policies [2]. The number of VOCs in the US list of priority pollutants [3] is higher. A major concern on toxic VOCs in municipal wastewater is their potential air emission from wastewater treatment plants. Toxic air emissions from publicly owned treatment works (POTW) are regulated in the USA [4], but a similar rule has not been issued by the European Union. Several authors have studied VOC concentration and fate in sewage treatment plants. Air emission (stripping and volatilisation), biodegradation and sorption are the *Corresponding author. Tel.: +34-934-006-174; fax: +34932-045-904. E-mail address: [email protected] (J. Caixach).
main processes affecting VOC fate in sewage treatment plants [5]. Air emission prevails in pre-treatment [5–7] and primary clarifiers [6,8], though sorption has also been considered by other authors [5,7]. Air emission, biodegradation and sorption compete in aerobic biological reactors, though sorption is a secondary mechanism for most VOCs [6]. Air emission prevails in secondary clarifiers [6–8]. A review work on toxic air emissions from publicly owned treatment works [1] gathers considerable information on VOC fate from that period. Further work has been done on stripping and volatilisation, including [9,10]. Different commercial and non commercial fate models exist for estimating VOC processes in wastewater treatment plants. According to a review [11], the most comprehensive models were BASTE [5], TOXCHEM+ [7], and WATER9 [6,12]. Purge and trap techniques, combined with HRGC/ MS are usually applied for VOC analysis in wastewaters—EPA Method 624 [13]—, though different,
0043-1354/03/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00336-1
ARTICLE IN PRESS 3914
A. Escalas et al. / Water Research 37 (2003) 3913–3920
more specific techniques are required for some polar compounds. Closed-loop stripping analysis (CLSA) combined with HRGC/MS has been used for analysing volatile and semivolatile organic compounds in aqueous matrices, and is a useful technique for analysing a wide volatility range of VOCs. CLSA provides high recoveries and high concentration factors, as well as satisfactory accuracy and repeatability [14], and has been adopted as a standard method [15]. Few studies have reported the temporal and spatial oscillations of VOC concentration in wastewater in sewage treatment plants [17,18]. In this study, CLSAHRGC/MS has been used for characterising a list of VOCs at time bands of high and low organic load in 4 sampling points along a wastewater treatment train. A two-step strategy was followed for characterising the temporal and spatial VOC regime along the wastewater treatment train, in order to reduce the number of samples to process by CLSA and HRGC/MS.
2. Experimental First, the dissolved organic carbon (DOC) regime was determined in a previous study [19] with a high number of samples. This allowed for the detailed characterisation of the general organic load regime at each sampling point and the setting of time bands of higher and lower organic load. Second, VOC discharges to sewers were assumed to follow the general organic load regime. Then, VOCs were analysed in a moderate number of liquid samples (28) taken at the different DOC time bands, according to a previous sampling design based on a factorial experimental design. The VOC concentration oscillations between time bands were analysed, in order to determine the effect of the day/night and the weekday/ weekend variables on the VOC concentrations. The validness of the assumption of similar behaviour for DOC and VOCs is also checked and discussed below. 2.1. Target compounds A set of VOCs from the list of EPA Method 624 [13] were selected. Other volatile and semivolatile organic compounds from the lists of EPA methods 524.2 and 625 [13] were detected in a previous qualitative study and included in the list. Finally, a list of non regulated terpenic compounds were also detected and added to the analyte list.
Spain. The plant flow diagram has been described elsewhere [19]. 4 sampling points were selected along the wastewater treatment train: influent (A), pre-treatment effluent (B), primary effluent (C), and plant effluent (D). The DOC regime was previously studied [19], and time bands of higher and lower DOC load were determined for each point, for both weekdays and the weekend. Liquid phase samples were taken for VOC analysis at the 4 sampling points. A 22 factorial design was prepared for each sampling point (16 samples). The week was divided into 4 different time bands, according to two design variables: the time of the day (variable H), and the day of the week (variable D). For both weekdays (D+) and weekend days (D), continuous time bands of higher (H+) and lower (H) DOC were determined at each sampling point. The 4 22 factorial design was reinforced with 12 additional samples at the plant influent, in order to have a better characterisation of the VOC load at that point. Grab samples were taken. Clean glass bottles were completely filled, and then capped with Teflon-sealed threaded stoppers. Samples were stored at 4 C until analysed. Table 1 displays, the codification used for the sampling design. Table 2 specifies the time bands corresponding to each experimental condition at each sampling point, as determined from DOC [19]. In this table, the high DOC hours basically correspond to daytime hours (12–22 h), though an important delay is observed in the occurrence of high DOC—until noon—, due to the flow through sewers and the distance from the plant to the city. For simplicity, high/low DOC time bands are also referred as day/night bands in this work.
Table 1 Conventions for the sampling design Variable
Symbol
Dummy variable P for sampling points
+
Dummy variable Q for sampling points
+
Sampling point
X ¼ ðP; QÞ + + ++
Time of the day
H
+
2.2. Sampling The sewage treatment plant of Manresa and Sant Joan de Vilatrorrada was selected for the study. This is a 134,000 population equivalents plant in Catalonia,
Value Meaning
Day of the week
D
+
A=Plant influent B=Pre-treatment effluent C=Primary effluent D=Plant effluent Time band of lower DOC Time band of higher DOC Weekend Weekday
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920 Table 2 Time bands for the 4 sampling conditions Sampling point
Experimental conditions (H, D)
+
+
++
A
2410 h
1222 h
2410 h
1222 h
B
210 h
1222 h
2410 h
1222 h
C
214 h
1624 h
814 h
166 h
D
186 h
816 h
820 h
226 h
3915
centration. The response factor was calculated for each analyte, relative to one of the 1-chloroalkanes. A characteristic m/z (quantitation ion) was selected for each compound. The unknown concentration of each analyte in a sample was obtained from its specific response factor, through the formula: Cz=AzCi/RzAi, where Cz and Ci are the concentrations of respectively the analyte and the internal standard; Az and Ai are the peak areas of respectively the analyte and the internal standard; and Rz is the response factor of the analyte.
