Onset of severe nitrification in mildly nitrifying chloraminated bulk waters and its relation to biostability

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 2

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

Onset of severe nitrification in mildly nitrifyingchloraminated bulk waters and its relation to biostability

Arumugam Sathasivana,*, Ian Fisherb, Tum Tamc

aDepartment of Civil and Construction Engineering, Curtin University of Technology, GPO Box U1987, Perth, WA 6845, AustraliabWatervale Systems, PO Box 318, Potts Point NSW 1335, AustraliacSydney Water Corporation, PO Box 73, West Ryde, NSW 2114, Australia

a r t i c l e i n f o

Article history:

Received 28 August 2007

Received in revised form 6 May 2008

Accepted 8 May 2008

Available online 21 June 2008

Keywords:

Nitrification

Chloramine

Ammonia Oxidising Bacteria (AOB)

Drinking Water

Bacterial Regrowth

Chloramine decay

Distribution system

Microbial decay factor

* Corresponding author. Tel.: þ618 92 267 29E-mail address: [email protected].

0043-1354/$ – see front matter ª 2008 Elsevidoi:10.1016/j.watres.2008.05.010

a b s t r a c t

Triggers of severe nitrification in distribution systems are still not clearly understood.

Recently, the biostability concept was proposed to explain the chloramine residual below

which signs of nitrification would be seen. To improve understanding, mildly nitrifying

bulk water samples (nitrite less than 0.010 mg-N/L) from Sydney Water distribution sys-

tems were incubated at constant temperatures and periodically analysed for nitrogenous

compounds and total chlorine. Total ammoniacal nitrogen in the sample was between

0.25 and 0.35 mg-N/L. Severe nitrification was triggered when chloramine residuals drop-

ped below about 0.4 mg/L – the critical threshold residual. In 45 such samples, the critical

threshold residual was 0.2–0.65 mg/L. The biostability concept was found to be useful in ex-

plaining the residual below which net growth of microorganisms begins. However, this

alone could not predict the critical threshold residual. Different means of overcoming

this problem are discussed. One of these is the use of the microbial decay factor method,

since microbiologically assisted chloramine decay in the samples studied was found to

be mostly the result of ammonia-oxidising bacterial activity. Nitrite levels in winter were

found to be poor indicators of nitrifying status. Overall the results were found to be useful

in controlling nitrification and to obtain early warning of severe nitrification.

ª 2008 Elsevier Ltd. All rights reserved.

1. Introduction Nitrification is a two-step microbial process. Ammonia is

Increased concern over disinfection by-products has prompted

manywaterutilities tochoose chloramine overchlorineasadis-

infectant in drinking water distribution systems. Additionally,

chloramine offers the advantage of greater stability than chlo-

rine. However, the use of chloramine presents some additional

challenges for water utilities. In addition to auto-decomposition

of chloramine, and its direct chemical reaction with waterborne

constituents, nitrification accelerates chloramine decay and

promotes bacterial regrowth. A survey of US water utilities

(Wilczak et al., 1996) indicated that nitrification may occur in

the systems of 63% of those utilities that use chloramine.

6; fax: þ618 92 662 681.au (A. Sathasivan).er Ltd. All rights reserved

initially converted by ammonia-oxidising bacteria (AOB) to ni-

trite and then nitrite is converted to nitrate by nitrite-oxidis-

ing bacteria (NOB). Before the extensive use of molecular

microbiological techniques, conclusions were made using

Nitrosomonas europea as a representative organism (Wolfe

et al., 1990). Recent use of a molecular microbiological ap-

proach has indicated N. europea is not the major species con-

trolling the nitrification process in chloraminated systems

where ammonium concentration is low (Regan et al., 2003;

Lipponen et al., 2004; Hoefel et al., 2005). In Finland and US dis-

tribution systems, Nitrosomonas oligotropha was found to be the

most abundant ammonia oxidiser. In distribution systems, it

.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 23624

is generally believed that mostly partial nitrification (ammo-

nia oxidation) takes place (Wolfe et al., 1990). Therefore, nitrite

is generally used as the indicator of nitrification status (Wolfe

et al., 1988) despite the fact that nitrite concentration

measured in bulk water does not necessarily correlate well

with AOB concentration measured by the most probable num-

ber (MPN) method in that water. The recent work of Regan

et al. (2003) and Hoefel et al. (2005) indicated that nitrite-

oxidising bacteria (NOB) are present in distribution system

bulk waters.

Nitrification can accelerate chloramine loss in distribution

systems (Cunliffe, 1991). The mechanism of chloramine decay

by nitrification is not well established, although chloramine

can be consumed in oxidation of nitrite to nitrate (Vikseland

et al., 2001). Although any presence of nitrite in the water

phase may suggest nitrifier activity, low levels of nitrite are

regarded as unimportant and actions are usually not taken

to eliminate it. However, once nitrification is severe, ammonia

and total chlorine can reach zero levels and nitrite can reach

very high concentrations. From a severely nitrifying status,

it is very difficult to re-establish an adequate chloramine

residual or preserve ammonia. For purposes of this paper,

severe nitrification is defined as the situation in which a water

sample immediately starts to show ammonia levels decreas-

ing towards zero after collection and incubation at 20–30 �C,

rather than just having a high initial nitrite level.

On many occasions, it has not been established whether

bacterial nitrification was the primary cause of the accelerated

decay or whether growth of the nitrifying bacteria was the

consequence of the chloramine decay. Data presented by Liu

et al. (2005) showed that nitrite concentration was dependent

upon the chloramine concentration. Other authors (Fleming

et al., 2005; Pintar et al., 2005) have also suggested dependence

of nitrifier activity on chloramine concentration. Wolfe et al.

(1988) reported that AOB can be present in concentrations as

high as 1.5 mg/L. Wilczak et al. (1996) reported that nitrifica-

tion was present even when chloramine residual was as

high as 2 mg/L. It was not clear whether severe nitrification

was present before this residual was reached in those samples

that exhibited nitrification.

Various means of identifying the time of onset of severe

nitrification have been proposed in the literature. In a study

to detect broad-based trends in nitrification occurrence,

Kirmeyer et al. (2004) proposed 0.05 mg/L NO2�-N as an

arbitrary critical threshold level. Pintar et al. (2005) suggested

chloramine drop as a better indicator. They did not suggest

a specific residual that provides an early warning or how

this could be used in practice. Cunliffe (1991) proposed

residual drop along with increased nitrite production as

a means of rapid diagnosis for severe nitrification. Molecular

microbiological analysis is being attempted by researchers

(Regan et al., 2003; Hoefel et al., 2005) to obtain an early

indication of its onset.

