This article was downloaded by: [41.71.155.

183]
On: 04 January 2013, At: 01:59
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Urban Water Journal

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/nurw20

The impact of variable hydraulic operation of water

distribution networks on disinfection by-product
concentrations
a a a a
Emesomake Idornigie , Michael R. Templeton , Cedo Maksimovic & Suzan Sharifan
a
Department of Civil and Environmental Engineering, Imperial College London, London, UK
Version of record first published: 21 Sep 2010.

To cite this article: Emesomake Idornigie , Michael R. Templeton , Cedo Maksimovic & Suzan Sharifan (2010): The impact of
variable hydraulic operation of water distribution networks on disinfection by-product concentrations, Urban Water Journal,
7:5, 301-307

To link to this article: http://dx.doi.org/10.1080/1573062X.2010.509438

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
 Urban Water Journal
Vol. 7, No. 5, October 2010, 301307

RESEARCH ARTICLE
The impact of variable hydraulic operation of water distribution networks on disinfection
by-product concentrations
Emesomake Idornigie, Michael R. Templeton*, Cedo Maksimovic and Suzan Sharifan
Department of Civil and Environmental Engineering, Imperial College London, London, UK
(Received 26 January 2010; nal version received 15 July 2010)

This study assessed the magnitude of variations in the concentrations of disinfection by-products, specically
trihalomethanes and haloacetic acids, which occur in two UK water distribution networks of dierent sizes due to
changes in the hydraulic operation of the networks. These operational changes included varying the ll levels of
water storage tanks and pumping conditions within the network. This was investigated through modelling which
combined a conventional distribution network ow model (EPANET) with previously developed models for
disinfection by-product formation, and using network information obtained from water company partners. The
modelling demonstrated that there is the potential for signicant variations in disinfection by-product
Downloaded by [41.71.155.183] at 01:59 04 January 2013

concentrations, up to a 40% change in some simulations, due to what may be considered routine variability in
the hydraulic operation of the network. Smaller networks may be especially susceptible to these variations. These
ndings may inuence how water companies plan disinfection by-product monitoring programmes for regulatory
reporting.
Keywords: drinking water; distribution network; disinfection by-product; hydraulic modelling

Introduction matter content, and bromide concentration (Sohn et al.

Disinfection by-products (DBPs) are formed from the
reaction of disinfectants (e.g., chlorine, ozone) with
natural organic matter and other naturally occurring
chemical species (e.g., bromide). Due to the potential
human toxicity of certain DBPs (e.g., Nieuwenhuijsen
et al. 2000, 2008) there are regulatory limits on the
concentrations of certain groups of DBPs in tap water
in many countries. For example, total trihalomethanes
(TTHMs) are regulated at a maximum concentration
of 100 mg/l in tap water in the UK currently, and UK
water companies must periodically collect samples
from their distribution networks and report the
concentrations of these compounds. In the US, the
Stage 2 Disinfectants/Disinfection By-Products Rule
of the Environmental Protection Agency stipulates
the calculation of locational running annual averages
(LRAA) for TTHMs and also haloacetic acids
(USEPA 2006). The LRAA requirement recognises
that DBP concentrations will vary within a distribution
network and the Rule therefore species that samples
should be taken from locations in the network that are
anticipated to have the highest DBP concentrations.
DBP formation is a function of a number of water
quality variables, including pH, temperature, organic
2001). The formation of DBPs also depends on
operational factors, specically the water age (i.e. the
time that the water spends in transit between the treat-
ment works and the consumer tap) and the chlorine
residual (Clark et al. 1994, Clark and Sivaganesan 1998,
Clark et al. 2010). These latter two variables are directly
linked to the hydraulic operation of the network, since
water age will vary with ow and chlorine decays over
time (Biswas et al. 1993, Powell et al. 2000, Courtis et al.
2009a,b).
While previous studies have shown that DBP
concentrations vary spatially and seasonally in dis-
tribution networks (e.g., Sadiq and Rodriguez 2004,
Courtis et al. 2009b), few have considered short-term
temporal variations in DBP concentrations, i.e. on the
order of minutes or hours, that may occur due to
routine hydraulic instabilities in the network. These
short-term variations in DBP concentrations are
unlikely to be a concern from a health standpoint,
since the toxicity of most DBPs is based on chronic
rather than acute exposure when considering the DBP
levels that are typical of drinking water (Nieuwenhuij-
sen et al. 2008). However, short-term variations could
be important for DBP sampling programme planning