3. Results and discussion 2.3. CLSA extraction CLSA extraction was carried out in a commercial apparatus (Brechbuler, . Zurich, . Switzerland), in accordance with the standard method 6040 B [15]. 30–450 ml of wastewater sample were diluted to 1 l. Then the samples were air-stripped for 70 min in a bath at 45 C, and the VOCs were adsorbed on a 5-mg activated carbon filter at 55 C. The filters were then extracted with 40 ml of carbon disulphide. 2.4. HRGC/MS conditions The CLSA extracts were processed by HRGC/MS. A Carlo Erba (Milan, Italy) 800 series chromatograph was used for analyte separation, coupled to a Fisons (Manchester, UK) MD800 quadrupole mass spectrometer, for identification and quantitation. The chromatograph was equipped with a 60 m 0.25 mm I.D. DB-5 column with 0.25 mm film thickness (J&W Scientific Folsom, CA, USA), fitted with a 2-m deactivated fusedsilica capillary pre-column. Injections were 1-ml cold oncolumn, using helium as carrier gas at a linear velocity of 30.8 cm s1 at 70 C. The column temperature program was: 35 C (held for 5 min) at 4 C min1 to 280 C (held for 10 min). The mass spectrometer conditions were: electron impact ionisation at 70 eV, acquisition mode: scan in the range of 35–350 m/z at 1 s/scan, ion source temperature of 200 C and interface temperature of 250 C. 2.5. Quantitation 1-chloroalkanes (C6, C10, C12, C16) were used as internal standards. The wastewater samples were spiked with internal standards at an individual concentration of 40 ng ml1 on the final extract. Two other internal standards (1-chloroalkanes C8 and C14) were added during the extraction procedure, for recovery control. Individual recovery factors were determined for each analyte, by CLSA-processing solutions of known con-
3.1. Influent VOC characterisation Table 3 shows the results of the influent characterisation. The compounds listed in key EU or US regulations are marked with X. The samples were qualitatively rich in target compounds. Only 6 of 47 targets were not detectable in any sample, while 35 analytes were detected in all experimental conditions. High concentration variability was observed, as indicated by the standard deviations and the 95% percentile. The flow-weighted mean concentration and mass flow of analytes at the plant influent were calculated. The total analyte weighted concentration at the plant influent was 232 mg l1, with a mean mass flow of 6380 g d1. A list of 15 VOCs accounted for 201 mg l1 or 87% of the total weighted concentration. 5 target compounds can be classified as carcinogen, mutagenic or toxic to reproduction (CTM) under article 5, paragraph 6 of the 1999/13/EC directive [20]. These compounds accounted for 926 g d1 or 14.5% of the total mass flow of analytes at the plant influent. The 1999/13/EC directive does not affect emissions from sewage treatment plants, and there are no specific EU regulations for air emissions from wastewater treatment plants. The US regulations can be used as a reference. 17 hazardous air pollutants (HAPs), as defined by EPA [16] were included in the target compound list. 16 of them were detected. They accounted for 2240 g d1 or 35% of the total mass flow of analytes at the plant influent. An annual extrapolation of these data yields 817 kg yr1 of HAPs. When compared to the US regulations, the mass flow of HAP analytes at the plant influent is much lower than the air emission limits set in the POTW rule [4] for new or reconstructed POTWs. These limits are 10 ton yr–1 (9072 kg yr–1) of a single HAP or 25 ton yr1 (22,680 kg yr–1) of the sum of 188 HAPs. Though a limited number of HAPs have been analysed in this study, the list of target compounds includes the most representative HAPs present in municipal wastewater, according to previous studies [1]. Besides, only a fraction of the influent VOC flow is emitted to the atmosphere
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920
3916
Table 3 VOC characterisation of the plant influent Compound
Mean (mg 11)
s (mg 11)
Samples with c> MDL
95% percentile
Priority EUa
Priority USAb
CMTc
HAPd
Chloroform 1,1,1-Trichloroethane Benzene Carbon tetrachloride Trichloroethylene Dimethyl disulphide Toluene Perchloroethylene Chlorobenzene Ethylbenzene m-Xylene and p-xylenee Styrene o-Xylene Cumene Bromobenzene a-Pinene Camphene o-Chlorotoluene n-Propylbenzene p-Chlorotoluene 1,3,5-Trimethylbenzene Dimethyl trisulphide tert-Butylbenzene 1,2,4-Trimethylbenzene Decane m-Dichlorobenzene p-Dichlorobenzene sec-Butylbenzene 1,2,3-Trimethylbenzene p-Cymene D-Limonene 1,8-Cineol o-Dichlorobenzene n-Butylbenzene Fenchone Linalool 1,3,5-Trichlorobenzene Camphor 1,2,4-Trichlorobenzene Menthol Naphtalene a-Terpineol 1,2,3-Trichlorobenzene Hexachloro-l,3-butadiene Dimethyl tetrasulphide Citronellol Bornyl acetate
9.5 1.3 7.6 1.0 6.1 1.6 23 1.4 0.25 1.7 6.6 0.27 2.2 0.11 oMDL 0.74 0.038 oMDL 0.9 oMDL 1.3 3.8 oMDL 14 49 oMDL 0.17 0.083 1.4 0.63 4.0 2.8 0.34 0.010 0.10 9.2 1.8 13.2 0.40 5.4 4.1 1.9 0.10 oMDL 3.9 3.5 2.4
7.6 2.0 9.1 1.4 15 2.0 24 1.5 0.19 2.7 8.1 0.38 2.4 0.27 — 0.86 0.049 — 2.8 — 3.6 5.4 — 43 96 — 0.13 0.30 2.7 1.2 3.8 2.2 1.2 0.039 0.14 6.3 6.6 11.4 1.6 4.2 3.4 1.5 0.12 — 6.0 3.2 2.4
15 15 16 16 16 16 16 16 16 16 16 16 16 14 0 16 9 0 16 0 16 15 0 16 15 0 14 5 16 14 16 15 5 1 7 16 15 15 6 16 15 16 12 0 7 12 15
0.8024 0.042–6.3 0.23–29 0.18–4.4 0.19–45 0.16–6.0 4.0–77 0.25–4.7 0.083–0.68 0.084–8.5 1.2–27 0.065–1.2 0.51–8.1 oMDL–0.78 — 0.12–2.9 oMDL–0.15 — 0.046–7.5 — 0.086–10 0.038–15 — 0.72–113 0–278 — oMDL–0.43 oMDL–0.76 0.075–8.3 oMDL–3.7 0.32–11.4 0.14–6.6 oMDL–3.2 oMDL–0.10 oMDL–0.36 0.76–20 0.011–16.7 0.056–36 oMDL–4.0 0.009–13 0.48–11 0.13–4.9 oMDL–0.38 — oMDL–16 oMDL–8.9 0.060–6.4
MDL: Method detection limit. a Priority substance in water, under the 2455/2001/EC decision or the list I of the 76/464/EEC directive. b Priority toxic pollutant [3]. c Carcinogen, mutagen or toxic to reproduction under the 1999/13/EC directive [20]. d Hazardous air pollutant in the USA [16]. e m- and p-xylene were coeluted.
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920 Table 4 Summary of weighted data for chemical groups at the influent (A) Group
Decane Chlorinated aliphatic hydrocarbons Light aromatic hydrocarbons Chlorinated aromatic hydrocarbons Sulphur compounds Terpenic compounds Totals
Mean weighted concentration (mg l1)
Mean weighted mass flow (g d1)
% of total
51 28
1414 761
22 12
90
2482
39
70
1
7
189
3
53
1467
23
232
6383
100
2.5
(the rest is biodegraded, sorbed or discharged). Therefore, the total HAP emissions would probably fall well below the limits set in the US POTW rule. The analytes were grouped by chemical constitution in six groups, and the aggregate weighted mean concentration of each grouped was computed (Table 4). Light aromatic hydrocarbons (39%), terpenic compounds (23%) and decane (22%) were the most abundant chemical groups of analytes at the plant influent. 3.2. Effects of time and day at the plant effluent The effect of the time of the day (H effect) and the day of the week (D effect) were computed as the differences between the mean concentrations obtained at the + and values of the independent variables [21]. Many effects were found close to zero at both the negative and the positive sides. Therefore, their statistical significance was evaluated. Effects are differences between means, and their statistical significance can be evaluated through t tests for difference between means. A t test was run for each analyte and effect, at a significance level of 0.05. A previous F test was run for each compound and effect, in order to determine if the variances were equal in the two populations being compared, and to determine the kind of t test to be applied (t test for equal or different variances). F and t tests were applied following the procedure described elsewhere [22]. The H effect was positive for 21 of 39 VOCs analysed, at a 0.05 significance level. The H effect was not significant for the remaining 18 VOCs, at 0.05 significance level. The sum of the significant positive effects
3917
was +146 mg l1. This means that the aggregated concentration of these 21 analytes was 146 mg l1 higher during the time bands of high DOC (187 mg l1) than during the time bands of low DOC (41 mg l1). This represents a 350% increase over the low DOC time band. The D effect was positive for 22 VOCs and negative for 17. A significance study through t tests led to the conclusion that the D effect was not significant for 34 of 39 target compounds, at 0.05 significance level. This means that for most VOCs, the concentration is not significantly higher during the weekdays than during the weekend, at the plant influent. This is not the case for the DOC, which was higher during the weekdays. However, the difference between DOC on weekdays and weekends was only 21 mg l1, vs. 35 mg l1 between the hours of high and low DOC. Analysing COD data from a whole year, the mean weekend COD was 402 mg l1 while the mean weekday COD was 472 mg l1. Though these values are significantly different at 0.05 significance level, the general organic load during the weekdays is not very different from the organic load during the weekend. Then the D effect results are not much different of those of the general organic load when expressed as COD. The HD interaction was not significant for 37 of 39 target compounds, at 0.05 significance level. In conclusion, for a vast majority of target compounds, neither D or HD were significant in determining the COV concentration at the plant influent. For a slight majority of VOCs (21/39), the H effect was significant and positive, this meaning that these compounds present higher concentrations at the hours of higher DOC. However, an important fraction of target compounds (18/39) were not sensible to the H effect, and did not present higher concentrations during the hours of higher DOC. The assumption of same VOC and DOC regime at the plant influent was correct for a majority of analytes, in the case of the day/night regime (H effect). It was not correct for the weekday/weekend regime (D effect), since most analytes were not affected by this factor. In this case, even DOC or COD presented smaller differences between the weekdays and the weekend.
