Water Research 36 (2002) 3479–3488

The decay of chlorine associated with the pipe wall in water

distribution systems
N.B. Hallama, J.R. Westa, C.F. Forstera,*, J.C. Powellb, I. Spencerc
a
School of Civil Engineering, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
b
Binnie Black and Veatch Ltd., Grosvenor House, 69 London Road, Redhill, Surrey RH1 1LQ, UK
c
Severn Trent Water Ltd., Leicester Water Centre, Gorse Hill, Amstey, Leicester LE7 7GU, UK
Received 12 June 2001; received in revised form 9 November 2001; accepted 1 February 2002

Abstract

Free chlorine decay rates in water distribution systems for bulk and wall demands should be modelled separately as
they have different functional dependencies. Few good quality determinations of in situ wall demand have been made
due to the difficulty of monitoring live systems and due to their complexity. Wall demands have been calculated from
field measurements at 11 locations in a distribution system fed from a single source. A methodology for the laboratory
determination has been evolved and shown to give results that are similar to the in situ results. Pipe materials were
classified as either having high reactivity (unlined iron mains) or low reactivity (PVC, MDPE and cement-lined ductile
iron). The results indicate that wall decay rates for the former are limited by chlorine transport and for the latter by pipe
material characteristics. The wall decay rate is inversely related to initial chlorine concentration for low reactivity pipes.
In general, water velocity increases wall decay rates though the statistical confidence is low for low reactivity pipes. A
moderate biofilm coating did not influence the wall decay rate for low reactivity pipes. r 2002 Elsevier Science Ltd. All
rights reserved.

Keywords: Chlorine; Wall decay rate; Distribution systems

1. Introduction model to quantify chlorine decay in distribution systems

uses a first-order decay equation with respect to
Despite the effective use of chlorine over many chlorine:
decades as a disinfectant for potable water supplies,
dC=dt ¼
kc
there are still regulatory failures in the quality of
drinking water, which cause concern to the water or
utilities. An understanding of how chlorine concentra-
tions will decay with time, due to reactions with trace Ct ¼ C0 expð
ktÞ:
compounds contained in the bulk water, material
When considering in situ chlorine decay, the reactions
accumulated at the pipe wall or with the pipe material
associated with the bulk liquid are frequently separated
itself is, therefore, central to managing water quality. It
from those associated with the pipe wall. Thus,
is widely accepted that chlorine in distribution systems
AWWARF [1] defined the overall decay constant as
will decay, either due to reactions with compounds
the sum of a bulk constant and a wall constant:
contained within the bulk water (bulk decay) or due to
reactions at the pipe wall (wall decay). The simplest k ¼ k b þ kw ;

where k is the first-order decay constant (h
1), kb the

*Corresponding author. bulk first-order chlorine decay constant (h
1) and kw the
E-mail address: [Link]@[Link] (C.F. Forster). wall first-order chlorine decay constant (h
1).

0043-1354/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 0 5 6 - 8
3480 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488

