a study on volatile organic sulfide causes of odors at philadelphia's northeast water pollution...

ARTICLE IN PRESS

0043-1354/$ - se

doi:10.1016/j.w

�Correspond

fax: +1215 743

E-mail addr

Water Research 39 (2005) 3781–3790

www.elsevier.com/locate/watres

A study on volatile organic sulfide causes of odors atPhiladelphia’s Northeast Water Pollution Control Plant

Xianhao Cheng�, Earl Peterkin, Gary A. Burlingame

Bureau of Laboratory Services, Philadelphia Water Department, 1500 E. Hunting Park Ave., Philadelphia, PA 19124, USA

Received 5 August 2004; received in revised form 16 May 2005; accepted 5 July 2005

Available online 19 August 2005

Abstract

Volatile organic sulfide (VOS) causes of odors were studied at Philadelphia’s Northeast Water Pollution Control

Plant between September 11 and November 25, 2003. Results showed that dimethyl sulfide (DMS) dominated the VOS

pool whenever VOS concentration rose above the background level (o50 mg/L). Methanethiol was generally less than

10% of VOS and it was mainly found at sites with limited or reduced dissolved oxygen (DO). Dimethyl disulfide

occupied �1% of the VOS and was often not detectable. Carbon disulfide was not detected. The concentrations of

DMS varied by three orders of magnitude, ranging from o5 to 1260mg/L. High concentrations of DMS, averaging 419

and ranging from o5 to 1000 mg/L, were generally found in return activated sludge. The DMS concentration in the

primary effluent ranged from o5 to 729 mg/L and averaged 245mg/L. Concentrations of DMS in the aeration tank (AT)

with high DO were from o5 to 997 mg/L with an average of 250 mg/L. However, DMS concentrations in the AT

representing anaerobic conditions were as high as 1260 mg/L. The estimated average purge efficiency of DMS was about

78%, which required a DMS production rate of 108mg/L h�1 to keep the analyzed concentration in the AT. While a

valid but weak statistical relationship between acetone and DMS was observed, there was no strong evidence to support

that the methylation of hydrogen sulfide with acetone could be the mechanism for the DMS formation. Instead, DMS

production was found to be associated with the characteristics of incoming wastewater. Thus, a VOS precursor was

believed to be present in the incoming wastewater, which warranted a need for further investigation.

Published by Elsevier Ltd.

Keywords: DMS; VOS odor; Wastewater

1. Introduction

Under the Title V Operating Permit of the Clean Air

Act of the Environmental Protection Agency in the

United States, odor emissions from wastewater treat-

ment plants have become an important issue to the

wastewater industry (WEF, 1995; Stuetz and Franz-

Bernd, 2001). The primary concern a wastewater plant

e front matter Published by Elsevier Ltd.

atres.2005.07.009

ing author. Tel.: +1215 685 1436;

5594.

ess: [email protected] (X. Cheng).

may face is the irritating nature of the odors, and the

controversy generated by the surrounding neighbor-

hood, as they may perceive the nuisance as something

harmful. Philadelphia’s Northeast Water Pollution

Control Plant (NEWPCP) has a history of odor

nuisance problems largely because the community

resides along one fence line. NEWPCP is the second

largest wastewater treatment plant in Pennsylvania,

treating a wide range of municipal and industry wastes

at an average rate of 210 MGD. Since the late 1980s,

managing and controlling odors have been a mission of

NEWPCP. However, monitoring odor source samples

www.elsevier.com/locate/watres

ARTICLE IN PRESSX. Cheng et al. / Water Research 39 (2005) 3781–37903782

and identifying the odor-causing mechanisms were

proving to be a challenge.

PWD’s odor assessment team, which included CH2M

HILL enginery consultants, identified the primary

nuisance odor as a ‘‘canned corn’’ quality odor coming

from the secondary biological process and causing

complaints from the adjacent community. Previous

studies showed that it was most likely caused by

dimethyl sulfide (DMS) but may also include dimethyl

disulfide (DMDS) and methanethiol (MT) (Burlingame

and Mosley, 1995; Burlingame et al., 2004). However,

confirmation was needed by actual chemical analysis

and mechanisms of the odor formation remained in need

of investigation. As the request, the Organics Labora-

tory in the Bureau of Laboratory Services (BLS) of

PWD established an analytical method, which ensured

an accurate and convenient analysis of volatile organic

sulfides (VOS) in wastewater together with other volatile

organic compounds (VOC) (Cheng et al., 2005).

While other processes generate nuisance odors at

NEWPCP, the aeration tank odor was both the most

persistent and troublesome, and least understood. Thus,

the present study was mainly focused on odors produced

in the aeration tank (AT). This study examined the

occurrence of odor-related VOC, evaluated the possible

mechanisms for VOS odor production, and determined

a way to predict odor generation.