Woolschlager et al. (2001) and Harrington et al. (2002) pro-

posed a simple formula to determine the point of biostability.

Fleming et al. (2005) applied this concept to determine the re-

sidual below which potential for nitrification exists. This was

done by balancing growth and disinfection by disinfectant.

The model is valid for batch culture conditions since only

growth and disinfection are considered. In their model,

ammonia was used as food for AOB growth and dichloramine

was used as the disinfectant. The resulting equation is:

mmðfree ammoniaÞðKs þ free ammoniaÞ ¼ kd � TCl (1)

where mm is the maximum specific growth rate of AOB (day 1);

free ammonia is the sum of ammonia (NH3) and ammonium

(NH4þ) concentrations (mg-N/L); Ks is the half saturation con-

stant for AOB; kd is the rate constant for inactivation of AOB

by disinfectant (L day�1 mg Cl2�1); TCl is the total chlorine con-

centration (mg-Cl2/L). As initial estimates, they used 2.0 mg/L

for mm/kd and 0.5 mg-N/L for Ks.

If chloramine residual or its decay rate could be used as one

of the controlling parameters for determining onset of severe

nitrification, tools such as the microbial decay factor method

(Sathasivan et al., 2005), which showed promise in predicting

chloramine residuals in Sydney distribution systems (Fisher

et al., 2006), could be useful more generally. In this method,

chemical and microbial components of total decay of chlora-

mine (measured as total chlorine) are separated by simple de-

cay measurements at a controlled temperature. The rate of

chloramine drop due to microorganisms could therefore be

determined by the microbial decay factor method.

Despite all these findings, most utilities are still struggling

to cope with the uncertainty of the time of onset of severe ni-

trification. It would be particularly useful to understand what

triggers it. Before the triggers of severe nitrification in a distri-

bution system can be identified, it is necessary to understand

the triggers in bulk water alone. In this paper, such an under-

standing is developed by comparing profiles of nitrogenous

compounds and chloramine residuals in many different bulk

water samples from three sub-systems of Sydney Water dis-

tribution systems. The implications for management of distri-

bution system residuals are considered subsequently.

2. Materials and methods

Samples were collected from three different sub-supply sys-

tems of Sydney Water Corporation within a period ranging

from January 2003 to December 2005. General characteristics

of these samples are outlined in Table 1. These distribution

systems typically have chloramine residuals less than

1.7 mg/L total chlorine and deliver water over medium dis-

tances. These systems contain about 200 service reservoirs

(tanks). The farthest reservoir is about 50 km downstream

from the treatment plant except one arm which is much lon-

ger. Retention time in the systems varies between 5 and 8 days

except the longer arm which had a retention time up to 15

days. Altogether, 70 samples were collected from reservoirs

and pipelines within the first 50 km. Results from 55 samples

that had nitrite less than 0.010 mg-N/L are reported in this pa-

per. Of these, 20 samples had nitrite less than 0.002 mg-N/L

(the detection limit).

2.1. Prospect, Macarthur and Woronora(Sydney Water) systems

Sydney Water Corporation supplies water to about 5 million

customers living in the Sydney metropolitan area. On average,

Table 1 – General characteristics of samples analysedfrom Sydney Water systems

Parameters Range

Total chlorine (mg/L) 0.6–1.3

Ammonia-N (mg/L) 0.25–0.35

Nitrite-N (mg/L) Less than 0.010

Nitrate-N (mg/L) Less than 0.15

Dissolved organic carbon (mg/L) 3–3.6

Temperature (�C) 12–25

Chemical decay coefficienta (kc, h�1) 0.0013–0.0017 (�0.0002b)

Microbial decay factor (Fm) 0–2.5

Distance from treatment plant

(the first chloramination point)

to sampling point km

3–50

a Decay coefficients were measured at 20 �C as defined in

Sathasivan et al. (2005).

b Average 95% confidence interval rounded to the nearest fourth

decimal.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 2 3625

about 1700 ML of water per day is delivered to customers. A

treatment plant at the head of each system treats water drawn

from large surface storages (dams) by coagulation/floccula-

tion/direct filtration, followed by chlorination and ammonia-

tion. Chlorine is applied as the primary disinfectant and

ammonia is added subsequently. The finished water quality

entering all three systems meets the following criteria: turbid-

ity less than 0.1 NTU, chloramine (as total chlorine) 1.5–

1.7 mg/L, pH 8.0� 0.2 and Cl/NH3 ratio of 4. Water temperature

in these systems varies between 12 and 25 �C seasonally.

The Prospect water filtration plant (WFP) is Sydney’s major

treatment plant, which supplies about 85% of its water re-

quirement (about 1400 ML per day). Most of the raw water is

obtained from Warragamba Dam. Macarthur WFP supplies

about 100 ML per day and derives water from Broughton’s

Pass weir, which can be supplied from several large storages

(Nepean, Cataract and Cordeaux Dams) via natural river chan-

nels. Woronora WFP supplies about 160 ML per day with raw

water derived from Woronora Dam.

2.2. Determination of onset of severe nitrification

For determining the onset of severe nitrification, duplicate

samples collected from reservoirs and pipelines were trans-

ported to the laboratory under dark conditions at constant

temperature, immediately after collection. Temperature of in-

cubation in different samples ranged between 20 and 25 �C,

but only one temperature was adopted for a given sample.

They were incubated in new 500 mL PET containers (pre-

cleaned with 10% NaOCl solution) under dark conditions in

duplicate, as outlined by Sathasivan et al. (2005), until all the

disinfectant residual was lost. To minimise the measurement

error, measurements were made on both duplicate samples

and average results are reported. Nitrite and ammonia were

measured using Flow Injection Analysis (FIA) as detailed in

APHA et al. (1998). Nitrite was measured by the sulphanila-

mide method. Ammonia was measured by the phenate

method. Nitrite had the lowest detection level of 0.002

mg-N/L. If the discrepancy between duplicate readings was

more than 0.002 mg/L then previously reserved duplicate

samples were analysed and the average of all the readings

was reported. A similar procedure was adopted for ammonia

measurement. Since this had a measurement error of 10%,

a discrepancy of 10% was allowed. Total chlorine residuals

were measured by the DPD colorimetric method using

a HACH pocket colorimeter. Averages of two readings from

duplicate samples were reported. Total chlorine measure-

ment had an experimental error of 0.03 mg/L.