This model has the advantages of simplicity and the to develop a greater understanding of the factors which
decay constants, being in the same units, allow easy influence wall decay and to enable the results of
comparison. Measurements of the decay of chlorine laboratory studies to be more directly applied to
associated with the pipe wall have been performed both distribution systems. The study aimed at examining a
in the laboratory and under field conditions. Powell et al. wider range of pipe types with more repeat wall decay
[2] reported the factors which control wall chlorine determinations for in situ conditions, compared with
decay through field surveys of the in situ chlorine decay work reported in the literature [6,9,10].
of a number of pipes in distribution.
Investigation into the in situ chlorine decay is still
limited by the practical considerations such as estima- 2. Materials and methods
tion of travel times, identification of suitable sites and
limits on the time to conduct such a survey. Most 2.1. Measurement of in situ wall decay
studies, therefore, consider only a small number of pipe
lengths and focus on a small number of related The in situ wall decay rate between any two points A
parameters. Whilst in situ decay studies have advantages and B was measured [3] as follows:
in being directly applicable to distribution, potential
errors associated with the measurement may be large. 1. The bulk decay rate constant was measured at point
The decay rate is effectively calculated by observing A.
the difference in chlorine concentration between two 2. The average upstream or initial chlorine concentra-
points and estimating the time of travel of the water tion, C0 ; was measured for a slug of water at point A
between those points. Errors are, therefore, associated over E20 min.
with the following [3]: 3. The travel time, t; was calculated from point A to
point B for that slug of water. This was obtained
* Chlorine measurement: 70.01 mg/l by double Hach either from flow meter records or from tracer studies.
procedure. 4. The downstream or final chlorine concentration, C;
* Travel time measurement: 715% for flow meters; was measured as the slug of water arrived at point B.
75% for tracer studies. An average over E20 min was taken to ensure that
* Bulk measurement: 70.01 l/h. the correct slug of water was measured. If a tracer
A further error may be associated with iron pipes, study was used to calculate the travel time, then the
which experience corrosion, reducing their effective period immediately prior to the trace being observed
internal diameter. This was estimated as
30% by was used as the final chlorine concentration.
Powell [3]. 5. The overall and wall first-order decay rate constants
The pipe wall decay of chlorine has been measured in were then calculated from
carefully controlled laboratory experiments, which aim
k ¼ 1=t  ln C=C0 ;
at minimising the potential errors. Times corresponding
to travel times can be carefully chosen, sample waters kw ¼ k
kb :
with little bulk decay can be used, the state of the
internal surface of the test section of pipe can be
Sites suitable for measurement of in situ wall decay
observed and chlorine concentrations are easier to
were based on the following criteria:
measure in laboratory conditions. The drawbacks of a
laboratory study are that results may be less applicable * proximity to a chlorination point to ensure sufficient
and are often performed with new sections of pipe and chlorine to measure;
under poor hydraulic regimes. The impact of parameters * sufficient length to observe measurable decay;
which influence the in situ decay is well documented and * uniform pipe material and diameter to aid the
have been shown to have significant effects on the interpretation of results; and
chlorine concentration in networks. The majority of * safe and accessible sampling locations.
research has been carried out using the first-order decay
rate equation and the factors which have been shown to
influence the wall decay are: Flow meters could only be used where an accurate
meter was directly connected to the surveyed pipe or to
* pipe material and diameter [4–6];
all connecting pipework, where all demands off the
* initial chlorine concentration [1];
surveyed section were metered or were negligible and
* corrosion [1,7]; and
where the internal pipe diameter was accurately known.
* biofilm [7,8].
Tracer studies could only be used where a chlorine
The aim of this study was to consider wall decay as dosing unit or a flow meter where chlorine could be
measured both in situ and under laboratory conditions added was near to the surveyed section.
 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488 3481

When measuring the in situ wall decay rate, the bulk fitted between a PVC base plate and a perspex, conical-
demand was subtracted from the overall decay. To shaped lid. The instrument was fitted with an inlet valve
ensure a low bulk demand, the quarter strength Ringer’s entering towards the base, an outlet valve and an air
solution was used. Three bulk decay tests were carried release valve towards the top and a magnetic stirring
out on Ringer’s solution dosed with sodium hypochlor- facility (Fig. 2). The procedure for the test was as
ite and it showed a bulk decay rate (kb ) of 0.006 h
1 or follows:
less. This was deemed acceptable, being o6% of the
measured average wall decay of 0.10 h
1 and less than 1. The instrument’s base and lid were treated so that
the minimum recorded decay constant of 0.01 h
1. they were demand free by placing them in distilled
All the in situ tests were conducted on the Melbourne water dosed to 10 mg/l free chlorine with concen-
fed, Leicester distribution system (Fig. 1) which supplies trated hypochlorite for 24 h, before rinsing thor-
a total of 1.1 million customers using a lowland river as oughly with distilled water. This is the same
its source water. Between June 1996 and March 1999, 54 procedure as used for the cleaning of equipment for
in situ decay rate measurements were undertaken from a the bulk decay rate.
total of 11 pipe sections (Table 1). 2. A 10-l drum of quarter strength Ringer’s solution
was dosed to the desired chlorine concentration with
2.2. Measurement of laboratory-based wall decay concentrated hypochlorite solution and left to stand
for 10 min to ensure homogeneity.
Laboratory-based wall demand tests were conducted 3. The demand instrument was filled by gravity through
on 450 mm length, 100 mm diameter sections of pipe the inlet valve. As the instrument filled, the stirrer

8 Ab Kettleby
7 6

Old Dalby
10
9 1
5
4
Melbourne WTW

Ragdale 2

Approximate Scale (km):