2. Site description and sample handling

The NEWPCP is a secondary wastewater treatment

facility with two sets of primary sedimentation tanks

(PST). The PST sludge blanket is kept below 2 ft and is

pumped from each tank sequentially approximately

once or twice a day. Wastewater, after PST, is delivered

to AT for secondary biological treatment. The AT is a

Set 2

25%PE

25% PE

A1 A2 A3 A4 A5 A

B1 B2 B3 B4 B5 B

C1 C2 C3 C4 C5 C

D1 D2 D3 D4 D5 D

Pre-Selector zone

Fig. 1. Aeration basin s

seven tank, four-pass per tank, step-feed process

(Fig. 1). It is operated in sludge re-aeration. The return

activated sludge (RAS) enters each of the seven tanks at

the upstream end of pass A(10). Fifty percent of the

primary effluent (PE) enters at the upstream end of pass

B, and 50% of PE enters at the upstream end of pass C

(Fig. 1). Aeration basin selector zones are designated to

be anoxic zones in the AT to allow for the selective

growth of faculative bacteria to assist in the wastewater

treatment process. Here, pre-selector zones (A1, A2, B9

and B10) have reduced aeration and selector zones (B1,

B2, C9 and C10) have no aeration. The dissolved oxygen

(DO) in AT is 1–2mg/L from A10 to A3, 0.5mg/L in

A1, A2, B9 and B10, 1–4mg/L from B8 to B3, 0–0.5mg/

L in B1, B2, C9 and C10, 1–3mg/L from C8 to C1 and

5–8mg/L from D1 to D10. Based on the designed flow

of 210 MGD, the detention time is �2.1 h in each PST.

The detention time of the wastewater in AT is �1.7 and

�3.8 h in the RAS.

The sample locations were designed to reflect what

was occurring in the AT but also involve PE and RAS as

references. Samples were collected at the locations that

represent each treatment process in AT in order of

process flow. Grab samples were collected in 250 mL

amber screw cap glass bottles with no headspace. The

samples were placed immediately in a cooler with ice,

chilled to 472 1C and transported to the lab. No sample

preservatives were used to adjust the pH or affect the

chemical nature of the samples. Samples were stored in

the refrigerator (470.5 1C) overnight to allow solids to

settle. The assumption was made that VOS in waste-

water was best represented by the liquid phase.

Depending on the concentration of individual samples,

different volumes of supernatant liquid were pipetted

into volumetric flasks and diluted with acidified (HCl)

deionized water. The final pH value was adjusted to

�1.4 with HCl.

RAS

Set 1 25%PE

25%PE

Selector Zone

6 A7 A8 A9 A10

6 B7 B8 B9 B10

6 C7 C8 C9 C10

6 D7 D8 D9 D10

ampling locations.

ARTICLE IN PRESSX. Cheng et al. / Water Research 39 (2005) 3781–3790 3783

3. Materials and methods

Instrumentation, chemicals, and analytical methods

used for the analysis of VOS and VOC were described by

Cheng et al. (2005). Briefly, instruments used in the

analysis consisted of a gas chromatograph (Agilent

6890A), mass spectrometer (Agilent 5973N), purge and

trap concentrator (O.I. Analytical 4560), and an

autosampler (Tekmar-Dohrmann SOLATEK 72). A

capillary column (60m� 250 mm� 1.40mm, J&W DB-

VRX) was used to separate the VOS and VOC

compounds. Calibration standards were obtained from

Protocol Analytical (472 Lincoln Boulevard, Middlesex,

NJ, USA) with a maximum uncertainty of 75%.

Internal standards and surrogates were purchased from

UltraScientific (250 Smith St., N. Kingstown, RI, USA).

Methanol (purge and trap grade) and HCl (certified

A.C.S. plus) were purchased from Fisher Scientific

(2000Park Lane, Pittsburgh, PA, USA). Detection

limits for VOS and VOC compounds are 5mg/L, except

for acetone, which is 13mg/L. Dissolved hydrogen

sulfide concentration (H2S) was analyzed by the Iodo-

metric Method (America Public Health Association,

1998), with a detection limit of 1mg/L. Because of the

limitation in sample management, all the values in this

study were from single analysis. Based on our previous

study, analysis of VOS in the liquid phase is quite

consistent with a coefficient of variance less than 4.1%

(Cheng et al., 2005).

4. Results

Results of the analyses are shown in Table 1. For

convenience, the results are arranged by the source and

through the treatment processes. All the calculations in

this study are based on Table 1. If the concentration of a

compound was less than the detection limit, then the

calculation of an average was based on the detection

limit rather than zero.

4.1. Return activated sludge (RAS)

The average concentration of DMS was 419mg/L,

ranging from o5 to 999 mg/L. MT ranged from o5 to

66mg/L, averaging 25mg/L. DMDS was only detected in

the September 16 sample at a concentration of 7mg/L,

which was about 1% of DMS in the same sample.