2.3. Microbial decay factor (Fm) method: determinationof chemical and microbial chloramine decay rate coefficients

Determining chemical and microbial decay rate coefficients

involves four major steps: sample preparation, incubation,

monitoring chloramine decay, and estimating decay (rate) co-

efficients from the resulting data. Sample preparation in-

volves splitting the sample into two sub-samples. The first

sub-sample was not processed at all and the second was

inhibited. Both sub-samples were then subjected to incuba-

tion at a constant temperature of 20 �C. For each sub-sample,

total chlorine was monitored over time and the decay coeffi-

cient was estimated. The term microbial decay factor (Fm)

was derived from this method, to characterize the relative

contribution of microbiologically assisted chloramine decay

to the total chloramine decay observed in bulk water. Fm is de-

fined as the ratio between the microbiologically assisted and

chemical decay coefficients of a given water sample measured

at 20 �C. Full details of the method were given by Sathasivan

et al. (2005).

2.4. Procedure for determining critical thresholdresidual – the residual at which severe nitrificationis triggered

In a bulk water sample that had detectable nitrification activ-

ity (nitrite between 0.002 and 0.010 mg-N/L), the total chlorine

profile appears to accelerate after the residual drops below

a particular value. This residual is termed the critical thresh-

old residual (CTR). In most of the samples, the CTR could be

readily identified. In some samples, however, it was not so ob-

vious, so a standardised procedure was adopted to identify it.

For any pair of measurements on a decay curve, the resid-

ual concentrations at time tn and tnþ1 are denoted cn and cnþ1,

respectively. Assuming the decay was first-order, the

decay coefficient for that time interval was calculated as

[(1/(tnþ1� tn))� ln(cn/cnþ1)]. For each consecutive pairs of mea-

surements, the calculated decay coefficient was plotted

against the corresponding average residual concentration

[(cnþ cnþ1)/2]. Then, the average first-order decay coefficient

for Phase 1 was estimated. Since the value of the decay rate

coefficient during Phase 1 is relatively stable, this could easily

be identified by visual inspection. Due to possible variation in

elemental decay rate coefficient between different pairs of

points (due to experimental error in chlorine measurement),

the residual corresponding to double the average first-order

decay coefficient in Phase 1 was adopted as the CTR of the

sample. This method is illustrated in Fig. 2 using the results

for Representative Sample A.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 5 10 15 20 25Time of incubation (days)

To

tal C

hlo

rin

e (

mg

/L)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Nit

ro

ge

no

us

Co

mp

ou

nd

s a

s N

(m

g/L

)

Fig. 1 – Total chlorine and inorganic nitrogenous

compound profiles for Representative Sample A (from

Macarthur system, Sydney): initial nitrite concentration

0.005 mg-N/L, incubation temperature 20 8C. Circles,

diamonds and squares are data points for total chlorine,

ammoniacal-N, and nitrite-N, respectively, for

unprocessed sample. Triangles represent total chlorine

profile in inhibited sample. Ammoniacal-N is the sum of

free and combined ammonia-nitrogen.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 23626

3. Results and discussion

3.1. General characteristics of mildly nitrifying samplesfrom Sydney

Mildly nitrifying samples were separated from the total set of

samples collected, using the criterion that their nitrite level

was below 0.010 mg-N/L. There were 55 such samples. Of

these, 35 samples had nitrite greater than or equal to

0.002 mg-N/L and 20 samples had less. The chloramine resid-

ual concentrations (as total chlorine) ranged between 0.7 and

1.7 mg/L. Ammonia concentration ranged between 0.25 and

0.35 mg-N/L. DOC concentrations were between 3 and

3.6 mg/L. Values of the chemical decay coefficients in the sam-

ples were 0.0014–0.0019 h�1 and the microbial decay factor

(Fm) was 0–2.5.

Freshly chloraminated finished water had the same chem-

ical and total chloramine decay coefficients (0.0013–

0.0016 h�1), its Fm was close to 0 and nitrite levels were below

detection limit. Similar analysis of distribution system sam-

ples indicated that, in mildly nitrifying samples, chemical de-

cay was almost identical to that in freshly chloraminated

finished water.

0.0000

0.0020

0.0040

0.0060

0.0080

0.0100

0.0120

0.0140

0.0160

00.10.20.30.40.50.60.70.80.9Middle total chlorine concentration (mg/L)

Elem

en

tal first o

rd

er d

ecay co

efficien

t (h

-1)

0.0056 h-1(Twice the average1st phase decay coefficient)

0.0028 h-1 (Average 1st

phase decay coefficient)

CTR

Fig. 2 – Determination of CTR for Sample A. chloramine is

represented by total chlorine: its residual is the average

residual for two consecutive measurement times and

elemental decay rate coefficient is the calculated decay rate

coefficient over the period between them. Dashed line

represents average decay rate coefficient over the first

phase. Dotted line represents double that value. CTR is

residual occurring when latter decay rate coefficient is

reached in Phase 2 (shown as thick vertical line). CTR for

this sample is 0.43 mg/L.

3.2. Typical profiles of total chlorine and nitrogenouscompounds in mildly nitrifying samples (nitriteconcentration 0.003–0.010 mg-N/L)

When mildly nitrifying bulk water samples having a tempera-

ture of around 20 �C were incubated at a constant tempera-

ture, consistent patterns of inorganic nitrogenous

compounds and residual profiles were obtained in 25 out of

35 samples. These patterns are explained in detail below for

Representative Sample A. When winter samples (temperature

at time of collection was less than 15 �C) were collected, the

pattern observed was slightly different. This is subsequently

explained for Representative Sample B. There were about 10

winter samples that had nitrite greater than 0.002 mg-N/L. In

two samples, despite the presence of detectable nitrite con-

centration, they did not show the signs of presence of nitrifi-

cation. This is explained for Representative Sample C.