Scalford
0 2 4
Road

Stoughton

1 Ab Kettleby to Melton
2 Melton to Ashfordby
3 Ab Kettleby to Long Clawson
11
4 Ab Kettleby to Saxelby 1
5 Saxelby 1 to Saxelby 2
Oadby Tank
6 Old Dalby 1 to Old Dalby 2
2 SR
7 Willoughby 1 to Willoughby 2
8 Willoughby 2 to Willoughby 3
Service Reservoir
9 Willoughby 3 to Willoughby 4
In Situ Pipe Survey
10 Willoughby 4 to Willoughby 5
Chlorination Point
11 Great Glen 1 to Great Glen 2

Fig. 1. Schematic of the in situ wall decay test locations (all fed from Ragdale service reservoir).
3482 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488

Table 1
Summary of the in situ wall decay tests

Location Pipe characteristics

No. of tests Diameter (mm) Length (m) Material Year laid

Ab Kettlby/Melton 7 298 3505 DICL Relined 1991

Melton/Ashfordby 7 180 2610 MDPE 1985
Ab Kettlby/Long Clawson 6 150 2300 PVC 1979
Ab Kettlby/Saxelby 1 5 200 1030 SI 1959
Saxelby 1/Saxelby 2 5 200 2284 SI 1959
Old Dalby Res./Old Dalby 3 150 2022 CI Pre-1960
Willoughby 1/Willoughby 2 6 150 2165 PVC 1973
Willoughby 2/Willoughby 3 7 100 925 CI 1960
Willoughby 3/Willoughby 4 4 100 515 CI 1960
Willoughby 4/Willoughby 5 2 100 305 CI 1960
Great Glen 1/Great Glen 2 2 200 865 PVC 1991

Air release valve

Outlet

Pipe section
100mm diam. x 450mm

Inlet from header tank

Magnetic stirrer
Fig. 2. Schematic diagram of the laboratory-based wall decay monitor.

was turned on to ensure continuous renewal of water samples were taken at times to ensure a regular
at the pipe surface. proportional drop in chlorine.
4. As the instrument became full, a certain volume
(20 ml) of water was run to waste before the first
sample was taken from the outlet valve using the 78% of the tests were carried out on pipe sections,
double Hach procedure [11]. All water released was which had not been in the system and were therefore
replaced from the original drum by gravity. considered ‘new’. They were treated, prior to testing, by
5. The air valve was opened and a small volume of submerging in distilled water dosed to 10 mg/l free
water released (o5 ml) to allow any trapped air to chlorine with hypochlorite solution for 48 h and rinsed
escape. with distilled water. A further 20% of the tests were
6. Further samples were taken in the same manner at carried out on ‘new’ pipe sections, which had been
intervals throughout the period of the test. Ideally, placed within the system with water passing through for
 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488 3483