Carbon disulfide was not detected in any of the samples

analyzed and, therefore, will not be discussed further.

Concentrations of H2S ranged from 1.08 to 2.16 mg/L.

VOC compounds were not found to be more than 25 mg/

L except for three methyl group-containing compounds,

acetone (Act.), cumene (Cum.), and styrene (Styr.).

Acetone concentrations were normally in less than

200 mg/L to non-detectable levels in RAS. High con-

centrations of acetone were found only in the November

18 sample at 3220mg/L, while cumene and styrene were

less than 7 mg/L in RAS samples.

4.2. Primary effluent (PE)

Concentrations of DMS in PE ranged from 5 to 729

(avg. 245 mg/L). Relatively low concentrations of MT

were also observed, which ranged from o5 to 45 (avg.

14mg/L). DMDS was not detectable, except on Septem-

ber 16, which had a concentration of 8 mg/L. Concen-

trations of H2S in PE ranged from o1 to 1.50 (avg.

1.13mg/L). Concentrations of methyl group-donating

compounds, however, were highest in PE samples.

Concentrations of acetone ranged from o12.5 to

22,500 (avg. 4302 mg/L). Concentrations of cumene

ranged from o5 to 59. Styrene was also found in most

PE samples, which ranged from o5 to 73mg/L.

4.3. Aeration tank locations (A2 and B2) with near

anaerobic conditions (DO�0 to 0:5 mg/L)

A2 is a location upstream of PE introduction and B2 a

location right after the PE introduction (Fig. 1).

Concentrations of DMS were highest in A2 samples,

which ranged from o5 to 1260 and averaged 386 mg/L.

Concentrations of DMS at B2 ranged from o5 to 770

and averaged 257mg/L. Concentrations of MT were also

relatively higher at A2 and ranged from o5 to 340 and

averaged 15 mg/L. Concentrations of MT ranged from

o5 to 42 and averaged 18mg/L at B2. DMDS

concentrations were relatively higher compared to other

sites, which ranged from o5 to 41 mg/L. At B2, however,

DMDS was only detected on September 16 (Table 1).

H2S concentrations were lower in A2 samples, although

relatively higher concentrations could be found on

certain days, which ranged from o1 to 3.29mg/L.

Concentrations of H2S were consistently higher in B2

samples, ranging from o1 to 2.22 with an average of

1.53mg/L. Concentrations of the methyl group-contain-

ing compounds were relatively lower in A2 samples.

Acetone was detected only on 2 days, while cumene was

close to the detection limit. Styrene was also not

detected in those samples (Table 1). The methyl group-

containing compounds were most detectable in B2

samples. Concentrations of acetone ranged from

o12.5 to 20,600 and averaged 2812mg/L. Concentra-

tions of cumene ranged from o5 to 59 and averaged

16mg/L. Styrene was still not detected in most of the

samples.

4.4. Aeration tank locations with DO higher than 1 mg/L

As exhibited in Fig. 1, A8, B8, C8, C2, and D10 were

well aerated. Concentrations of DMS were lower in A8

samples, ranging from o5 to 649 with an average of

ARTICLE IN PRESS

Table 1

Wastewater concentrations of volatiles at NEWPCP (mg/L)

Locat. Comp. 9/16 10/7 10/14 10/21 10/28 11/4 11/12 11/18 11/25

RAS DMS 691 o5 709 817 216 196 128 999 12

MT 66 7 7 15 16 3 46 65 4

DMDS 7 o5 o5 o5 o5 o5 o5 o5 o5

CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa 1.48 1.08 1.44 1.52 2.04 2.16 2.24 1.84 2.34

Act. o12.5 90 o12.5 o12.5 186 105 o12.5 3220 o12.5

Cum. o5 o5 o5 o5 o5 7 o5 o5 o5

Styr. o5 o5 o5 o5 o5 o5 o5 o5 o5

Primary tank (set 1) DMS 456 5 689 729 40 164 62 386 19

MT 39 13 o5 16 o5 8 21 9 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 o1 1.16 1.16 1.10 o1 1.14 1.20 o1

Act. 1960 1570 8890 663 3470 2760 308 22500 1470

Cum. 7 7 6 8 11 9 8 10 8

Styr. 9 7 7 9 28 24 10 58 26

Primary tank (set 2) DMS 287 o5 479 696 23 75 91 183 28

MT 35 o5 o5 18 o5 o5 45 10 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 1.12 1.5 1.5 o1 o1 1.5 o1 1.06

Act. 1460 1050 5110 575 2710 881 o12.5 21000 1070

Cum. 7 o5 6 8 10 8 o5 11 o5

Styr. 9 o5 o5 8 17 21 o5 73 19

Aeration tank site A8 DMS 409 o5 563 239 47 111 118 649 o5

MT 31 o5 o5 o5 o5 o5 27 27 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 o1 o1 o1 2.00 1.38 o1 2.26 1.96