3.3. VOC potential sources and day/night regime at the plant influent A bibliographic study was carried out on the sources of the 39 analytes included in the H effect analysis. 11 of 39 compounds were associated to petroleum solvents, applied in industrial degreasing and vehicle washing [23]. This group includes decane and all aromatic hydrocarbon analytes, except benzene. 10 of 11 petroleum solvent compounds presented significant positive H
ARTICLE IN PRESS 3918
A. Escalas et al. / Water Research 37 (2003) 3913–3920
effects, in accordance with the diurnal activity of most industries in Manresa. The terpenic compounds can be associated to aroma compounds included in powder detergents, liquid detergents and cleaners, and fabric conditioners. Powder detergents contain 0.6–1.2 mg/kg of fragrance additives, 25% of them being terpenic compounds, while fabric conditioners may present a higher content [24]. 9 of 11 terpenic compounds presented positive H effects, though only 5 were significant at 0.05 significance. No terpenic compound presented a significant negative H effect. These results fit reasonably well with the day/night cycle of activity, both domestic, commercial and industrial in Manresa. 6 analytes are chlorinated solvents from diverse origin: water disinfection (chloroform), dry cleaning (perchloroethylene), industrial degreasing (trichloroethylene and 1,1,1-trichloroethane) and other uses (chlorobenzene and o-dichlorobenzene). The H effect was not significant for chloroform, as expected from the continuous nature of tap water chlorination. Trichloroethylene presented a high, statistically significant positive H effect, what is in accordance with its use as industrial solvent. The other chlorinated solvents presented lower concentrations and no significant H effects. Dimethyl di-, tri- and tetrasulphide are biogenic odor compounds, related to the biodegradation of organic matter. All presented significant positive H effects, in accordance with the higher organic matter concentration in wastewater during daytime. Styrene is a cross-linking agent and sec-butylbenzene a plasticiser in the polymer industry. Both presented significant positive H effects. pdichlorobenzene is used as antimoth and deodoriser in toilets. Given the contact with water in the last application, this seems the most probable main source for p-dichlorobenzene in wastewater. This compound did not present a significant H effect at 0.05 confidence level. A main source could not be assigned to the other analytes, and they did not have significant H effects. 3.4. Evolution along the wastewater treatment train For this study, the analytes being detected at all points A, B, and C were only considered (34 analytes). Table 5 displays the arithmetic means of VOC concentrations for the 34 analytes at each sampling point. In order to study the evolution of VOC concentration along the wastewater treatment train, the means of analyte concentrations at points B, C, and D were compared with those of point A, through t tests for mean comparison. The total analyte concentration increased from the plant influent (187 mg l1) to the pre-treatment effluent (215 mg l1). 18 of 34 analytes increased their concentration between A and B, while the remaining 16 analyte concentrations decreased. A decrease should be ex-
Table 5 Mean concentrations of analytes at each sampling point Concentration (mg 11) A
B
C
D
Chloroform 1,1,1-Trichloroethane Benzene Carbon tetrachloride Trichloroethylene Dimethyl disulphide Toluene Perchloroethylene Chlorobenzene Ethylbenzene m-Xylene and p-xylene Styrene o-Xylene Cumene a-Pinene n-Propylbenzene 1,3,5-Trimethylbenzene Dimethyl trisulphide 1,2,4-Trimethylbenzene Decane p-Dichlorobenzene 1,2,3-Trimethylbenzene p-Cymene D-Limonene 1,8-Cineol Linalool 1,3,5-Trichlorobenzene Camphor Menthol Naphtalene a-Terpineol Dimethyl tetrasulphide Citronellol Bornyl acetate
9.5 1.3 7.6 1 6.1 1.6 23 1.4 0.25 1.7 6.6 0.27 2.2 0.11 0.74 0.92 1.3 3.8 14 49 0.17 1.4 0.63 4 2.8 9.2 1.8 13 5.4 4.1 1.9 3.9 3.5 2.4
4.4 1.3 5.8 0.59 1.4 8.4 19 1.5 0.31 2.9 13 0.086 5.3 0.15 0.64 1.4 2.7 12 35 35 0.13 0.62 0.22 1.5 6.6 9.8 0.14 12 7.7 5.8 2.2 9.5 6.3 1.2
5.5 0.28 0.61 0.83 2.2 2.8 11 2.3 0.11 1.3 5.9 0.68 2.4 0.11 0.56 0.49 0.73 6 7 17 0.19 0.73 0.36 1.8 2.1 6.4 0.2 6.5 3.7 7.5 0.98 5.2 1.8 0.94
2.9 0.26 1.3 0.26 0.56 0.48 7.3 0.2 0.12 0.36 1.1 0.059 0.38 0.032 0.098 0.088 0.097 0.94 1.2 oMDL 0.052 0.08 0.048 0.086 0.19 0.31 0.026 1.4 0.44 0.21 0.032 0.3 oMDL 0.096
Totals
187
215
106
21
pected, since 20–30% of influent VOCs are emitted through pre-treatment and primary treatment [1], and a maximum 10% is emitted from the primary clarifier [25]. In a t test at 0.05 significance level, 29 of 34 analytes presented equal mean concentrations. Chloroform (cA>cB) and—with cB>cA—dimethyl disulphide, oxylene, dimethyl trisulphide, and 1,8-cineol, presented significantly different concentrations. Since the supernatant of the digested sludge is recycled to the pretreatment, the increase in the concentration of sulphur compounds could be attributed to their biogenic generation in the anaerobic digesters. Besides, a higher number of samples would be necessary at the point B, given the high concentration variance. A comparison between the influent (A) and the primary effluent (C) led to results similar to those of
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920
3919
100 90
Mean concentration, mg l
-1
80 70 60 50 40 30
Aromatic HC Decane Terpenic compounds Chlorinated aliphatic HC Sulphur compounds
20 10 0
A
B
Chlorinated aromatic HC C
D HC: hydrocarbons
Fig. 1. Evolution of analyte concentrations, grouped by their chemical constitution, along the sewage treatment plant.
the A–B comparison, in spite of the considerable difference between A and C total analyte concentration. No significant difference was observed for any analyte between points A and C, at 0.05 significance level. Again, the VOC losses through the pre-treatment and the primary treatment should be noticeable. However, given the high variances of the analyte concentrations, a higher number of samples would be necessary at C for detecting the differences. Finally, all analyte concentrations were lower at the plant effluent (D) than at the plant influent (A), at a 0.05 significance level, as expected. The overall percent removal was 89%. For the 34 analytes considered, a daily discharge through the plant effluent was estimated as 681 g d1, distributed as follows: decane (o0.5 g d1), chlorinated aliphatic hydrocarbons (125 g d1), light aromatic hydrocarbons (391 g d1), halogenated aromatic hydrocarbons (6.1 g d1), sulphur compounds (78 g d1), and terpenic compounds (81 g d1). Fig. 1 represents the mean concentration of these groups of analytes at the four sampling points in the plant.