a period of 3 weeks in order to produce a biofilm. The pipework diameters and friction factors are known.
pipe was placed into the instrument while the surface Tracer studies offer the flexibility to determine travel
was still wet, ensuring that the state of any biofilm was times to virtually any point in the system.
not disturbed. The remainder of the tests (2%) were Chlorine tracing was carried out using the following
carried out on pipe sections which had been within the methodology:
system for a number of years. These were cut out and
the biofilm, which had dried during transportation, was 1. Either chlorine was injected into the network to
rinsed with distilled water to moisten it before the produce a spike of chlorinated water. The spike was
section was placed in the demand instrument. The in the range of 0.5–1.0 mg/l free chlorine for a
authors accept that this technique disturbs the inside of duration of 15 min. This was achieved either by
the pipe to some degree. However, measurement on site increasing the target on a nearby chlorination
was considered impractical. equipment or by connecting an emergency-dosing
Six pairs of laboratory-based wall decay tests were equipment to the mains via a universal cock where a
carried out on the same section of pipe under the same flow meter existed.
conditions of initial chlorine dose and temperature to Or the chlorine-dosing equipment was turned off
determine the reproducibility of the tests. Results for a period of 30 min producing a significant dip in
showed that the standard deviation of the errors was free chlorine concentration. If the usual chlorine
70.013 h
1 which, when analysed assuming a Student’s concentrations would allow, this method avoided
t distribution, gave a standard error of 70.037 h
1 and using any high concentration of chlorine in the
95% confidence limits of 70.016 h
1. system.
A number of differences exist between the laboratory 2. Hydrant sampling was implemented downstream of
model and the field tests, which affect the associated the chlorine injection sites, either with a continuous
errors. In the laboratory studies: monitor (EIT Model 5150), or with the personnel
continually taking spot samples. The chlorine spikes,
* In the majority (98%) of cases, new pipes were used, or dips, were, thus, followed down a length of main
eliminating the impact of the age and condition of the into the system. Hydrant flows were set to 6 l/min or
pipes. less in order to minimise disruption to the natural
* The hydraulic regime was only an approximation of flow in the main.
system conditions. The magnetic stirrer used to 3. The travel time between two points was calculated as
ensure water movement was temperamental. How- the period between the rising limb of the spike and
ever, tests concluded that its use had no significant falling limb of the dip. This was chosen so that the
impact on wall decay. equipment could, immediately, be moved down-
* Water samples were taken from the top of the stream if circumstances necessitated.
monitor through a sample line away from the pipe
wall. Samples were, therefore, not taken at the wall
but from the bulk liquid. 3. Results and discussion
* Initial chlorine concentrations in the laboratory
experiment were greater than observed in distribution The in situ wall decay constants varied from 0.00 to
systems. Higher concentrations ensured that any 1.64 h
1. 70% of the constants were o0.4 h
1 and an
small errors from the pocket colorimeters were examination of Fig. 3 shows that most of the sites
minimized. surveyed had low decay constants. However, five sites
* Water temperature of the tests averaged 191C, were associated with some higher (>0.4 h
1) values and
warmer than under most field conditions. two sites; Willoughby 3/Willoughby 4 and Old Dalby
* Laboratory tests were conducted over a longer time Reservoir/Old Dalby; gave decay constants which were
period than field tests in order to ensure that the all >0.4 h
1. Both of these were cast iron (CI) mains,
majority of chlorine in the water had decayed. the former with a nominal diameter of 100 mm, the
latter with one of 150 mm.
2.3. Tracer studies The laboratory experiments were conducted on new
polyvinyl chloride (PVC), cement lined cast iron (CICL)
Good hydraulic information is vital in considering and medium density polyethylene (MDPE) pipe sec-
chlorine decay within a system in order to gather tions. The wall decay constants ranged from 0.01 h
1 to
realistic travel time information. Flow meters within a 0.78 l/h (Fig. 4). These results showed a slightly greater
system are often limited as they are expensive to install range of decay rates than the in situ tests on the same
and their accuracy is often difficult to assess. Pressure materials, which ranged from 0.00 to 0.26 h
1.
monitoring provides an alternative. However, this is The average in situ decay of CI and spun iron
only accurate where pressure gradients are high and (SI) pipes surveyed was significantly greater than the
3484 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488

cement-lined ductile iron (DICL), PVC and MDPE (0.05 h
1) mains, though still significantly lower than
pipework (Fig. 5). Chlorine is a strong oxidising agent, unlined iron pipes. Therefore, the pipe reactivity, as
which reacts readily with iron. However, the variability judged by the wall decay constants, was found to be
in the cast iron decay constants was large, varying CI>SI>DICL>PVC>MDPE. It is clear that the
between 0.03 and 1.64 h
1. This variation will, in part, cement provides a layer of protection for the iron
be related to the condition of the pipe itself. The average against attack by chlorine. Previous in situ studies have
wall decay of the DICL (0.13 h
1) main was slightly also shown that unlined cast iron pipes have decay rates
higher than those for the PVC (0.09 h
1) or the MDPE between 4 and 100 times greater than lined or plastic
pipes [1].
For the laboratory studies, the pipes were separated
into relatively reactive pipes (CI and SI) and relatively
1.8 unreactive pipes (MDPE, PVC and CICL). Fig. 6 shows
that laboratory experiments produced similar results to
the field tests, suggesting a good relationship between
1.6
the two methods of measuring wall decay. The average
wall decay as measured in laboratory and field tests for
1.4 the different materials, respectively, were 0.12 and

1.2
Wall decay constant (l/h)

50
1
45

0.8 40

35
Frequency (%)

0.6 30

25
0.4
20

15
0.2
10

5
0
0 1 2 3 4 5 6 7 8 9 10 11 0
Sites <0.05 0.15-0.20 0.30-0.35 0.45-0.50 0.60-0.65 0.75-0.80
Wall decay constant (l/h)
Fig. 3. In situ wall decay constants at the various sites (as
numbered in Fig. 1). Fig. 4. Variation in the laboratory wall decay constant.