Act. o12.5 o12.5 o12.5 o12.5 83 o12.5 o12.5 1950 o12.5

Cum. o5 o5 o5 o5 o5 o5 o5 o5 o5


Aeration tank site A2 DMS 236 o5 1260 1190 39 82 48 580 31

MT 34 o5 340 47 21 o5 13 256 26

DMDS 8 o5 41 24 o5 o5 o5 o5 o5

CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 o1 1.22 o1 3.29 o1 o1 1.56 1.92

Act. o12.5 o12.5 o12.5 o12.5 o12.5 o12.5 447 5910 o12.5

Cum. o5 o5 o5 6 o5 o5 o5 7 o5

Styr. o5 o5 o5 o5 o5 o5 o5 10 o5

Aeration tank site B2 DMS 363 o5 770 609 82 103 67 289 27

MT 33 o5 10 23 15 o5 25 42 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa 1.18 1.08 2.22 2.14 2.23 1.46 o1 1.14 1.30

Act. 568 367 2290 231 525 380 o12.5 20600 334

Cum. 7 59 26 12 10 7 o5 14 o5

Styr. o5 o5 o5 o5 7 9 o5 46 o5

Aeration tank site B8 DMS 287 o5 541 307 27 126 23 490 13

MT 31 o5 o5 18 o5 o5 5 77 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 1.4 1.46 1.55 1.78 1.82 1.44 1.38 1.34

X. Cheng et al. / Water Research 39 (2005) 3781–37903784

ARTICLE IN PRESS

Table 1 (continued )

Locat. Comp. 9/16 10/7 10/14 10/21 10/28 11/4 11/12 11/18 11/25

Act. 31 o12.5 592 o12.5 o12.5 175 o12.5 14000 290

Cum. o5 10 6 6 o5 8 o5 17 o5

Styr. o5 o5 o5 o5 10 8 o5 37 o5

Aeration tank site C8 DMS 413 o5 984 629 132 127 65 434 16

MT 33 13 11 16 o5 o5 14 60 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 o1 1.66 1.7 1.56 1.78 1.30 1.26 1.74

Act. 593 324 1980 143 1760 558 120 17000 525

Cum. 7 36 29 9 15 7 7 16 o5

Styr. o5 o5 o5 o5 10 8 o5 37 o5

Aeration tank site C2 DMS 468 o5 967 359 66 72 29 434 o5

MT 32 o5 o5 19 o5 o5 19 64 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 1.06 1.30 2.10 1.80 1.18 1.08 1.46 1.08

Act. 200 o12.5 1140 o12.5 1000 415 o12.5 12800 361

Cum. o5 o5 o5 o5 o5 7 o5 8 o5

Styr. o5 o5 o5 o5 o5 o5 o5 12 o5

Aeration tank site D10 DMS 366 o5 997 195 72 40 79 302 o5

MT 30 o5 o5 16 o5 o5 52 37 o5


CS2 o5 o5 o5 o5 o5 o5 o5 o5 o5

H2Sa o1 o1 o1 o1 1.54 1.88 o1 1.80 1.74

Act. o12.5 o12.5 o12.5 o12.5 588 27 o12.5 4040 110

Cum. o5 o5 o5 o5 o5 o5 o5 o5 o5


aDissolved inorganic sulfide in mg/L

X. Cheng et al. / Water Research 39 (2005) 3781–3790 3785

132 mg/L. DMS concentrations were higher through B8

to C2, ranging from o5 to 541, o5 to 984 and o5 to

967 mg/L, in B8, C8 and C2, respectively. Average

concentrations of DMS (202, 311, and 267 mg/L) were

found in these samples. MT concentrations were also

lower in A8 samples, ranging from o5 to 31 (avg. 13mg/L),

though not detected in most samples. Concentrations of

MT increased slightly from B8 to C2, ranging from o5

to 77, o5 to 60 and o5 to 64, with averages of 17, 18

and 18m/L, respectively. There was little difference in the

H2S concentrations between these sites, averaging

around 1.4mg/L. However, H2S concentrations were

less than the detection limit in most of the samples from

A8. Concentrations of methyl group-donating com-

pounds were less than their detection limit in most

samples from A8, except for the samples collected on

November 11 and 18 in which acetone concentrations

were comparable with other sites. Concentrations of

methyl group-donating compounds increased from A8

to C2, which agrees with the progressive introduction of

PE effluent to AT (Table 1). Concentrations of acetone

ranged from 143 to 17,000 and o12.5 to 12,800

and averaged 2556 and 1773 in C8 and C2 samples,

respectively. Concentrations of cumene were not

detectable in most samples from C2. However, cumene

was frequently found at C8, ranging from o5 to 36 and

averaged 15mg/L. Styrene was not detected in most cases

at these sites.