*Corresponding author. Email: m.templeton@imperial.ac.uk

ISSN 1573-062X print/ISSN 1744-9006 online

2010 Taylor & Francis
DOI: 10.1080/1573062X.2010.509438
http://www.informaworld.com
 302 E. Idornigie et al.
for regulatory reporting by water utilities and also
simply in terms of gaining an improved understanding
the variability in physicochemical water quality that is
inherent in a network.
Short-term variations in DBPs are hypothesised to
arise due to a range of dierent types of hydraulic
instabilities in water distribution networks. Water
networks are rarely operated at steady-state (Bargiela
and Hainsworth 1989, Al-Omari and Hanif Chaudhry
2001); hydraulic variations may occur in response to
water demand and also due to operational variations
in the network, such as water storage tanks operating
at dierent ll levels, sections of the network being
emptied for cleaning or during re-hydrant usage, or
routine on/o cycles of pumps, e.g., in response to
demand. The objective of this research study therefore
was to investigate the magnitude of changes in
disinfection by-product concentrations caused by these
Downloaded by [41.71.155.183] at 01:59 04 January 2013
hydraulic instabilities and to consider the relative
importance of these variations for dierent sizes of
distribution networks.
Methodology
Network data and information were obtained from
water company partners for two UK water distribu-
tion networks, to be referred to as Network A and
Network B. Network B was a larger network in
Figure 1. Network A considered in this study.
terms of the total length of pipes (Figures 1 and 2,
Table 1).
EPANET software (Rossman, 2009) was used to
model the water ages and chlorine residuals in each
network. Data obtained from the water company
partners for the network volumes and pump curves
dened the tank volume levels and discharge-head
relationships to use as a baseline representation of the
current operational practice. Patterns for daily demand
variations and pump operations with respect to time
were also dened.
Two DBP formation models were used to predict
DBP concentrations in the network. The rst was the
modelling approach developed by Gang et al. (2002),
shown below:
k k t
Rt S
Ct Co ffe 1  f e g
DBP DC  Co
where C(t) chlorine concentration (mg/l) at any
time t; Co the initial chlorine dose to give a chlorine
residual of approximately 1 mg/l after 5 days of
reaction; f fraction of chlorine dose attributed to
rapid reactions; kR rst order rate constant for
rapid reactions (0.5 day71); ks rst order rate
constant of slow reactions (0.05 day71); D DBP
yield coecient, which is dened as the ratio of
 Urban Water Journal 303

where TOC total organic carbon concentration

Figure 2. Network B considered in this study.
Downloaded by [41.71.155.183] at 01:59 04 January 2013
(mg/l); Cl2 initial chlorine dose (mg/l); T tempe-
rature (8C); t water age (h). For the purposes of this
modelling, the following water quality values were
applied to each network: TOC 2.5 mg/l, pH 7.5,
temperature 108C, and bromide 0.05 mg/l. An initial
chlorine dose, leaving the treatment works, of 1 mg/l
was assumed for each network. The same water quality
parameters and initial chlorine dose were applied to
both networks to remove these as confounding factors
and allow consideration of solely the hydraulic eects
on the magnitude of the DBP concentration variations.
A large number of DBP formation models have
been developed (Sadiq and Rodriguez 2004), but the
intention of using two models in this study (rather than
just one) was to compare the prediction outputs
from using dierent DBP formation models. The
water age and chlorine residual for each node in the