10 .000
Average 0.67 Average 0.33 Average 0.13 Average 0.05 Average 0.09
Wall Chlorine Decay Constant (hr -1)

1.00 0

0.10 0

0.01 0

0.00 1
CI SI DICL MDPE PVC
Pipe Material

Fig. 5. Effect of pipe material on wall decay constant kw :

 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488 3485

0.6

0.5
Wall decay constant (l/h)

0.4

0.3

0.2

0.1

0
0 1 2 3
Pipe material
Fig. 6. The impact of pipe material on the laboratory-measured wall decay constant (1—CICL; 2—MDPE; 3—PVC).

0.13 h
1 (CICL and DICL), 0.09 and 0.09 h
1 (PVC) 0.3 mg/l. Similar relationships were not evident for
and 0.05 and 0.05 h
1 (MDPE). Laboratory studies have MDPE, DICL or iron pipes. It is possible that the
shown that new plastic pipes have almost no chlorine small range of initial chlorine concentrations studied
demand, other than a small initial demand on first was responsible. Alternatively, there might have been
contact with chlorine, which would be satisfied very insufficient time for stabilisation of the wall decay
quickly in the system [5]. constant to occur, as indicated by the work reported
The bulk chlorine decay was observed to have an previously [1]. If either of the above explanations is true,
inverse relationship with the initial chlorine concentra- however, it does pose the question why was the
tion. A similar inverse relationship with wall decay on relationship evident for PVC pipes?
one pipe stretch dosed to between 0.3 and 5.0 mg/l has In the laboratory experiments, with samples being
been reported by AWWARF [1] but stabilisation of the dosed to initial chlorine concentrations of between 0.16
wall decay, following a change in initial chlorine and 1.95 mg/l, relationships between initial chlorine
concentration, took E1 month. In the current study, a concentration and wall decay constants were also
strong relationship between initial chlorine concentra- observed (Fig. 8). A linear function was not found to
tion and wall decay for all PVC pipes surveyed in situ fit the data adequately. A power function provided the
was found. The regression equation shown in Fig. 7 is best fit when applied to each material type (Table 2). The
y ¼ 0:26
0:73x and analyses showed that the relation- relationships were all significant at the 99% level. The
ship was significant at the 99% level. This suggests that inverse relationship suggests that there is a limiting rate
the wall decay constant would be zero at initial chlorine at which the chlorine can be taken up by the pipe wall,
concentrations >0.36 mg/l. The logical proposition hence higher chlorine concentrations yield lower kw
would be that wall decay would tend to, but not reach, values. Unlined iron pipes, with a highly reactive
zero. However, neither exponential nor power functions surface, would not be expected to be limited in the
adequately fitted this data. The linear function proposed same way, hence it is not unreasonable that no such
in Fig. 7 should, therefore, only be considered represen- relationship was found with the in situ measurements
tative between initial chlorine concentrations of 0.15 and under this type of pipe. No such comparison was
3486 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488

0.25 1.8

1.6

0.2 1.4

Wall Decay Constant (l/h)

Wall decay constant (l/h)

1.2

0.15 1

0.8

0.1 0.6

0.4
0.05
0.2

0
0 0 0.1 0.2 0.3 0.4 0.5
0 0.1 0.2 0.3 0.4 Velocity (m/s)

Initial chlorine concentration (mg/l) Fig. 9. Relationship between velocity and the in situ wall decay
Fig. 7. Relationship between the initial chlorine concentration constants for SI (K) and CI (~) pipes.
and the in situ wall decay constant for PVC pipes.