4. Conclusions 1. 47 regulated and non regulated volatile organic compounds (VOCs) were characterised by closedloop stripping analysis (CLSA) and high resolution gas chromatography coupled to mass spectrometry (HRGC/MS) in 28 aqueous samples from 4 sampling points along a sewage treatment plant.
2. The sum of the weighted mean concentration of analytes was 232 mg l1 at the plant influent. The mass flow of analytes at the influent was 6383 g d1. The overall VOC removal along the wastewater treatment train was 89%. 3. At the plant influent, a majority of analytes (21 of 39) followed the general organic load regime, based on DOC. This set of compounds presented an overall 350% increase in concentration during high DOC time bands vs. low DOC time bands (H effect). However, 18 analytes were not significantly affected by the time of the day. No analyte presented a trend significantly opposite to the general organic load regime. 4. Petroleum solvents, and terpenic compounds from surfactant-containing preparations accounted for 79% of the analyte concentration at the plant influent. Most of these compounds followed the general organic load regime, characterised by a mostly diurnal activity in the city. 5. The day of the week (D effect: weekdays vs. weekend) did not significantly affected the majority of analytes (34 of 39) at the plant influent. The HD interaction was not significant for 37 of 39 analytes.
Acknowledgements We appreciate technical support from Ms. Anna M. ! from Aigues Lupon, . de Manresa SA. We also thank
ARTICLE IN PRESS 3920
A. Escalas et al. / Water Research 37 (2003) 3913–3920
Mr. Ricard Tomas, from the same company, for access to the plant. [13]
References [1] WEF-ASCE. Toxic Air Emissions from Wastewater Treatment Facilities. Alexandria VA, USA: Water Environment Federation-American Society of Civil Engineers; 1995. [2] Decision 2455/2001/EC of the European Parliament and of the Council of 20 November 2001 establishing the list of priority substances in the field of water policy and amending Directive 2000/60/EC. Off J Eur Communities L 331; 15/12/2001: 1–5. [3] EPA. Toxics criteria for those states not complying with Clean Water Act section 303(c)(2)(B). US Environmental Protection Agency: 40 CFR 131.36. [4] EPA. National emission standards for hazardous air pollutants: publicly owned treatment works. US Environmental Protection Agency. Federal Register 1999, 64, 206, 57572. [5] Govind R, Lai L, Dobbs R. Integrated model for predicting the fate of organics in wastewater treatment plants. Environ Prog 1991;10(1):13–23. [6] EPA. Air emissions models for waste and wastewater. Document EPA-453/R-94-080. Research Triangle Park, NC, USA: EPA Office of Air Quality Planning and Standards; 1994. [7] Melcer H, Bell JP, Thompson DJ, Yendt CM, Kemp J, Steel P. Modeling volatile organic contaminants in wastewater treatment plants. J Environ Eng 1994;120(3): 588–609. [8] Corsi RL, Card TR. Estimation of VOC emissions using the Baste model. Environ Prog 1991;10(4):290–9. [9] Lee KC, Rittmann BE, Shi J, McAvoy D. Advanced steady-state model for the fate of hydrophobic and volatile compounds in activated sludge. Water Environ Res 1998;70(6):1118–31. [10] Chern JM, Chou SR. Volatile organic compound emission rates from mechanical surface aerators: mass-transfer modelling. Ind Eng Chem Res 1999;38(8):3176–85. [11] Tata P. Methodology for the determination of VOC emissions. Prepared for inclusion in AMSA MACT Guidance Document. Metropolitan Water Reclamation District of Greater Chicago, 1999. [12] EPA. WATER9 [on line]. EPA Office of Air Quality Planning and Standards. URL: http://www.epa.gov/ttn/
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21] [22]
[23]
[24]
[25]
chief/software/water/index.html. [Consulted on 3 April 2003]. EPA. EPA Methods and guidance for analysis of water [CD-ROM]. Washington DC: EPA Office of Water; 1999. Malleret L, Bruchet A, Hennion MC. Picogram determination of ‘‘Earthy-Musty’’ odorous compounds in water using modified closed loop stripping analysis and large volume injection GC/MS. Anal Chem 2001;73(7): 1485–90. APHA-AWWA-WPCF (1995). Standard methods for the examination of water and wastewater. Washington, DC: American Public Health Association—American Water Works Association—Water Pollution Control Federation; 1995. EPA. EPA—Air Toxics Website original list of hazardous air pollutants [on line]. URL: http://www.epa.gov/ttn/atw/ orig189.html. Consulted on 8 January 2003. Dunovant VS, Clark CS, Que Hee SS, Hertzberg VS, Trapp JH. Volatile organics in the wastewater and airspaces of three wastewater treatment plants. J Water Pollut Control Fed 1986;58(9):886–95. Simonich SL, Begley WM, Debaere G, Eckhoff WS. Trace analysis of fragrance materials in wastewater and treated wastewater. Environ Sci Technol 2000;34(6):959–65. Escalas A, Droguet M, Guadayol JM, Caixach J. Estimating DOC regime in a wastewater treatment plant by UV deconvolution. Water Res 2003;37(11): 2627–35. Council Directive 1999/13/EC of 11 March 1999 on the limitation of emissions of volatile organic compounds due to the use of organic solvents in certain activities and installations. Off J Eur Communities L 085; 29/03/1999: 1–22. Montgomery DC. Design and analysis of experiments, 4th ed. New York: Wiley; 1997. p. 228. Montgomery DC, Runger GC. Applied statistics and probability for engineers, 1st ed. Mexico D.F., Mexico: McGraw-Hill; 1996. p. 396–399, 410–412 and 431–432 [in Spanish]. Pax!eus N. Vehicle washing as a source of organic pollutants in municipal wastewater. Water Sci Technol 1996;33(6):1–8. Gavald"a A. Personal communication from Albert Gavald"a, aroma expert from Aromas Maluquer SA, Saragossa 91, 08912 Badalona, Spain, 2000. Pincince DR. Transfer of oxygen and emissions of volatile organic compounds at clarifier weirs. J Water Pollut Control Fed 1995;63(2):77–119.