Wastewater from D10 was directed to the final

sedimentation tanks. Concentrations of DMS were

relatively lower at this site compared to other sites in

the aeration tanks, which ranged from o5 to 997 and

averaged 229 mg/L. MT could only be detected on

certain days and DMDS was not detected, except for a

September sample. H2S was lower than its detection

limit on most of the days studied. Acetone was not

detected or could only be detected at lower levels when

H2S was detectable. Cumene and styrene were not

detected.

5. Discussion

5.1. Occurrence summary of VOS and VOC

The concentration of DMS can be close to or higher

than 1000mg/L in the VOS pool at the NEWPCP. It was

the dominant compound when the VOS concentration


was higher than �50mg/L. DMS then account for more

than 90% of the analyzed VOS in the samples other than

that collected on October 7 and November 25. A

previous research found DMS concentration in the

influent of a wastewater treatment plant ranged between

3 and 27 mg/L (Hvitvet-Jacobsen, 2000), which agrees

with the DMS concentration range (o31mg/L) in the

samples collected on October 7 and November 25, in the

NEWPCP. Therefore, a concentration of about 50mg/L

can be practically considered as the background level in

the NEWPCP. DMDS was not detected in most samples

collected during the study period. Carbon disulfide was

not detected in any of the samples. A noticeable

phenomenon was that MT was mostly detected at the

sites with no or limited DO (RAS, A2 and B2). At other

sites, it was mostly undetected or just above the

detection level. This agrees with a laboratory incubation

study using freshwater sediment (Lomans et al., 1999),

in which no MT was detected in the slurries incubated

under air (with shaking). Although biogeochemical

conditions during a wastewater treatment process could

be more complicated than in a controlled laboratory

culture, it might indicate that MT was only a transient

form of VOS in the NEWPCP, which was generated in a

low oxygen environment. The only VOC compound

detected, with an odor of concern, was styrene.

26.5˚C

20˚C

21.6˚C21.6˚C

1

Dates o

9/16/03 10/7/03 10/14/03 10/21/03 10

DM

S C

once

ntra

tions

(µg

/L)

0

200

400

600

800

1000

1200

1400

PERASA2 B2 ATD10

Fig. 2. Distribution of DMS in d

However, the concentration of styrene was close to its

detection limit on most of the days studied. Therefore,

DMS was the only ‘‘canned corn type’’ odorous

compound analyzed that played the major role for

odor. Concentrations of DMS fluctuated significantly by

three orders of magnitude. It confirms a previous study

that DMS is the VOS compound most likely contribut-

ing to the AT nuisance ‘‘canned corn’’ odor (Porter

et al., 2004).

5.2. Evaluation of DMS odor source

Since the concentration of DMS accounted for more

than 90% of the VOS pool analyzed when VOS

concentration was above the background level, VOS

odors could be represented by DMS. The distribution of

DMS in different treatment units is exhibited in Fig. 2.

The average concentration of DMS in PE samples was

14.4% of the collective DMS analyzed through the

studied treatment processes. It was comparable to

the average concentration of DMS in the AT (14.1%).

The DMS concentration in RAS was the highest among

the treatment units, making up 25.1%. Concentrations

of DMS were relatively higher in AT’s pre-selector zones

(18.1%) with reduced air (A2) and selector zones

(16.1%) with no aeration (B2), compared to other sites

4.4˚C22.2˚C

14.4˚C

13.8˚C

7.2˚C

f Sampling

/28/03 11/4/03 11/12/03 11/18/03 11/25/03

ifferent treatment processes.


in the AT. The lowest DMS concentration was typically

observed at the down stream site of the AT (D10,

12.1%), where wastewater was exported directly to the

final sedimentation tank. As a whole, concentrations of

DMS were high in the process from pass A, which only

had RAS, and the mixing of PE with RAS, but then

decreased slightly during the aeration purge. The

consistent existence of DMS in the AT, with concentra-

tions close to or even higher than that in the PE samples,

demonstrated that DMS was produced within the

aeration tank at the same time when it was being

stripped out by the aeration process.