0.9 either turbulence or diffusion, was rejected because,

0.8 under high chlorine concentrations, a greater concentra-
tion difference would exist between the pipe center and
0.7
the pipe wall where the majority of decay was occurring.
Wall Decay Constant (l/h)

0.6 This would increase the driving force controlling the rate
0.5
at which chlorine would diffuse to the pipe wall. Hence,
it would be expected that greater chlorine concentrations
0.4 would increase, not decrease, transport to the pipe wall
0.3 and in turn decay. Turbulence would be the same under
any chlorine concentration.
0.2
Flow velocity can affect turbulence, diffusion and the
0.1 boundary layer thickness, all of which influence the rate
0 at which chlorine is transferred to the pipe wall. A
0 0 .5 1 1 .5 2 positive relationship would be expected where transfer
Initial chlorine concentration (mg/l) of chlorine to the pipe wall is limiting as velocity
increases the rate at which chlorine would be trans-
Fig. 8. Relationships for initial chlorine concentration and the
ferred. Fig. 9 shows that there was a linear relationship
laboratory-measured wall decay constant for CICL ( ),
MDPE ( ) and PVC ( ). between flow velocity and the in situ decay constants for
unlined iron pipes. This relationship was significant at
the 99% level. This linear equation suggests that if velocity
Table 2 were reduced below 0.15 m/s in unlined iron pipes, wall
Regression equations for the relationships between initial decay would be eliminated. This is unrealistic, as some
chlorine concentrations and the wall decay constant for transfer of chlorine would still occur resulting in positive
different pipe materials wall decay. A linear response is, therefore, perhaps
Pipe material Regression equation R2 fortuitous. However, neither an exponential nor a power
function was found to represent the data adequately. This
CICL y ¼ 0:05x
1:17 0.48 relationship should, therefore, only be considered repre-
PVC y ¼ 0:02x
1:55 0.67
sentative of velocities between 0.2 and 0.45 m/s.
MDPE y ¼ 0:02x
1:75 0.52
A similar positive relationship was found for PVC,
MDPE and DICL pipes. However, the fit was poor
(R2 ¼ 0:26), as might have been be expected with
conducted under laboratory conditions as no pipes were unreactive pipes where the pipe reactivity and not
available for testing. transfer of chlorine would be rate limiting.
The hypothesis that the rate at which chlorine can be The flow in all the pipes surveyed was in the
transferred to the pipe wall was limiting decay, due to transitional region of the Moody diagram, with the
 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488 3487

Reynolds number ranging by an order of magnitude present on the pipe surface. Only one of the samples
from 1.7  104 to 1.7  105. showed a decay constant, which was outside the range of
The relationships determined between the Reynolds values obtained when no biofilm was present. These
numbers and the decay constants indicated that results, perhaps, go some way to explaining why TOC
turbulence had a similar effect on wall decay as on had no influence on wall decay. A larger amount of
velocity. Generally, the relationships were poor, parti- biofilm, as has been reported by some authors [13,14],
cularly for SI, PVC, MDPE and DICL pipes. may have a more significant effect on chlorine decay.
Pipe diameter is inversely proportional to the surface
to volume ratio. The higher the surface to volume ratio,
the larger the surface area across which mass transfer 4. Conclusions
can occur and the greater the number of reactive sites
available for each unit volume of water passing down * An acceptable method for measuring the in situ
the pipe. Previous research has observed decay rates to chlorine decay in distribution was developed. A
decrease with pipe diameter [5,6]. Measuring the impact laboratory experiment to measure the wall decay
of diameter on in situ chlorine decay was difficult, as, gave results representative of the distribution system.
ideally, pipe stretches with differing diameters but * The pipe material has a strong influence on the wall
otherwise identical properties were needed. As can be decay in distribution. Pipes can generally be classified
seen from Table 1, these criteria did not exist in the into reactive pipes (unlined iron) and unreactive pipes
survey area. No relationship was evident for any pipe (PVC, MDPE, cement-lined iron).
type, although only a small range of pipe diameters was * Results suggest that the wall decay of unreactive
considered. pipes is limited by the reactivity of the pipe while
Temperature will affect a range of parameters, which reactive pipes are limited by the transfer of chlorine
can influence chlorine decay. The data from the surveys to the pipe wall.
suggested that a higher wall decay did occur at higher * An inverse relationship was found over a small
temperatures, but the relationships were such that no observed C0 range (0.15–0.3 mg/l) for in situ PVC
conclusions could be drawn. The TOC data showed a pipes and for unreactive pipes in laboratory tests over
considerable amount of scatter and, again, no conclu- a wider range of C0 (0.16–1.95 mg/l).
sions could be drawn about its effect on the wall decay * A positive relationship existed between flow velocity
constants. and wall decay for all pipe sections surveyed. The
A biofilm, which develops on the internal surface of a relationship between Reynolds number and wall
pipe may create a chlorine demand itself or it may decay is less significant
protect a surface from chlorine penetration, thus * Relationships were not observed between wall decay
reducing decay. and the parameters of TOC, pipe diameter and
The only previous research on the chlorine demand of temperature. There is a suggestion, however, that
a biofilm was conducted by Lu et al. [8] and Kiene et al. increased temperatures may make the decay more
[7], who developed a biofilm on 4 mm polystyrene beads variable.
and measured the resulting demand. The biofilm was * A moderate amount of biofilm (51–670 pg ATP/cm2)
shown to exert a significant but small chlorine demand on the wall of unreactive pipes did not influence the
of E0.11–0.18 h
1 for colonized beads and 0.10 h
1 for wall decay rate.
uncolonized beads. No measure of the biofilm quantity
on the beads was given. A limitation of this work is that
it considers the biofilm grown on polystyrene beads and
Acknowledgements
not on the actual sections of pipe and ignores the
interaction between the material and biofilm. The
The authors would like to acknowledge the financial
laboratory monitor was, therefore, used to measure the
support given by Severn Trent Water Ltd.
chlorine decay due to a biofilm in contact with a pipe
surface.
The same pipe sections used to assess the effect of pipe
material on the the chlorine demand (PVC, MDPE and References
CICL) were placed within parts of the distribution
[1] AWWARF. Characteristics and modelling of chlorine
system as vertical pipe reactors [12] and a biofilm were
decay in distribution systems. USA: AWWA, 1996.
allowed to develop. The amount of biofilm on the pipe [2] Powell JC, Hallam NB, West JR, Forster CF, Simms J.
sections measured for chlorine decay ranged from 51 to Factors which control bulk chlorine decay rates. Water
670 pg ATP/cm2. This moderate quantity of biofilm did Res 2000;34:117–26.
not significantly influence the decay of chlorine at the [3] Powell JC. Modelling chlorine in water distribution
pipe wall. 12 samples were tested when biofilm was networks. Ph.D. thesis, University of Birmingham, 1998.
3488 N.B. Hallam et al. / Water Research 36 (2002) 3479–3488