Water Research 37 (2003) 3913–3920
Time and space patterns of volatile organic compounds in a sewage treatment plant A. Escalasa, J.M. Guadayola, M. Cortinab, J. Riverab, J. Caixachb,* b
a Department of Chemical Engineering, Universitat Polit"ecnica de Catalunya, Colom 1, 08222 Terrassa, Spain Mass Spectrometry Laboratory, Department of Ecotechnologies, IIQAB-CSIC, Jordi Girona Salgado 18, 08034 Barcelona, Spain
Received 11 February 2003; received in revised form 28 May 2003; accepted 3 June 2003
Abstract 47 regulated and non-regulated volatile organic compounds (VOCs) were characterised by closed-loop stripping analysis (CLSA) and high resolution gas chromatography coupled to mass spectrometry (HRGC/MS) in 28 aqueous samples from 4 sampling points along a sewage treatment plant in Manresa, Catalonia, Spain. A 4 22 factorial design (16 samples) was first prepared for the sampling, and reinforced with 12 additional samples at the plant influent. The total analyte weighted mean concentration was 232 mg l1 at the plant influent, with a mass flow of 2231 kg yr1. Petroleum solvents and terpenic compounds accounted for 79% of the influent analyte concentration. VOC concentration in influent was clearly higher for most VOCs from 12 to 22 h (high organic load hours), and lower from 24 to 10 h (lower organic load). Differences between time bands were confirmed through t tests. Differences between weekdays and the weekend were not so clear, and could not be confirmed through t tests. VOC concentrations along the plant are discussed. Overall analyte removal in the plant was 89%. r 2003 Elsevier Ltd. All rights reserved. Keywords: VOCs; CLSA; HRGC/MS; Wastewater; Experimental design
1. Introduction VOCs are present in municipal wastewater due to discharges from domestic, commercial and industrial sources [1]. A reduced number of VOCs were included in the European Union list of priority pollutants for water policies [2]. The number of VOCs in the US list of priority pollutants [3] is higher. A major concern on toxic VOCs in municipal wastewater is their potential air emission from wastewater treatment plants. Toxic air emissions from publicly owned treatment works (POTW) are regulated in the USA [4], but a similar rule has not been issued by the European Union. Several authors have studied VOC concentration and fate in sewage treatment plants. Air emission (stripping and volatilisation), biodegradation and sorption are the *Corresponding author. Tel.: +34-934-006-174; fax: +34932-045-904. E-mail address: [email protected] (J. Caixach).
main processes affecting VOC fate in sewage treatment plants [5]. Air emission prevails in pre-treatment [5–7] and primary clarifiers [6,8], though sorption has also been considered by other authors [5,7]. Air emission, biodegradation and sorption compete in aerobic biological reactors, though sorption is a secondary mechanism for most VOCs [6]. Air emission prevails in secondary clarifiers [6–8]. A review work on toxic air emissions from publicly owned treatment works [1] gathers considerable information on VOC fate from that period. Further work has been done on stripping and volatilisation, including [9,10]. Different commercial and non commercial fate models exist for estimating VOC processes in wastewater treatment plants. According to a review [11], the most comprehensive models were BASTE [5], TOXCHEM+ [7], and WATER9 [6,12]. Purge and trap techniques, combined with HRGC/ MS are usually applied for VOC analysis in wastewaters—EPA Method 624 [13]—, though different,
0043-1354/03/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00336-1
ARTICLE IN PRESS 3914
A. Escalas et al. / Water Research 37 (2003) 3913–3920
more specific techniques are required for some polar compounds. Closed-loop stripping analysis (CLSA) combined with HRGC/MS has been used for analysing volatile and semivolatile organic compounds in aqueous matrices, and is a useful technique for analysing a wide volatility range of VOCs. CLSA provides high recoveries and high concentration factors, as well as satisfactory accuracy and repeatability [14], and has been adopted as a standard method [15]. Few studies have reported the temporal and spatial oscillations of VOC concentration in wastewater in sewage treatment plants [17,18]. In this study, CLSAHRGC/MS has been used for characterising a list of VOCs at time bands of high and low organic load in 4 sampling points along a wastewater treatment train. A two-step strategy was followed for characterising the temporal and spatial VOC regime along the wastewater treatment train, in order to reduce the number of samples to process by CLSA and HRGC/MS.
2. Experimental First, the dissolved organic carbon (DOC) regime was determined in a previous study [19] with a high number of samples. This allowed for the detailed characterisation of the general organic load regime at each sampling point and the setting of time bands of higher and lower organic load. Second, VOC discharges to sewers were assumed to follow the general organic load regime. Then, VOCs were analysed in a moderate number of liquid samples (28) taken at the different DOC time bands, according to a previous sampling design based on a factorial experimental design. The VOC concentration oscillations between time bands were analysed, in order to determine the effect of the day/night and the weekday/ weekend variables on the VOC concentrations. The validness of the assumption of similar behaviour for DOC and VOCs is also checked and discussed below. 2.1. Target compounds A set of VOCs from the list of EPA Method 624 [13] were selected. Other volatile and semivolatile organic compounds from the lists of EPA methods 524.2 and 625 [13] were detected in a previous qualitative study and included in the list. Finally, a list of non regulated terpenic compounds were also detected and added to the analyte list.
Spain. The plant flow diagram has been described elsewhere [19]. 4 sampling points were selected along the wastewater treatment train: influent (A), pre-treatment effluent (B), primary effluent (C), and plant effluent (D). The DOC regime was previously studied [19], and time bands of higher and lower DOC load were determined for each point, for both weekdays and the weekend. Liquid phase samples were taken for VOC analysis at the 4 sampling points. A 22 factorial design was prepared for each sampling point (16 samples). The week was divided into 4 different time bands, according to two design variables: the time of the day (variable H), and the day of the week (variable D). For both weekdays (D+) and weekend days (D), continuous time bands of higher (H+) and lower (H) DOC were determined at each sampling point. The 4 22 factorial design was reinforced with 12 additional samples at the plant influent, in order to have a better characterisation of the VOC load at that point. Grab samples were taken. Clean glass bottles were completely filled, and then capped with Teflon-sealed threaded stoppers. Samples were stored at 4 C until analysed. Table 1 displays, the codification used for the sampling design. Table 2 specifies the time bands corresponding to each experimental condition at each sampling point, as determined from DOC [19]. In this table, the high DOC hours basically correspond to daytime hours (12–22 h), though an important delay is observed in the occurrence of high DOC—until noon—, due to the flow through sewers and the distance from the plant to the city. For simplicity, high/low DOC time bands are also referred as day/night bands in this work.
Table 1 Conventions for the sampling design Variable
Symbol
Dummy variable P for sampling points
+
Dummy variable Q for sampling points
+
Sampling point
X ¼ ðP; QÞ + + ++
Time of the day
H
+
2.2. Sampling The sewage treatment plant of Manresa and Sant Joan de Vilatrorrada was selected for the study. This is a 134,000 population equivalents plant in Catalonia,
Value Meaning
Day of the week
D
+
A=Plant influent B=Pre-treatment effluent C=Primary effluent D=Plant effluent Time band of lower DOC Time band of higher DOC Weekend Weekday
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920 Table 2 Time bands for the 4 sampling conditions Sampling point
Experimental conditions (H, D)
+
+
++
A
2410 h
1222 h
2410 h
1222 h
B
210 h
1222 h
2410 h
1222 h
C
214 h
1624 h
814 h
166 h
D
186 h
816 h
820 h
226 h
3915
centration. The response factor was calculated for each analyte, relative to one of the 1-chloroalkanes. A characteristic m/z (quantitation ion) was selected for each compound. The unknown concentration of each analyte in a sample was obtained from its specific response factor, through the formula: Cz=AzCi/RzAi, where Cz and Ci are the concentrations of respectively the analyte and the internal standard; Az and Ai are the peak areas of respectively the analyte and the internal standard; and Rz is the response factor of the analyte.