Production of DMS from this source could be

estimated by the distribution of acetone. As reflected

by the direction of the flow (Fig. 1), the concentration of

acetone decreased by an average of 71% from B2, where

the PE is introduced to the AT, to B8, where wastewater

has been aerated through B bay. From C8, where

wastewater is introduced from the PE to the AT, to C2

and D10, the concentration of acetone decreased by an

average of 85% (Table 1). A simple logic for this is that

if the AT is not the DMS source, DMS concentration

DMS in B Bay

100 1 2 3 4 5 6 7 8 9

100 1 2 3 4 5 6 7 8 9

DM

S C

once

ntra

tions

(µg

/L)

0

200

400

600

800

1000

DMS at B2

DMS at B8

Acetone in B Bay

Days of sampling

Ace

tone

Con

cent

ratio

ns (

µg/L

)

0

500

1000

1500

2000

2500

3000

14000

16000

18000

20000

22000

Acetone at B2 Acetone at B8

1

1

1

1

Fig. 3. Variation of acetone concen

will decrease more quickly than acetone due to purging,

because the boiling point of DMS (36.21) is lower than

acetone (56.51). We did not observe much decrease of

DMS concentration as expected from the distribution of

acetone (Fig. 3). Therefore, a major amount of DMS

must be produced in the AT. Assuming that the loss

through the aeration process was much higher than

biological and chemical processes, the DMS production

rate in the AT can be estimated from the loss of acetone.

According to the above calculation, an average of

78% of the acetone could be lost in the AT from the PE

input to the end of the AT (D10). Assuming a same rate

of loss for DMS, the concentration of DMS would be

�45 mg/L at D10 if there was no in-site production. The

residence time of flow from the PE to D10 was about

1.7 h. Subtracting the average concentration of DMS at

D10 (229 mg/L) by 45mg/L and dividing by 1.7, the

production rate came out to be 108 mg/Lh�1). Despite

this being a semi-quantitative estimation, it shows

the range of DMS production in the AT. DMS was

also subject to anaerobic and aerobic degradation

(Lomans et al., 1999), but this error might be minimized

100 1 2 3 4 5 6 7 8 9

100 1 2 3 4 5 6 7 8 9

DMS from C to D Bay

0

200

400

600

800

1000

1200

DMS at C8 DMS at C2 DMS at D10

Acetone from C to D Bay

Days of sampling

0500

1000150020002500300035004000

2000

4000

6000

8000Acetone at C8 Aceone at C2 Acetone at D10

trations in the aeration tank.

ARTICLE IN PRESS

0 20 40 60 80 100 120

0

2

4

6

8

10

12

14

16

Acetone Concentrations (µM)

DM

S C

once

ntra

tions

(µM

)

Fig. 4. Linear regression of DMS and decrease in acetone for

aeration tank.

X. Cheng et al. / Water Research 39 (2005) 3781–37903788

somewhat with the biological and chemical losses of

acetone. As a consequence, it could be predicted that

VOS production was strongly associated with the

characteristics of the incoming wastewater, resulting in

the in situ production of DMS upon the mixing of

incoming wastewater with RAS in AT.

5.3. An exploration of the VOS production mechanism in

NEWPCP

The mechanisms for the generation or production of

VOS has been studied in marine and fresh water

environments because of its impacts on global warming

and acid rain (Fitzgerald, 1991; Richards et al, 1994;

Lomans et al., 2001a, b). However, it is not well studied

in the wastewater environment. As far as it can be seen,

there have been no published studies on the formation of

VOS related with VOC in wastewater. Due to the local

industry-related specific source water and biogeochem-

ical environment, the mechanism of VOS production in

wastewater can be significantly different from that in

natural environments, and can vary in different waste-

water treatment plants. There are at least three possible

mechanisms for the production of VOS in aquatic

environments. Among them are: (1) methylation of H2S

(Drotar et al., 1987; Bak et al., 1992), (2) microbial

decomposition of sulfur-containing amino acids (Zinder

and Brock, 1978; Oho et al., 2000), and (3) chemical and

biological degradation of biochemical precursors (Kiene

and Capone, 1988; Ginzburg et al., 1998). In anaerobic

freshwater sediments, formation of MT and DMS has

been demonstrated to occur mainly by methylation of

sulfide (Ginzburg et al., 1998). The referenced study

showed a close relationship between hydrogen sulfide

and DMS/MT. Methoxylated aromatic compounds

were believed to be the major source of methyl groups

for the methylation of hydrogen sulfide. However, the

present study did not find a meaningful relationship

between hydrogen sulfide and DMS or MT at any of the

locations (Table 1). In addition, a bench study on

NEWPCP wastewater showed that the addition of

hydrogen sulfide to RAS did not increase the DMS

production, even though the H2S concentration mea-

sured in the headspace of the reactor was many times

greater than would have been observed in the treatment

process (Porter et al., 2004). This raised the question as

to whether the conventional mechanism of hydrogen

sulfide methylation was applicable to NEWPCP.

No significant contribution by methoxylated aromatic

compounds was found in analyzed volatiles. Instead, it

was found that acetone dominated the VOC pool. High

concentrations of acetone existed consistently in the

plant influents and it was reduced to its detection limit

or non-detectable at site D10 and in RAS samples in

most cases, except on November 18, 2003, when the

acetone concentration was extremely high (Table 1).