[4] Brammer LF, Tansley NS. Chlorine residual modelling in network and applications to Paris distribution network.
distribution: the improvement of taste and the mainte- Proceedings of the AWWARF and USEPA Conference
nance of effective disinfection. International Conference on Water Quality Modelling in Distribution Systems,
on Computer Applications for Water Supply and Dis- Cincinatti, Ohio, USA, 4–5 February, AWWA, 1991.
tribution, De Montford University, Leicester, UK, 7–9 p. 77–87.
September, 1993. [10] Zhang GR, Kiene L, Wable O, Chan US, Duguet JP.
[5] Gotoh K. Residual chlorine decreasing rate coefficients for Modelling of chlorine residual in the water distribution
various pipe materials. Water Supply 1989;7:164–72. system of Macao. Environ Technol 1992;13(10):937–46.
[6] Sharp WW, Pfeffer J, Morgan M. In situ chlorine decay [11] Powell JC, West JR, Hallam NB, Forster CF, Simms J.
rate testing. Proceedings of the AWWARF and USEPA Performance of various kinetic models for chlorine
Conference on Water Quality Modelling in Distribution decay. ASCE-J Water Resources Plann Manage 2000;126:
Systems, Cincinatti, Ohio, USA, 4–5 February, AWWA, 13–20.
1991. p. 311–22. [12] Hallam NB, West JR, Jago P, Forster CF, Simms J. The
[7] Kiene L, Lu W, Levi Y. Relative importance of the potential for biofilm growth in water distribution systems.
phenomena responsible for chlorine decay in drinking Water Res 2001;35:4063–71.
water distribution systems. Water Sci Technol [13] LeChevallier MW, Babcock TW, Lee RG. Examination
1998;38:219–27. and characterisation of distribution system biofilms. Appl
[8] Lu W, Kiene L, Levi Y. Chlorine demand in water Env Microbiol 1987;53:2714–24.
distribution systems. Water Res 1999;33:827–35. [14] Pederson K. Biofilm development on stainless steel
[9] Wable O, Dumoutier N, Duguet JP, Jarrige PA, Gelas G, and PVC surfaces in drinking water. Water Res 1990;24:
Depierre JF. Modelling chlorine concentrations in a 239–43.