Fig. 1 shows the behaviour of Representative Sample A. At

the beginning of Phase 1, nitrite-N and ammonia-N concentra-

tions were 0.005 and 0.29 mg/L, respectively. These levels

remained relatively stable (ammonia only dropped by

0.03 mg/L and nitrite concentration increase was not notice-

able) until Phase 2 started. In samples that had nitrite concen-

tration more than 0.005 mg-N/L, a slight increase in nitrite

was noticed in Phase 1, although the drop in ammonia was

similar to that for Representative Sample A. When Phase 2

started, nitrite-N concentration started to increase and am-

monia-N started to drop at a much faster rate. The rate of am-

monia-N drop in Phase 2 was 0.018 mg-N/day. This was six

times the ammonia drop rate of Phase 1. Similarly, as shown

in Fig. 2, the chloramine decay coefficient in Phase 2

(0.0140 h�1) was about five times greater than in Phase 1

(0.0028 h�1).

Fig. 2 illustrates how CTR was calculated for Representative

Sample A using the procedure set out in Section 2. In Fig. 2, the

elemental first-order decay coefficient and average residual

for each consecutive pair of data in Fig. 1 were plotted. The av-

erage Phase 1 decay coefficient was calculated to be 0.0028 h�1

by selecting the points which are mostly close together or by

excluding the points which obviously deviate from the origi-

nal first-order relationship. The residual at which the decay

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0.00 2.00 4.00 6.00 8.00 10.00 12.00Time of Incubation (days)

To

ta

l c

hlo

rin

e (

mg

/L)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40 Nit

ro

ge

no

us

co

mp

ou

nd

s (

mg

-N

/L)

Fig. 3 – Total chlorine and inorganic nitrogenous

compound profiles for Representative Sample B (from

Prospect system, Sydney): initial nitrite and ammonia

concentrations were 0.003 and 0.29 mg-N/L, respectively,

initial temperature was 12 8C, and incubation temperature

was 23 8C. Circles, diamonds, and squares are data points

for total chlorine, ammoniacal-N, and nitrite-N,

respectively. Triangles represent total chlorine data points

in inhibited sample. Ammoniacal-N is the sum of free and

combined ammonia-nitrogen.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 2 3627

coefficient reached double the average Phase 1 decay coeffi-

cient (0.0056 h�1) is 0.43 mg/L – the CTR value.

Representative Sample B was collected when the water

temperature was 12 �C. When incubated at 23 �C, the sample

showed behaviour similar to that of samples with substan-

tially greater nitrite concentration. The results are presented

for Representative Sample B in Fig. 3. Table 2 shows the

Table 2 – Water quality at the time Representative Samples we

Sample A

Physical and chemical characteristics

Temperature (�C) 18.2

Total chlorine (mg-Cl2/L) 0.85

Ammoniacala-N (mg/L) 0.29

Cl/N ratio 3.3

Nitrite-N (mg/L) 0.005

Nitrate-N (mg/L) 0.10

DOC (mg/L) 3.0

pH 8.0

Decay characteristicsb

Total decay coefficient, kT (h�1) 0.0028� 0.0002

Chemical decay coefficient, kc (h�1) 0.0017� 0.0002

Microbial decay coefficient, km (h�1) 0.0011� 0.0004

Microbial decay factor, Fm (¼km/kc) 0.65

Critical threshold residual (CTR) (mg/L) 0.43

a Ammoniacal nitrogen is the sum of nitrogen in the form of free ammo

b Decay characteristics were measured at 20 �C (decay characteristics fo

(2005) and bacterial activity was inhibited using silver nitrate. Decay coe

�95% confidence interval rounded to the fourth decimal place are report

c Below detection level of 0.002 mg-N/L.

characteristics of the sample at the time of collection. In these

samples, nitrite concentration increased from 0.003 to

0.02 mg/L within a few days of incubation and stayed rela-

tively stable (an increase of another 0.005–0.025 mg-N/L) until

CTR was established. Nitrifying bacteria in these samples

have probably become active (due to suitable temperature

for their growth or activity). However, neither chloramine de-

cay coefficient nor ammonia concentrations were adversely

affected within the first phase. As soon as the second phase

started, ammonia concentrations were substantially reduced.

In Sample B, two phases were again evident (Fig. 3). The de-

cay coefficients are summarised in Table 2. The first phase de-

cay coefficient was 0.0052 h�1, while the chemical decay

coefficient was 0.0015 h�1. Therefore, the microbial decay rate

coefficient was 0.0037 h�1, so that microbial decay coefficient

was 2.5 times that of chemical decay coefficient (i.e. Fm¼ 2.5).

As soon as the second phase started, the decay rate coefficient

increased by about four fold to 0.0225 h�1 and nitrite increased

to 0.052 mg-N/L. The CTR in this sample was around 0.49 mg/L,

little different from the CTR of 0.43 mg/L in the mildly nitrifying

Representative Sample A (Figs. 1 and 2). After the residual had

reached the CTR of 0.49 mg/L, nitrite increased, chloramine de-

cay rate and ammonia drop accelerated. Behaviour similar to

this was observed in 20 samples that were collected in winter.

Eight of them had nitrite greater than or equal to 0.002 mg-N/L

and 12 of them had less. Therefore, nitrite concentration in win-

ter samples is an unreliable indicator.

3.3. Confirmation of AOB presence in RepresentativeSample A

Although it is evident from Fig. 1 that there is mild nitrification

occurring in the original sample, additional experiments were

performed to confirm the existence of activity by ammonia

oxidisers. In one set of experiments, two sub-samples were

re collected

Sample B Sample C

12 15

1.19 1.7

0.29 0.35

4.10 4.85

0.003 0.001c

0.12 0.08

3.3 3.2

7.9 8.1

0.0052� 0.0003 0.0013� 0.0001

0.0015� 0.0002 0.0013� 0.0002

0.0037� 0.0005 0� 0.0003

2.5 0

0.50 0

nia and that in the form of combined chlorine.

r Sample C were measured at 23 �C) as defined in Sathasivan et al.

fficient was estimated by considering only the first phase. Mean and

ed.