3. Results and discussion 2.3. CLSA extraction CLSA extraction was carried out in a commercial apparatus (Brechbuler, . Zurich, . Switzerland), in accordance with the standard method 6040 B [15]. 30–450 ml of wastewater sample were diluted to 1 l. Then the samples were air-stripped for 70 min in a bath at 45 C, and the VOCs were adsorbed on a 5-mg activated carbon filter at 55 C. The filters were then extracted with 40 ml of carbon disulphide. 2.4. HRGC/MS conditions The CLSA extracts were processed by HRGC/MS. A Carlo Erba (Milan, Italy) 800 series chromatograph was used for analyte separation, coupled to a Fisons (Manchester, UK) MD800 quadrupole mass spectrometer, for identification and quantitation. The chromatograph was equipped with a 60 m 0.25 mm I.D. DB-5 column with 0.25 mm film thickness (J&W Scientific Folsom, CA, USA), fitted with a 2-m deactivated fusedsilica capillary pre-column. Injections were 1-ml cold oncolumn, using helium as carrier gas at a linear velocity of 30.8 cm s1 at 70 C. The column temperature program was: 35 C (held for 5 min) at 4 C min1 to 280 C (held for 10 min). The mass spectrometer conditions were: electron impact ionisation at 70 eV, acquisition mode: scan in the range of 35–350 m/z at 1 s/scan, ion source temperature of 200 C and interface temperature of 250 C. 2.5. Quantitation 1-chloroalkanes (C6, C10, C12, C16) were used as internal standards. The wastewater samples were spiked with internal standards at an individual concentration of 40 ng ml1 on the final extract. Two other internal standards (1-chloroalkanes C8 and C14) were added during the extraction procedure, for recovery control. Individual recovery factors were determined for each analyte, by CLSA-processing solutions of known con-
3.1. Influent VOC characterisation Table 3 shows the results of the influent characterisation. The compounds listed in key EU or US regulations are marked with X. The samples were qualitatively rich in target compounds. Only 6 of 47 targets were not detectable in any sample, while 35 analytes were detected in all experimental conditions. High concentration variability was observed, as indicated by the standard deviations and the 95% percentile. The flow-weighted mean concentration and mass flow of analytes at the plant influent were calculated. The total analyte weighted concentration at the plant influent was 232 mg l1, with a mean mass flow of 6380 g d1. A list of 15 VOCs accounted for 201 mg l1 or 87% of the total weighted concentration. 5 target compounds can be classified as carcinogen, mutagenic or toxic to reproduction (CTM) under article 5, paragraph 6 of the 1999/13/EC directive [20]. These compounds accounted for 926 g d1 or 14.5% of the total mass flow of analytes at the plant influent. The 1999/13/EC directive does not affect emissions from sewage treatment plants, and there are no specific EU regulations for air emissions from wastewater treatment plants. The US regulations can be used as a reference. 17 hazardous air pollutants (HAPs), as defined by EPA [16] were included in the target compound list. 16 of them were detected. They accounted for 2240 g d1 or 35% of the total mass flow of analytes at the plant influent. An annual extrapolation of these data yields 817 kg yr1 of HAPs. When compared to the US regulations, the mass flow of HAP analytes at the plant influent is much lower than the air emission limits set in the POTW rule [4] for new or reconstructed POTWs. These limits are 10 ton yr–1 (9072 kg yr–1) of a single HAP or 25 ton yr1 (22,680 kg yr–1) of the sum of 188 HAPs. Though a limited number of HAPs have been analysed in this study, the list of target compounds includes the most representative HAPs present in municipal wastewater, according to previous studies [1]. Besides, only a fraction of the influent VOC flow is emitted to the atmosphere
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920
3916
Table 3 VOC characterisation of the plant influent Compound
Mean (mg 11)
s (mg 11)
Samples with c> MDL
95% percentile
Priority EUa
Priority USAb
CMTc
HAPd
Chloroform 1,1,1-Trichloroethane Benzene Carbon tetrachloride Trichloroethylene Dimethyl disulphide Toluene Perchloroethylene Chlorobenzene Ethylbenzene m-Xylene and p-xylenee Styrene o-Xylene Cumene Bromobenzene a-Pinene Camphene o-Chlorotoluene n-Propylbenzene p-Chlorotoluene 1,3,5-Trimethylbenzene Dimethyl trisulphide tert-Butylbenzene 1,2,4-Trimethylbenzene Decane m-Dichlorobenzene p-Dichlorobenzene sec-Butylbenzene 1,2,3-Trimethylbenzene p-Cymene D-Limonene 1,8-Cineol o-Dichlorobenzene n-Butylbenzene Fenchone Linalool 1,3,5-Trichlorobenzene Camphor 1,2,4-Trichlorobenzene Menthol Naphtalene a-Terpineol 1,2,3-Trichlorobenzene Hexachloro-l,3-butadiene Dimethyl tetrasulphide Citronellol Bornyl acetate
9.5 1.3 7.6 1.0 6.1 1.6 23 1.4 0.25 1.7 6.6 0.27 2.2 0.11 oMDL 0.74 0.038 oMDL 0.9 oMDL 1.3 3.8 oMDL 14 49 oMDL 0.17 0.083 1.4 0.63 4.0 2.8 0.34 0.010 0.10 9.2 1.8 13.2 0.40 5.4 4.1 1.9 0.10 oMDL 3.9 3.5 2.4
7.6 2.0 9.1 1.4 15 2.0 24 1.5 0.19 2.7 8.1 0.38 2.4 0.27 — 0.86 0.049 — 2.8 — 3.6 5.4 — 43 96 — 0.13 0.30 2.7 1.2 3.8 2.2 1.2 0.039 0.14 6.3 6.6 11.4 1.6 4.2 3.4 1.5 0.12 — 6.0 3.2 2.4
15 15 16 16 16 16 16 16 16 16 16 16 16 14 0 16 9 0 16 0 16 15 0 16 15 0 14 5 16 14 16 15 5 1 7 16 15 15 6 16 15 16 12 0 7 12 15
0.8024 0.042–6.3 0.23–29 0.18–4.4 0.19–45 0.16–6.0 4.0–77 0.25–4.7 0.083–0.68 0.084–8.5 1.2–27 0.065–1.2 0.51–8.1 oMDL–0.78 — 0.12–2.9 oMDL–0.15 — 0.046–7.5 — 0.086–10 0.038–15 — 0.72–113 0–278 — oMDL–0.43 oMDL–0.76 0.075–8.3 oMDL–3.7 0.32–11.4 0.14–6.6 oMDL–3.2 oMDL–0.10 oMDL–0.36 0.76–20 0.011–16.7 0.056–36 oMDL–4.0 0.009–13 0.48–11 0.13–4.9 oMDL–0.38 — oMDL–16 oMDL–8.9 0.060–6.4
MDL: Method detection limit. a Priority substance in water, under the 2455/2001/EC decision or the list I of the 76/464/EEC directive. b Priority toxic pollutant [3]. c Carcinogen, mutagen or toxic to reproduction under the 1999/13/EC directive [20]. d Hazardous air pollutant in the USA [16]. e m- and p-xylene were coeluted.