This demonstrated that acetone was imported from

influents, consumed in the aeration tanks, and exhausted

in the RAS (Fig. 3). It is believed that acetone can

provide its methyl group to hydrogen sulfide to generate

DMS in 1:1mole ratio. However, there was no simple

relationship between DMS and acetone in each flow

path of the AT (Fig. 3). A statistical analysis showed a

valid but weak relationship between the average

concentration of DMS in the AT and average loss of

acetone in concentrations from influents on different

days (Fig. 4), if the datum from November 18, 2003 was

excluded. However, the extremely high concentration of

acetone in the influents on November 18, 2003 did not

produce corresponding VOS. Even though concentra-

tions of hydrogen sulfide up to 1800 mg/L still existed in

the AT, the concentration of DMS was only

5027102mg/L. Other factors such as temperature may

play a role in affecting the demethylation process of

methyl-donating compounds (Lomans et al., 1997). In a

cited study (Lomans et al., 2001a, b), optimal growth

temperatures of demethylation bacteria were between 34

and 37 1C, and maximum demethylation activity,

measured as MT formed, was at 25 1C in controlled

cultures. This study did find temperature on November

18, 2003 to be relatively low and higher concentrations

of DMS were found mostly on the days with the air

temperature higher than 20 1C (Fig. 2). It was possible

that variation in temperature influenced the demethyla-

tion activity so as to influence the formation of VOS.

The statistical relationship between DMS and acetone

could neither reject nor accept the hypothesis that

methylation of sulfide was the mechanism of VOS

production in NEWPCP.

Sulfur-containing amino acids were not analyzed in

the wastewater samples in this study. However, mea-

surement by a research lab showed that the total

dissolved protein (TDP) in the PE influents and AT

ranged from 18 to 52mg/L on June 17, 2003 in


NEWPCP. Concentrations of TDP decreased down

stream. This showed that a maximum contribution of

dissolved protein was 34mg/L if all of the loss of TDP

contributed to the formation of VOS, during the aeration

process. The sulfur-containing amino acids, cysteine and

methionine, which can be precursors of DMS, are minor

compounds in the amino acid pool and are seldom found

in most natural environments. A study in sediments did

not find the accumulation of cysteine and methionine

during the degradation of sediments (Dauwe and

Middelburg, 1998). It might indicate that these sulfur-

containing organic compounds were not readily accumu-

lated in degraded materials like wastewater. The average

ratio of sulfur in the sulfur-containing amino acids is

�24%. Assuming the concentration of cysteine and

methionine is 2% of the TDP, the contribution of sulfide

from sulfur-containing amino acids could be �160mg/L.

This is much less than the dissolved sulfide analyzed in

the samples when considering the fact that production of

DMS is dynamic and the loss in the aeration process can

be much more than what was measured in liquid phase.

Although it is possible that an industry could have been

dumping waste rich in certain proteins, we do not have

such evidence in this area.

Studies on the chemical and biological degradation of

biochemical precursors have mainly focused on dimethyl

sulfoniopropionate (DMSP), an osmolyte of algae and

phytoplankton. Since a planktonic algae bloom has not

been found at NEWPCP, it is unlikely that this could be a

major mechanism for the VOS odor in NEWPCP. The

methylation of hydrogen sulfide was more likely a feasible

mechanism for the production of DMS at NEWPCP

though not as the major mechanism by the evidence

presented above. However, discharge from industries

could contain the substances that were not captured by

this study. A detail investigation on local manufactures

that discharge their organic sulfide-related wastewater to

NEWPCP is necessary before further study. Obviously,

there is still a need for a more concrete identification of

the source/mechanism. Management and remediation of

the odor problem will be more cost effective if the correct

source/mechanism can be identified.

6. Conclusions

On the basis of the results from this study the

following conclusions can be drawn:

�
DMS was the major source of nuisance VOS odor in
Philadelphia’ NEWPCP.

�
VOS production was strongly associated with the
characteristics of the incoming wastewater, which,

when mixed with RAS in the aeration tank, resulted

in the in situ production of DMS.

�
Methylation of hydrogen sulfide only, in the aeration
tank, could not explain the offensive VOS odor and

there may be a VOS precursor discharged into

NEWPCP and converted to DMS under favorable

conditions.

Acknowledgments

Data of total dissolved hydrogen sulfide concentra-

tion were provided by Dr. Alicia Elsetinow. This

research was done under the support of Debra McCarty,

Deputy Commissioner, and Leonard Gipson, Waste-

water Plants Manager, and Robert Lendzinski, North-

east Plant Manager of the Philadelphia Water

Department. The authors acknowledge the sampling

support provided by Thomas Healey and Philadelphia’s

Industrial Waste Unit staff.

References

America Public Health Association, 1998. America Water

Works Association; Water Environment Federation. Stan-

dard Methods for the Examination of Water and Waste-

water, 20th ed. Washington, DC.