1

)

0.12

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 23628

prepared to confirm the existence of nitrifying organisms. The

first sub-sample was filtered through a 0.2 mm polycarbonate

membrane. To the second sub-sample, silver nitrate was

added to inhibit microorganism activity as described in Satha-

sivan et al. (2005). Nitrite profiles in both inhibited and filtered

samples (Fig. 4) were similar and depleted to below detection

levels within 7 days of incubation, indicating that inhibition

by silver is as effective as filtration in reducing the ammo-

nia-oxidising bacterial activity in the sample. Therefore, only

chemical reactions could be of significance in these processed

samples. The depletion of nitrite, possibly, was a result of ni-

trite oxidation by chloramine. It also seems that nitrite oxida-

tion rate in the sample was very slow. In the original sample,

therefore, nitrite production must be present to balance the

nitrite loss by chemical nitrite oxidation. In addition, if there

were NOB in the sample (as proposed by Regan et al., 2003),

then there must have been further nitrite production to com-

pensate for NOB oxidation that might occur in the first phase

of decay in the unprocessed sample. Together these results

imply the presence of microbial nitrite production (including

the contribution of AOB) during the first phase of chloramine

decay in the Representative Sample.

3.4. Confirmation that the accelerated decay inRepresentative Sample A was mainly the result of AOBactivity

In another set of experiments, microbial and chemical decay

coefficients were measured for sub-samples of randomly se-

lected mildly nitrifying bulk water samples, to which ammo-

nia had been added to double the total ammoniacal nitrogen

level of the original sample. The results of the experiment

obtained for Representative Sample A are shown in Fig. 5

and Table 3. In the ammonia-added sub-sample, nitrite in-

creased to 0.010 mg-N/L within a few days. After the initial in-

crease, nitrite concentration stayed steady until CTR for this

sample was reached. Similarly, the total chlorine decay coeffi-

cient increased from 0.0028 h�1 to 0.0039 h�1 but the chemical

decay coefficient decreased from 0.0017 h�1 to 0.0013 h�1

0

0.02

0.04

0.06

0.08

0.1

0.12

0.0 5.0 10.0 15.0 20.0Incubation Time (days)

Nitrite (m

g-N

/L

)

Fig. 4 – Comparison of nitrite profile in original sample

with that of inhibited and filtered sub-samples of Sample

A. Nitrite measurement detection limit is 0.002 mg-N/L; i.e.

nitrite levels on days 10 and 12 of incubation are negligible

for inhibited and filtered sample. Circle, triangle, and

square represent nitrite profiles in unprocessed, inhibited

and filtered samples, respectively.

(Table 3), probably as a result of increased chlorine to ammo-

nia ratio (Jafvert and Valentine, 1992). Therefore, the magni-

tude of the microbial decay coefficient increased from

0.0011 h�1 to 0.0026 h�1. These along with nitrite profiles

clearly indicated the presence of ammonia-oxidising bacterial

(AOB) activity. AOB in the sample responded well to the addi-

tion of ammonia to double the total ammoniacal nitrogen con-

centration. In the original sample, neither nitrite level nor

total chloramine decay coefficient changed despite the con-

tinued drop in residual and increase in free ammonia-N before

CTR was reached. But the addition of ammonia-N to double

the total ammoniacal nitrogen doubled nitrite-N and nearly

doubled the microbial decay coefficient. Chloramine levels

in both samples are approximately similar. At similar chlora-

mine levels, chemical nitrite oxidation would have been sim-

ilar. Had there been any NOB, the concentration in the

ammonia-added and original sub-samples would also be sim-

ilar. Therefore, chemical and microbial nitrite oxidation

would be similar. Therefore, nitrite production must have

been nearly doubled by the addition of ammonia, to double

the total ammoniacal nitrogen. Similar results were obtained

when ammonia was added to double the total ammoniacal ni-

trogen in another seven randomly selected samples that had

nitrite concentration between 0.004 and 0.007 mg-N/L. The ob-

servation that the microbiologically assisted chloramine de-

cay coefficient has also doubled in all these samples

supports the conclusion that microbiologically accelerated

chloramine decay rate is mainly controlled by activity of AOB.

3.5. Profiles of total chlorine and nitrogenouscompounds in samples having nitrite concentration belowdetection level (nitrite less than 0.002 mg-N/L)

Profiles in eight of 20 samples for which nitrite was below de-

tection level were similar to those shown for Representative

Sample C in Fig. 6. All these samples were collected at the

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 5 10 15 20

Incubation time (days)

To

tal ch

lo

rin

e,

To

tal am

mo

niacal n

itro

gen

(m

g/L

0.00

0.02

0.04

0.06

0.08

0.10

Nitrite-N

(m

g/L

)

Fig. 5 – Profiles of chloramine and nitrite-N concentration

in samples for which 0.3 mg-NH3-N mg/L ammonia was

added to Representative Sample A. Circles, diamonds and

squares are data points for total chlorine, ammonia-N and

nitrite-N, respectively, for uninhibited sample. Triangles

represent total chlorine profile of inhibited sample.

Table 3 – Decay characteristics of differently processed sub-samples of Representative Sample A

Sample First-order decay coefficient (h�1)� c.i.b Fma (¼km/kc)

Unprocessed (kT)a Inhibited (kc)a km (h�1)a (¼kT� kc)

Original sample 0.0028� 0.0002 0.0017� 0.0002 0.0011� 0.0004 0.65

NH3 added sample 0.0039� 0.0002 0.0013� 0.0001 0.0026� 0.0003 2.0

a Fm is the microbial decay factor. See Sathasivan et al. (2005) for further details of calculation of these values.

b c.i. is the 95% confidence interval of decay coefficient (estimation is rounded to the fourth decimal place).

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 2 3629

outlet of treatment plants. Characteristics of Representative

Sample C are given in Table 2. In these samples, nitrite levels

remained below detection levels for the whole incubation pe-

riod of two months. There was no detectable microbiological

decay (i.e. profiles of total chlorine in inhibited and unpro-

cessed samples were similar) and there was no sign of sudden

increase in chloramine decay rate or nitrite production. This

sample had an Fm value of 0, meaning there was no microbial

chloramine decay. The chemical and total chloramine decay

profiles were very similar, with both rate coefficients being

slightly higher at the beginning than they were later. The later

decay coefficient was 0.0013 h�1.

In two samples that had nitrite greater than or equal to

0.002 mg-N/L and were collected from long pipelines (data

not presented), behaviour similar to that of samples having ni-

trite levels below detection level was evident, even though the

nitrite level was 0.004 mg-N/L. Even in unprocessed sub-sam-

ples, nitrite levels decreased to below detection levels. There

could be two reasons for this: experimental error in initial ni-

trite-N concentration measurement or initial nitrite concen-

trations in the samples could have been the result of biofilm

activity in the system. Repeated measurements confirmed

that the measurement was correct. In addition, total and

chemical decay coefficients in these sub-samples were

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00Time of incubation (days)

To

ta

l C

hlo

rin

e (m

g/L

)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Nitro

ge

no

us

C

om

po

un

ds

a

s N

(m

g/L

)

Fig. 6 – Total chlorine and inorganic nitrogenous

compound profiles for Representative Sample C (from

Woronora system, Sydney) within 3 km from the

treatment plant: initial nitrite concentration 0.001 mg-N/L,

incubation temperature 23 8C. Circles, diamonds, and

squares are data points for total chlorine, ammoniacal-N,

and nitrite-N, respectively, in the unprocessed sample.