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920 Table 4 Summary of weighted data for chemical groups at the influent (A) Group
Decane Chlorinated aliphatic hydrocarbons Light aromatic hydrocarbons Chlorinated aromatic hydrocarbons Sulphur compounds Terpenic compounds Totals
Mean weighted concentration (mg l1)
Mean weighted mass flow (g d1)
% of total
51 28
1414 761
22 12
90
2482
39
70
1
7
189
3
53
1467
23
232
6383
100
2.5
(the rest is biodegraded, sorbed or discharged). Therefore, the total HAP emissions would probably fall well below the limits set in the US POTW rule. The analytes were grouped by chemical constitution in six groups, and the aggregate weighted mean concentration of each grouped was computed (Table 4). Light aromatic hydrocarbons (39%), terpenic compounds (23%) and decane (22%) were the most abundant chemical groups of analytes at the plant influent. 3.2. Effects of time and day at the plant effluent The effect of the time of the day (H effect) and the day of the week (D effect) were computed as the differences between the mean concentrations obtained at the + and values of the independent variables [21]. Many effects were found close to zero at both the negative and the positive sides. Therefore, their statistical significance was evaluated. Effects are differences between means, and their statistical significance can be evaluated through t tests for difference between means. A t test was run for each analyte and effect, at a significance level of 0.05. A previous F test was run for each compound and effect, in order to determine if the variances were equal in the two populations being compared, and to determine the kind of t test to be applied (t test for equal or different variances). F and t tests were applied following the procedure described elsewhere [22]. The H effect was positive for 21 of 39 VOCs analysed, at a 0.05 significance level. The H effect was not significant for the remaining 18 VOCs, at 0.05 significance level. The sum of the significant positive effects
3917
was +146 mg l1. This means that the aggregated concentration of these 21 analytes was 146 mg l1 higher during the time bands of high DOC (187 mg l1) than during the time bands of low DOC (41 mg l1). This represents a 350% increase over the low DOC time band. The D effect was positive for 22 VOCs and negative for 17. A significance study through t tests led to the conclusion that the D effect was not significant for 34 of 39 target compounds, at 0.05 significance level. This means that for most VOCs, the concentration is not significantly higher during the weekdays than during the weekend, at the plant influent. This is not the case for the DOC, which was higher during the weekdays. However, the difference between DOC on weekdays and weekends was only 21 mg l1, vs. 35 mg l1 between the hours of high and low DOC. Analysing COD data from a whole year, the mean weekend COD was 402 mg l1 while the mean weekday COD was 472 mg l1. Though these values are significantly different at 0.05 significance level, the general organic load during the weekdays is not very different from the organic load during the weekend. Then the D effect results are not much different of those of the general organic load when expressed as COD. The HD interaction was not significant for 37 of 39 target compounds, at 0.05 significance level. In conclusion, for a vast majority of target compounds, neither D or HD were significant in determining the COV concentration at the plant influent. For a slight majority of VOCs (21/39), the H effect was significant and positive, this meaning that these compounds present higher concentrations at the hours of higher DOC. However, an important fraction of target compounds (18/39) were not sensible to the H effect, and did not present higher concentrations during the hours of higher DOC. The assumption of same VOC and DOC regime at the plant influent was correct for a majority of analytes, in the case of the day/night regime (H effect). It was not correct for the weekday/weekend regime (D effect), since most analytes were not affected by this factor. In this case, even DOC or COD presented smaller differences between the weekdays and the weekend.
3.3. VOC potential sources and day/night regime at the plant influent A bibliographic study was carried out on the sources of the 39 analytes included in the H effect analysis. 11 of 39 compounds were associated to petroleum solvents, applied in industrial degreasing and vehicle washing [23]. This group includes decane and all aromatic hydrocarbon analytes, except benzene. 10 of 11 petroleum solvent compounds presented significant positive H
ARTICLE IN PRESS 3918
A. Escalas et al. / Water Research 37 (2003) 3913–3920
effects, in accordance with the diurnal activity of most industries in Manresa. The terpenic compounds can be associated to aroma compounds included in powder detergents, liquid detergents and cleaners, and fabric conditioners. Powder detergents contain 0.6–1.2 mg/kg of fragrance additives, 25% of them being terpenic compounds, while fabric conditioners may present a higher content [24]. 9 of 11 terpenic compounds presented positive H effects, though only 5 were significant at 0.05 significance. No terpenic compound presented a significant negative H effect. These results fit reasonably well with the day/night cycle of activity, both domestic, commercial and industrial in Manresa. 6 analytes are chlorinated solvents from diverse origin: water disinfection (chloroform), dry cleaning (perchloroethylene), industrial degreasing (trichloroethylene and 1,1,1-trichloroethane) and other uses (chlorobenzene and o-dichlorobenzene). The H effect was not significant for chloroform, as expected from the continuous nature of tap water chlorination. Trichloroethylene presented a high, statistically significant positive H effect, what is in accordance with its use as industrial solvent. The other chlorinated solvents presented lower concentrations and no significant H effects. Dimethyl di-, tri- and tetrasulphide are biogenic odor compounds, related to the biodegradation of organic matter. All presented significant positive H effects, in accordance with the higher organic matter concentration in wastewater during daytime. Styrene is a cross-linking agent and sec-butylbenzene a plasticiser in the polymer industry. Both presented significant positive H effects. pdichlorobenzene is used as antimoth and deodoriser in toilets. Given the contact with water in the last application, this seems the most probable main source for p-dichlorobenzene in wastewater. This compound did not present a significant H effect at 0.05 confidence level. A main source could not be assigned to the other analytes, and they did not have significant H effects. 3.4. Evolution along the wastewater treatment train For this study, the analytes being detected at all points A, B, and C were only considered (34 analytes). Table 5 displays the arithmetic means of VOC concentrations for the 34 analytes at each sampling point. In order to study the evolution of VOC concentration along the wastewater treatment train, the means of analyte concentrations at points B, C, and D were compared with those of point A, through t tests for mean comparison. The total analyte concentration increased from the plant influent (187 mg l1) to the pre-treatment effluent (215 mg l1). 18 of 34 analytes increased their concentration between A and B, while the remaining 16 analyte concentrations decreased. A decrease should be ex-
Table 5 Mean concentrations of analytes at each sampling point Concentration (mg 11) A
B
C
D
Chloroform 1,1,1-Trichloroethane Benzene Carbon tetrachloride Trichloroethylene Dimethyl disulphide Toluene Perchloroethylene Chlorobenzene Ethylbenzene m-Xylene and p-xylene Styrene o-Xylene Cumene a-Pinene n-Propylbenzene 1,3,5-Trimethylbenzene Dimethyl trisulphide 1,2,4-Trimethylbenzene Decane p-Dichlorobenzene 1,2,3-Trimethylbenzene p-Cymene D-Limonene 1,8-Cineol Linalool 1,3,5-Trichlorobenzene Camphor Menthol Naphtalene a-Terpineol Dimethyl tetrasulphide Citronellol Bornyl acetate
9.5 1.3 7.6 1 6.1 1.6 23 1.4 0.25 1.7 6.6 0.27 2.2 0.11 0.74 0.92 1.3 3.8 14 49 0.17 1.4 0.63 4 2.8 9.2 1.8 13 5.4 4.1 1.9 3.9 3.5 2.4
4.4 1.3 5.8 0.59 1.4 8.4 19 1.5 0.31 2.9 13 0.086 5.3 0.15 0.64 1.4 2.7 12 35 35 0.13 0.62 0.22 1.5 6.6 9.8 0.14 12 7.7 5.8 2.2 9.5 6.3 1.2
5.5 0.28 0.61 0.83 2.2 2.8 11 2.3 0.11 1.3 5.9 0.68 2.4 0.11 0.56 0.49 0.73 6 7 17 0.19 0.73 0.36 1.8 2.1 6.4 0.2 6.5 3.7 7.5 0.98 5.2 1.8 0.94
2.9 0.26 1.3 0.26 0.56 0.48 7.3 0.2 0.12 0.36 1.1 0.059 0.38 0.032 0.098 0.088 0.097 0.94 1.2 oMDL 0.052 0.08 0.048 0.086 0.19 0.31 0.026 1.4 0.44 0.21 0.032 0.3 oMDL 0.096
Totals
187
215
106
21
pected, since 20–30% of influent VOCs are emitted through pre-treatment and primary treatment [1], and a maximum 10% is emitted from the primary clarifier [25]. In a t test at 0.05 significance level, 29 of 34 analytes presented equal mean concentrations. Chloroform (cA>cB) and—with cB>cA—dimethyl disulphide, oxylene, dimethyl trisulphide, and 1,8-cineol, presented significantly different concentrations. Since the supernatant of the digested sludge is recycled to the pretreatment, the increase in the concentration of sulphur compounds could be attributed to their biogenic generation in the anaerobic digesters. Besides, a higher number of samples would be necessary at the point B, given the high concentration variance. A comparison between the influent (A) and the primary effluent (C) led to results similar to those of
ARTICLE IN PRESS A. Escalas et al. / Water Research 37 (2003) 3913–3920
3919
100 90
Mean concentration, mg l
-1
80 70 60 50 40 30
Aromatic HC Decane Terpenic compounds Chlorinated aliphatic HC Sulphur compounds
20 10 0
A
B
Chlorinated aromatic HC C
D HC: hydrocarbons
Fig. 1. Evolution of analyte concentrations, grouped by their chemical constitution, along the sewage treatment plant.