Bak, F., Finster, K., Rothfuss, F., 1992. Formation of dimethyl

sulfide and methanethiol from methoxylated aromatic

compounds and inorganic sulfide by newly isolated anaero-

bic bacteria. Arch. Microbiol. 157, 529–534.

Burlingame, G.A., Mosley, C.J., 1995. Odor profiling of

environmental odors. Oper. Forum, Water Env. Fed. 12

(12), 10–14.

Burlingame, G.A., Suffet, I.H., Khiari, D., Bruchet, A.L., 2004.

Development of an odor wheel classification scheme for

wastewater. Water Sci. Technol. 49 (9), 201–209.

Cheng, X.-H., Peterkin, E., Narangajavana, K., 2005. Auto-

mated analysis of volatile organic sulfides with GC/MS in

wastewater and sludges, in publication.

Dauwe, B., Middelburg, J.J., 1998. Amino acids and hexosa-

mines as indicators of organic matter degradation state in

North Sea sediments. Limol. Oceanogr. 43, 782–798.

Drotar, A., Burton, G.A., Tavernier, J.E., Fall, R., 1987.

Widespread occurrence of bacterial thiol methyltransferases

and the biogenic emission of methylated sulfur gases. Appl.

Environ. Microbiol. 53, 1626–1631.

Fitzgerald, J.W., 1991. Marine aerosols: a review. Atmos.

Environ. 25, 533–545.

Ginzburg, B., Chalifa, I., Gun, I., Dor, I., Hadas, O., Lev, O.,

1998. DMS formation by dimethylsulfoniopropionate route

in freshwater. Environ. Sci. Technol. 32, 2130–2136.

Hvitvet-Jacobsen, T., 2000. Sewer Processes, Microbial and

chemical Process Engineering of Sewer Networks. CRC

Press, Washington, DC.

Kiene, R.P., Capone, D.G., 1988. Microbial transformations of

methylated sulfur compounds in anoxic salt marsh sedi-

ments. Microb. Ecol. 15, 275–291.

Lomans, B.P., Smolders, A.J.P., Intven, L., Pol, A., Op den

Camp, H.J.M., van der Drift, C., 1997. Formation of


dimethyl sulfide and methanetiol in anoxic fresh-

water sediments. Appl. Environ. Microbiol. 63,

4741–4747.

Lomans, B.P., Op den Camp, H.J.M., Pol, A., Vogels, GD.,

1999. Anaerobic versus aerobic degradation of dimethyl

sulfide and methanethiol in anoxic freshwater sediments.

Appl. Environ. Microbiol. 65, 433–438.

Lomans, B.P., Leijdekkers, P., Wesselink, J.J., Kakkes, P., Pol,

A., van der Drift, C., Vogels, G.D., Op den Camp, H.J.M.,

2001a. Obligate sulfite-dependent degradation of methoxy-

lated aromatic compounds and formation of methanethiol

and dimethyl sulfide by a freshwater sediment Isolate,

parasporobacterium paucivorans gen. nov., sp. nov. Appl.


Lomans, B.P., Luderer, R., Steenbakkers, P., Pol, A., van der

Drift, C., Vogels, G.D., Op den Camp, H.J.M., 2001b.

Microbial productions involved in cycling of dimethyl

sulfide and methanethiol in freshwater sediments. Appl.


Oho, T., Saito, T., Koa, T., 2000. Formation of MM from

L-methionine by porphyromonas gingivalis. Infect. Immun.

68, 6912–6916.

Porter, R., Witherspoon, J., Daigger, G., Fahnestock, L.,

Novak, J., Glindemann, D., Burlingame, G.A., Choudhary,

S.A., Lendzinski, R., Suffet, M., Rosenfeld, P., 2004.

Assessment of odor formation mechanisms in an activated

sludge basin at the Northeast Wastewater Treatment Plant.

Presented at Water Environment Federation, 77th Annual

Technical Exhibition and Conference, October 2004 (and in

Proceedings).

Richards, S.R., Rudd, J.W.M., Kelly, C.A., 1994. Organic

volatile sulfur in lakes ranging in sulfate and dissolved salt

concentration over five orders of magnitude. Limnol.

Oceanogr. 39, 562–572.

Stuetz, R., Franz-Bernd, F. (Eds.), 2001. Odors in Wastewater

Treatment. IWA Publishing, Alliance House, 12 Caxton

Street, London SW1H0QS, UK.

WEF, 1995. Odor Control in Wastewater Treatment Plants.

Manual of Practice, No. 22. Water Environment Federa-

tion, Alexandria, Virginia, USA.

Zinder, S.H., Brock, T.D., 1978. Methane, carbon dioxide, and

hydrogen sulfide production from the terminal methiol

group of methioning by anaerobic lake sediments. Appl.

Environ. Micriobiol. 35, 344–352.

a study on volatile organic sulfide causes of odors at philadelphia's northeast water pollution...

Documents