Triangles represent total chlorine data points in inhibited

samples. Ammoniacal-N is the sum of free and combined

ammonia-nitrogen.

0.0019 and 0.0017 h�1 respectively, indicating minimal micro-

bial activity in this sample. Hence, it was concluded that ni-

trite presence in the sample was the result of biofilm activity

in the system rather than in the bulk water sample itself.

In contrast, there were 12 samples having nitrite levels at

or below detection level when collected in winter (water tem-

perature at the time of collection was 12 �C) and incubated at

23 �C, which showed behaviour similar to that of samples with

substantially greater nitrite concentration. The results are

similar to the one presented for Representative Sample B in

Fig. 3. In these samples, nitrite concentration increased from

below detection to higher nitrite concentration (up to about

0.008 mg-N/L) within a few days of incubation and stayed rel-

atively stable until the CTR was reached. Nitrifying bacteria in

these samples had probably become active (due to suitable

temperature). Therefore, lower nitrite concentration in winter

should not be interpreted as absence of nitrifiers.

In summary, there were two distinct phases in the decay of

chloramine (as total chlorine) in bulk water that showed evi-

dence of mild nitrification. The phases are demarcated by

the critical threshold residual (CTR). In the first phase, total

chlorine decayed slowly, and neither ammonia nor nitrite

was much affected. Ammonia utilisation rate and nitrite pro-

duction rate were very low in the first phase. In the second

phase, ammonia loss accelerated and nitrite production rate

increased substantially. It should, however, be noted that a de-

tectable nitrite level in a sample taken from a distribution sys-

tem did not always mean the presence of mild nitrification in

the sample itself. Further, nitrite levels below the detection

limit in winter do not always mean absence of nitrification.

From the experiments on Sydney Water distribution system

samples, it was also concluded that microbial decay rate

was mostly due to AOB.

3.6. Variation of CTR within and between Sydneysystems

Phases 1 and 2 were observed in almost all 46 samples that

showed mild nitrification activity. It therefore appears to be

a very common phenomenon in samples having mild nitrify-

ing activity. This behaviour was independent of incubation

temperature (within the range of 20–25 �C) and that of initial

nitrite concentration (within the range of detection limit to

0.010 mg-N/L). To determine whether the CTR was dependent

on the origin of the samples, the average and mean values of

CTR were calculated for samples grouped according to the sys-

tem from which they were collected. Macarthur, Prospect, and

Woronora systems had CTR values at 0.40� 0.13, 0.46� 0.11,

and 0.37� 0.12 mg/L, respectively; i.e. the mean values of the

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 0.1 0.2

Free ammonia (NH3+NH

4

+) as mg-N/L

To

ta

l c

hlo

rin

e c

on

ce

ntra

tio

n (

mg

/L)

Fig. 7 – Total chlorine concentration versus free ammonia

concentration. Solid line represents the curve proposed in

this paper (Ks [ 0.18 mg/L; mm/kd [ 2.0 mg/L). The filled

circles represent points at which severe nitrification was

triggered when many different mildly nitrifying bulk water

samples were incubated at constant temperature.

Triangles (and a dashed line through them) represent

points on decay curve for a sample that had 0.2 mg/L CTR.

The filled triangle represents the point at which severe

nitrification was triggered.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 23630

CTR were not significantly different between the systems

from which samples were collected. The data were therefore

pooled and variation of CTR for all samples that showed evi-

dence of nitrification was determined. The mean and stan-

dard deviation of the CTR were then 0.43 and 0.12 mg/L,

respectively.

3.7. The relationship between CTR and biostableresidual concentration

The first phase is associated with stable nitrite and ammonia

levels along with slower chloramine decay. Microbial decay

did not show major change. The observation of negligible

growth in samples such as Sample A (Fig. 1) during the first

phase may be due to the growth and disinfection of AOB bal-

ancing each other or due to higher disinfection than growth.

This is similar to the arguments of Fleming et al. (2005) and

Woolschlager et al. (2002) in relation to biostability. Below the

biostable residual concentration (BRC), there is a possibility of

bacterial regrowth i.e. triggering of severe nitrification.

Whether it is seen or not should depend on the presence of

a mass of organisms sufficient to affect measured parameters

such as nitrite, ammonia, or chloramine decay. Once this crit-

ical mass of bacteria is reached, such signs would be seen. In

contrast, the CTR is the chloramine residual at which the decay

rate of a mildly nitrifying sample suddenly increases with time;

i.e. a mass of bacteria sufficient to affect the decay rate has al-

ready been reached. Therefore, the CTR should always be lower

than the BRC in mild nitrifying samples, because net bacterial

growth must have started at a concentration slightly above

the residual at which severe nitrification signs are seen. The

progress, in terms of total chlorine residuals and free ammo-

nia, of one batch sample is shown in Fig. 7 as triangles. Open tri-

angles indicate that mild nitrification is prevailing. The filled

triangle indicates the point (total chlorine and free ammonia)

at which nitrification is triggered. Therefore, the total chlorine

concentration of filled triangle represents CTR value for a given

free ammonia concentration. Similarly for other samples, filled

circles show the CTR and free ammonia concentration for

many different samples. Consequently, the BRC curve for this

(ammonia, chloramine, pH and temperature) regime should

lie above all filled data points.

The relationship between BRC and free ammonia is given

by Eq. (1). This equation is valid for batch culture conditions,

because only bulk water growth and disinfection are consid-

ered. For determination of BRC, an envelope was drawn about

0.05 mg/L more than CTR values for each free ammonia con-

centration. Curve fitting to the envelope of CTR produced

a Ks value of 0.18 mg/L and mm/kd of 2.0 mg/L. Fleming et al.