the A–B comparison, in spite of the considerable difference between A and C total analyte concentration. No significant difference was observed for any analyte between points A and C, at 0.05 significance level. Again, the VOC losses through the pre-treatment and the primary treatment should be noticeable. However, given the high variances of the analyte concentrations, a higher number of samples would be necessary at C for detecting the differences. Finally, all analyte concentrations were lower at the plant effluent (D) than at the plant influent (A), at a 0.05 significance level, as expected. The overall percent removal was 89%. For the 34 analytes considered, a daily discharge through the plant effluent was estimated as 681 g d1, distributed as follows: decane (o0.5 g d1), chlorinated aliphatic hydrocarbons (125 g d1), light aromatic hydrocarbons (391 g d1), halogenated aromatic hydrocarbons (6.1 g d1), sulphur compounds (78 g d1), and terpenic compounds (81 g d1). Fig. 1 represents the mean concentration of these groups of analytes at the four sampling points in the plant.
4. Conclusions 1. 47 regulated and non regulated volatile organic compounds (VOCs) were characterised by closedloop stripping analysis (CLSA) and high resolution gas chromatography coupled to mass spectrometry (HRGC/MS) in 28 aqueous samples from 4 sampling points along a sewage treatment plant.
2. The sum of the weighted mean concentration of analytes was 232 mg l1 at the plant influent. The mass flow of analytes at the influent was 6383 g d1. The overall VOC removal along the wastewater treatment train was 89%. 3. At the plant influent, a majority of analytes (21 of 39) followed the general organic load regime, based on DOC. This set of compounds presented an overall 350% increase in concentration during high DOC time bands vs. low DOC time bands (H effect). However, 18 analytes were not significantly affected by the time of the day. No analyte presented a trend significantly opposite to the general organic load regime. 4. Petroleum solvents, and terpenic compounds from surfactant-containing preparations accounted for 79% of the analyte concentration at the plant influent. Most of these compounds followed the general organic load regime, characterised by a mostly diurnal activity in the city. 5. The day of the week (D effect: weekdays vs. weekend) did not significantly affected the majority of analytes (34 of 39) at the plant influent. The HD interaction was not significant for 37 of 39 analytes.
Acknowledgements We appreciate technical support from Ms. Anna M. ! from Aigues Lupon, . de Manresa SA. We also thank
ARTICLE IN PRESS 3920
A. Escalas et al. / Water Research 37 (2003) 3913–3920
Mr. Ricard Tomas, from the same company, for access to the plant. [13]
References [1] WEF-ASCE. Toxic Air Emissions from Wastewater Treatment Facilities. Alexandria VA, USA: Water Environment Federation-American Society of Civil Engineers; 1995. [2] Decision 2455/2001/EC of the European Parliament and of the Council of 20 November 2001 establishing the list of priority substances in the field of water policy and amending Directive 2000/60/EC. Off J Eur Communities L 331; 15/12/2001: 1–5. [3] EPA. Toxics criteria for those states not complying with Clean Water Act section 303(c)(2)(B). US Environmental Protection Agency: 40 CFR 131.36. [4] EPA. National emission standards for hazardous air pollutants: publicly owned treatment works. US Environmental Protection Agency. Federal Register 1999, 64, 206, 57572. [5] Govind R, Lai L, Dobbs R. Integrated model for predicting the fate of organics in wastewater treatment plants. Environ Prog 1991;10(1):13–23. [6] EPA. Air emissions models for waste and wastewater. Document EPA-453/R-94-080. Research Triangle Park, NC, USA: EPA Office of Air Quality Planning and Standards; 1994. [7] Melcer H, Bell JP, Thompson DJ, Yendt CM, Kemp J, Steel P. Modeling volatile organic contaminants in wastewater treatment plants. J Environ Eng 1994;120(3): 588–609. [8] Corsi RL, Card TR. Estimation of VOC emissions using the Baste model. Environ Prog 1991;10(4):290–9. [9] Lee KC, Rittmann BE, Shi J, McAvoy D. Advanced steady-state model for the fate of hydrophobic and volatile compounds in activated sludge. Water Environ Res 1998;70(6):1118–31. [10] Chern JM, Chou SR. Volatile organic compound emission rates from mechanical surface aerators: mass-transfer modelling. Ind Eng Chem Res 1999;38(8):3176–85. [11] Tata P. Methodology for the determination of VOC emissions. Prepared for inclusion in AMSA MACT Guidance Document. Metropolitan Water Reclamation District of Greater Chicago, 1999. [12] EPA. WATER9 [on line]. EPA Office of Air Quality Planning and Standards. URL: http://www.epa.gov/ttn/
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21] [22]
[23]
[24]
[25]
chief/software/water/index.html. [Consulted on 3 April 2003]. EPA. EPA Methods and guidance for analysis of water [CD-ROM]. Washington DC: EPA Office of Water; 1999. Malleret L, Bruchet A, Hennion MC. Picogram determination of ‘‘Earthy-Musty’’ odorous compounds in water using modified closed loop stripping analysis and large volume injection GC/MS. Anal Chem 2001;73(7): 1485–90. APHA-AWWA-WPCF (1995). Standard methods for the examination of water and wastewater. Washington, DC: American Public Health Association—American Water Works Association—Water Pollution Control Federation; 1995. EPA. EPA—Air Toxics Website original list of hazardous air pollutants [on line]. URL: http://www.epa.gov/ttn/atw/ orig189.html. Consulted on 8 January 2003. Dunovant VS, Clark CS, Que Hee SS, Hertzberg VS, Trapp JH. Volatile organics in the wastewater and airspaces of three wastewater treatment plants. J Water Pollut Control Fed 1986;58(9):886–95. Simonich SL, Begley WM, Debaere G, Eckhoff WS. Trace analysis of fragrance materials in wastewater and treated wastewater. Environ Sci Technol 2000;34(6):959–65. Escalas A, Droguet M, Guadayol JM, Caixach J. Estimating DOC regime in a wastewater treatment plant by UV deconvolution. Water Res 2003;37(11): 2627–35. Council Directive 1999/13/EC of 11 March 1999 on the limitation of emissions of volatile organic compounds due to the use of organic solvents in certain activities and installations. Off J Eur Communities L 085; 29/03/1999: 1–22. Montgomery DC. Design and analysis of experiments, 4th ed. New York: Wiley; 1997. p. 228. Montgomery DC, Runger GC. Applied statistics and probability for engineers, 1st ed. Mexico D.F., Mexico: McGraw-Hill; 1996. p. 396–399, 410–412 and 431–432 [in Spanish]. Pax!eus N. Vehicle washing as a source of organic pollutants in municipal wastewater. Water Sci Technol 1996;33(6):1–8. Gavald"a A. Personal communication from Albert Gavald"a, aroma expert from Aromas Maluquer SA, Saragossa 91, 08912 Badalona, Spain, 2000. Pincince DR. Transfer of oxygen and emissions of volatile organic compounds at clarifier weirs. J Water Pollut Control Fed 1995;63(2):77–119.