(2005) obtained Ks and mm/kd values of 0.5 and 2.0 mg/L using

data from a pilot-scale distribution system in the US. The dif-

ference could be explained by the difference in nature of mi-

croorganisms present in each system and by the difference

in methods of data collection. Fleming et al. (2005) obtained

the data from a pilot-scale system in the US, whereas data

in this paper were obtained from batch experiments on sam-

ples from Sydney Water distribution systems. N. oligotropha

was found to be present in US and Finland distribution sys-

tems (Regan et al., 2003; Lipponen et al., 2004). N. oligotropha

isolated from a river estuary had Ks values ranging between

0.42 and 1.05 mg/L (Stehr et al., 1995). This implies that AOB

in the studied system had higher affinity for free ammonia

than reported for N. oligotropha. Due to low availability of

free ammonia in the system, the starvation conditions might

have selected a species having higher affinity. This is in line

with the findings of Bollmann et al. (2002) that some species

of AOB have the capability to grow well under starvation

conditions.

From the progress of one sample’s chloramine and free

ammonia concentrations (shown as triangles), it can be seen

that despite the fact that the residual has dropped below the

BRC (0.69 mg/L), nitrification did not trigger. The residual

had to drop to 0.2 mg/L (CTR) before signs of nitrification

were seen. Similar results were observed for other samples,

but only the CTR points are marked in Fig. 7. These results,

therefore, indicate that BRC cannot predict CTR in a mildly ni-

trifying sample.

When additional ammonia was added to Representative

Sample A, to double total ammoniacal nitrogen, the CTR

changed from 0.43 to 0.33 mg/L. The resulting BRC would

have risen to 1.38 mg/L (for free ammonia concentration of

about 0.45 mg-N/L, using Ks and mm/kd values obtained in

this study). Initial residual was 0.85 mg/L (well below the

BRC). Bacterial growth in the ammonia-added sample should

have started as soon as the ammonia was added. Therefore,

severe nitrification in the ammonia-added sample should

have been triggered earlier. However, in both the original

(Fig. 1) and the ammonia-added sample (Fig. 5) it took about

10 days to show signs of severe nitrification. This behaviour

could not be explained with the biostability concept. A possi-

ble reason is that presence of higher free ammonia may have

led to selection of another species than that growing in the

original Sample A.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 2 3631

In summary, the biostability concept helps to understand

whether there is any potential for nitrification for a given re-

sidual and free ammonia regime. This does not necessarily in-

dicate when severe nitrification would be triggered in mildly

nitrifying samples tested under batch conditions. This was at-

tributed to the need for a critical mass of bacteria to be present

to show signs of severe nitrification. The maximum CTR ob-

served in mildly nitrifying water samples was correlated

with BRC.

3.8. Implications for residual management indistribution systems

Previous discussion demonstrated that residual drop below the

critical threshold residual (CTR) in a mildly nitrifying bulk water

sample obtained from Sydney distribution systems can trigger

severe nitrification. Whether a similar CTR value exists in other

systems needs to be established. However, similar behaviour

consisting of sudden increase in chloramine decay rate, rapid

drop in ammonia, and rapid increase in nitrite could be

expected in mildly nitrifying water from other distribution sys-

temsat the timeseverenitrification is triggered.Taken together,

these indicators can provide earlier warning that active man-

agement of a chloraminated distribution system is required to

avoid severe nitrification. Use of the BRC to establish the maxi-

mum CTR in distribution system samples is encouraging.

Although the mean CTR was 0.43 mg/L, the highest value

observed was 0.65 mg/L, and the 99.5 percentile value was

0.74 mg/L. As there are other possible factors contributing to

nitrification in distribution systems, even this maximum value

is not necessarily a sufficiently high target residual. If water in

a small section of pipe has a residual lower than the CTR, then

that section could act as a breeding and seeding ground for ni-

trifiers with subsequent chloramine losses in other parts of the

distribution system downstream. Distribution system pipes

and reservoirs host biofilm and biofilm can be expected to con-

tinuously shed bacteria. Biofilm or particle-associated bacteria

are widely reported to have higher resistance to chloramine

(LeChevallier et al., 1988) implying that a higher BRC or CTR

may prevail in systems dominated by biofilms. If chloramine

in such places drops below the CTR, nitrification could be trig-

gered in bulk water in these local pockets. Therefore, maintain-

ing total chlorine residual at more than the maximum CTR

(preferably above 1 mg/L to provide an allowance for other con-

tributing factors) in all parts of the distribution system, and at

all times, should be consideredas a major componentof any ni-

trification control strategy. This guideline value was selected

based on the behaviour of Sydney Water samples.

Seeding from local pockets of nitrification or shedding

from biofilm can increase the bacterial numbers or the activity

of AOB in the bulk water. This could manifest as slightly in-

creased use of ammonia and production of nitrite or faster

chloramine decay rates (similar to Sample B in Fig. 3). If signs

of a bacterial number increase are seen, such as an increased

microbial chloramine decay coefficient in bulk water, espe-

cially when the sample is in the first phase, dilution with

good quality water is preferable to simply ‘‘topping up’’ with

free chlorine residual. This is because dilution with good qual-

ity water will help reduce the bacterial numbers. It may be

hard to identify the increased utilisation of ammonia or

production of nitrite when samples are in the first phase or

in a low temperature condition. The use of the microbial decay

factor method (Sathasivan et al., 2005) can assist in under-

standing the status of microbial decay rate irrespective of pre-

vailing temperature, as shown earlier for Sample B (Fig. 3).

The behaviour that the chloramine decay coefficient in-

creases five fold after the CTR is reached provides an opportu-

nity to obtain early warning and to take necessary action. In

the distribution system reservoirs, the change in decay rate

will be slower than that observed in batch systems. This can

provide more time to react to changes in decay rates.

4. Conclusion

To improve understanding of conditions that trigger severe ni-

trification, mildly nitrifying bulk water samples (nitrite less

than 0.010 mg-N/L) from Sydney Water’s chloraminated dis-

tribution systems were incubated at constant temperatures

and periodically analysed for nitrogenous compounds and to-

tal chlorine. A method was developed to identify a sample’s

total chlorine residual at which the decay rate (and nitrite

level) suddenly increased.

The biostability concept is important for understanding the

residual below which net growth of microorganisms could oc-

cur in bulk waters. However, it was not capable of predicting

the CTR, the residual at which signs of severe nitrification

were first detectable, in all mildly nitrifying waters tested.

This was mainly attributed to the requirement for a critical

mass of microorganisms to develop after the biostable resid-

ual concentration (BRC) is reached, before an impact on mea-

sured parameters can be seen. The equation defining

biostability was successfully fitted to the maximum values

of CTR from all samples. In addition, in ammonia-added

mildly nitrifying samples, the biostability concept did not ex-

plain the observed behaviour. Selection of different species

due to increased ammonia availability could be one of the

reasons.

It was also found that in randomly selected mildly nitrifying

bulk water samples tested from this distribution system, mi-

crobiologically assisted chloramine decay was mainly due to

ammonia-oxidising bacterial (AOB) activity. This implies that

when samples are mildly nitrifying, the microbiologically

assisted chloramine decay coefficient might be used as an indi-

cator of AOB presence. Further, utility managers can obtain an

early warning of severe nitrification from the trend of residuals

alone. If the residual approaches the maximum CTR, or if the

residual decay rate suddenly increases for no obvious reasons

(such as temperature increase or longer retention time), then it

is highly likely that even a mildly nitrifying water will turn into

a severely nitrifying one. Therefore, it is very important to

maintain the residual well above the maximum CTR, especially

in pockets of nitrification such as dead-end pipes.

Acknowledgement

Sydney Water Corporation approved the use of data from

Sydney distribution systems in Australia and funded the

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 2 3 – 3 6 3 23632

collection of samples and associated experiments. The assis-

tance of George Kastl, Len Rogerson, Gary Schultz, Philip

Broad and Clive Copelin is also gratefully acknowledged.

r e f e r e n c e s

APHA, AWWA, WEF, 1998. Standard Methods for the Examinationof Water and Wastewater, 20th ed. Washington, DC.

Bollmann, A., Bar-Gilissen, M., Laanbroek, H.J., 2002. Growth atlow ammonium concentrations and starvation response aspotential factors involved in niche differentiation amongammonia-oxidizing bacteria. Appl. Environ. Microbiol. 68 (10),4751–4757.

Cunliffe, D.A., 1991. Bacterial nitrification in chloraminated watersupplies. Appl. Environ. Microbiol. 57 (11), 3399–3402.

Fisher, I., Kastl, G., Sathasivan, A., 2006. Projecting chlorine andchloramine profiles in a distribution system, paper e6177 (CD-ROM). In: Enviro06 Australian WA Federal Convention.Melbourne, Australia, May 2–6, 2006.

Fleming, K.K., Harrington, G.W., Noguera, D.R., 2005. Nitrificationpotential curves: a new strategy for nitrification prevention. J.AWWA 97 (8), 90–99.

Harrington, G.W., Noguera, D.R., Kandou, A.I., Vanhoven, D.J.,2002. Pilot scale evaluation of nitrification control strategies. J.AWWA 94 (11), 78–89.

Hoefel, D., Monis, P.T., Grooby, W.L., Andrews, S., Saint, C.P., 2005.Culture-independent techniques for rapid detection ofbacteria associated with loss of chloramine residual ina drinking water system. Appl. Environ. Microbiol. 71 (11),6479–6488.

Jafvert, C., Valentine, R., 1992. Reaction scheme for thechlorination of ammoniacal water. Environ. Sci. Technol. 26(3), 577–586.

Kirmeyer, G.J., Martel, K., Thompson, G., Radder, L., Klement, W.,LeChevallier, M., Baribeau, H., Flores, M., 2004. OptimizingChloramine Treatment, second ed. AWWA ResearchFoundation, Denver.

LeChevallier, M.W., Cawthon, C.D., Lee, R.G., 1988. Factorspromoting survival of bacteria in chlorinated water supplies.Appl. Environ. Microbiol. 54 (3), 649–654.

Liu, S.B., Taylor, J.S., Randall, A.A., Dietz, J.D., 2005. Nitrificationmodelling in chloraminated distribution system. J. AWWA 97(10), 98–108.

Lipponen, M.T.T., Suutaria, M.H., Martikainen, P.J., 2004.Occurrence of nitrifiers and diversity of ammonia-oxidizingbacteria in developing drinking water biofilms. Water Res. 38,4424–4434.

Pintar, K.D., Anderson, W.B., Slawson, R.M., Smith, E.F., 2005.Assessment of a distribution system nitrification criticalthreshold concept. J. AWWA 97 (7), 116–1219.

Regan, J.M., Harrington, G.W., Baribeau, H., Leon, R.D.,Noguera, D.R., 2003. Diversity of nitrifying bacteria in full-scale chloraminated distribution systems. Water Res. 37 (1),197–205.

Sathasivan, A., Fisher, I., Kastl, G., 2005. A simple methodfor measuring microbiologically assisted chloraminedecay in drinking water. Environ. Sci. Technol. 39 (14),5407–5413.

Stehr, G., Bottcher, B., Dittberner, P., Rath, G., Koops, H.P., 1995.The ammonia-oxidising nitrifying population of the River Elbeestuary. FEMS Microbial. Ecol. 17, 177.

Vikseland, P.J., Ozekin, K., Valantine, R.J., 2001. Monochloraminedecay in model distribution system waters. Water Res. 35 (7),1766–1776.

Wilczak, A., Jacangelo, J.G., Marcinko, J.P., Odell, L.H., Kirmeyer, G.J., Wolfe, R.L., 1996. Occurrence of nitrification inchloraminated distribution systems. J. AWWA 88 (7), 74–85.

Wolfe, R.L., Means III, E.G., Davis, M.K., Barrett, S.E., 1988.Biological nitrification in covered reservoirs containingchloraminated water. J. AWWA 80 (9), 109–114.

Wolfe, R.L., Lieu, N.I., Izaguirre, G., Means, E.G., 1990. Ammoniaoxidizing bacteria in a chloraminated distribution system:seasonal occurrence, distribution, and disinfection resistance.Appl. Environ. Microbiol. 56 (2), 451–462.

Woolschlager, J.E., Rittmann, B., Piriou, P., Schwartz, B., 2001.Using a comprehensive model to identify the majormechanisms of chloramine decay in distribution systems.Water Sci. Technol.: Water Supply 1 (4), 103–110.

Woolschlager, J.E., Rittmann, B., Piriou, P., Schwartz, B., 2002.Developing a simple factor to evaluate microbiologicalstability in distribution systems. In: Proceedings of theAnnual Meeting of the AWWA. Denver, Colorado. ISBN 1-58321-197-7.