networks, obtained from the EPANET modelling,

were input into each of the DBP formation models to
obtain the DBP concentrations at each node in the
Table 1. Summary of physical characteristics of the two
studied networks. networks.
Earlier studies have showed that the application of
Network Parameter Network A Network B this modelling approach provides reasonably accurate
Number of junctions 336 5871 predictions of DBP concentration values when com-
Number of tanks 2 9 pared to eld sampling data (e.g., Sohn et al. 2001,
Number of pipes 345 4496 Sadiq and Rodriguez 2004). However, the main
Number of pumps 6 6 purpose of this study was to investigate the magnitude
Number of valves 2 1552
Total length of pipes (km) 84 548 of variations in the DBP concentrations under dierent
hydraulic conditions, rather than the prediction of the
concentrations themselves.
the concentration (mg/l) of DBP formed to the It should also be noted that the concentrations of
concentration of chlorine consumed (mg/l). The yield TTHMs and nine haloacetic acids (HAA9) modelled
coecients of 30 mg TTHM/mg Cl2 and 17 mg HAA/ in this study only accounted for DBP formation within
mg Cl2 described in Gang et al. (2002) were used in this the distribution network and did not take into account
study. Using default values was felt to be acceptable the formation in the treatment works prior to
for the purposes of this study, where the primary distribution. Therefore, the DBP values reported
objective was to observe the magnitude of variations in below are lower than would be measured in eld
DBP concentrations due to short-term hydraulic samples.
variability, rather than accurate prediction of the
DBP concentration values; if the latter had been the
primary focus, then the chlorine decay coecients and Results and discussion
DBP yield coecients would have been measured for Figure 3 illustrates the degree of variation in TTHM
each of the particular water matrices for the two concentrations based on current hydraulic variations
networks considered in the study. in the networks (i.e. current water demand patterns).
The second DBP formation model that was used in As expected, TTHM concentrations were higher for
this study was developed by Serodes et al. (2003) for node locations that were further away from the
THM and HAA concentration prediction, shown treatment works in the network, due to increased
below: water age and hence increased DBP formation time.
DBP variations in Network A, the smaller of the two
THMs 16:9 16:0TOC 3:319Cl2  1:1135T networks, were more dramatic than in Network B, in
terms of percent variation (i.e. deviation from the
1:139t
mean DBP concentration).
Pump on/o cycles were expected to have an eect
HAAs 8:202 4:869TOC 1:053Cl2 0:64t on TTHM concentrations at downstream locations by
 304 E. Idornigie et al.
aecting the water age patterns. Figure 4 illustrates our
analysis of this for Network B, where Node A5D
represents a location upstream of the pump and Node
Downloaded by [41.71.155.183] at 01:59 04 January 2013
Figure 3. Variations in TTHM concentrations in Network
A (a) and Network B (b) based on current hydraulic
operation practice of the networks.
Figure 4. Impact of on/o pumping cycles on TTHM concentrations at two nodes in Network B. Node A5D was located
upstream of the pump and Node A74 downstream of the pump.
A74 was downstream of the pump. Both nodes were
located near to a storage tank and TTHM concentra-
tions were also simultaneously aected by the regular
storage tank discharge cycles. There was a noticeable
disparity between the TTHM variations at these nodes
(i.e. upstream versus downstream of the pump), with a
steady increase in TTHM concentrations in Node A74
which coincided directly with the pump on/o cycles.
However, the TTHM concentration patterns for the
two nodes essentially overlapped otherwise, suggesting
that storage tank operating levels have a more
pronounced impact on DBP levels than individual
pump cycles.
To investigate the eect of varying storage tank
operating levels on DBP concentrations, the water
levels in a storage tank in Network A (shown with an
asterisk in Figure 1) were intentionally varied over a
narrow range, i.e. a 1-m dierence between maximum
and minimum water level in the tank, and a wider
range, i.e. a 3-m dierence between maximum and
minimum water level in the tank. The current
operating scenario was also considered, with a range
of tank operating levels of approximately 1.5 m. The
resulting variations in TTHM and HAA concentra-
tions in Network A at Node 312, which was located far
from the manipulated storage tank, and at Node 46,
located nearer the storage tank, were then calculated,
using the Serodes et al. (2003) DBP formation models
(Figure 5). In addition, Figure 6 shows the percent
variations in TTHM concentrations (i.e. percent
dierence in TTHM concentrations from the mean
concentration value for that node) in Network A at
Node 46 for the wide tank level operating range
scenario (Scenario 1) and the narrow tank level
 Urban Water Journal 305
Downloaded by [41.71.155.183] at 01:59 04 January 2013

Figure 5. Variations in TTHM concentrations (a) and HAA concentrations (b) in Network A at Node 312 (far from the storage
tank), and variation in TTHM concentrations (c) and HAA concentrations (d) at Node 46 (near the storage tank), for the wide
tank level operating range scenario (Scenario 1), the narrow tank level operating range scenario (Scenario 2), and the base case,
using the Serodes et al. (2003) DBP formation models.

operating range scenario (Scenario 2). For the wide
tank level operating range and the node near the tank,
percent variations up to 40% were predicted, as either
an increase or decrease, depending on high or low tank
level, respectively. For the narrow water tank level
operating range the percent variations in TTHM
concentrations were only up to 30%, and usually less
than 15%, for the node near to the tank (Node 46).
For the node further from the tank (Node 312), the
same trends with regard to tank operating level were
observed, however the percent deviations in TTHM
concentrations were smaller.
HAA9 concentrations were lower than TTHM
concentrations but showed similar levels and patterns
of variability as discussed earlier for the TTHMs
(Figure 5). It should be noted that some HAAs are
known to degrade in distribution systems under certain
conditions (Bayless and Andrews 2008), and this was
not taken into account in the models used in this study;
it is uncertain to what degree this degradation could
aect the short-term temporal variability in HAA
concentrations. However, while this is a limitation that
would aect the precision of prediction of the absolute
values of DBP concentrations, it was not felt to be
essential to include it in the modelling for the purpose
of estimating the magnitude of variation in DBP
concentrations on short time scales.
The DBP variability predictions using the Gang
et al. (2002) and Serodes et al. (2003) models were also
compared. In general, the Serodes et al. (2003) model
predicted less variability than the Gang et al. (2002)
model. For example, the average percent variation in
TTHM concentrations in Network A for the scenario
of wide tank level operating range was 25% for the
Serodes et al. (2003) model versus 38% for the Gang
et al. (2002) model. This dierence can be explained by
the fact that the Serodes et al. (2003) is a multiple
linear regression model which takes into account the
relative impacts of various water quality parameters
(e.g., total organic carbon, bromide) in addition to
 306 E. Idornigie et al.
Downloaded by [41.71.155.183] at 01:59 04 January 2013
Figure 6. Percent variations from the mean in TTHM
concentrations in Network A at Node 46 for the wide tank
level operating range scenario (a) and the narrow tank level
operating range scenario (b).
water age and chlorine residual, whereas the Gang
et al. (2002) model is more directly inuenced by the
value for the chlorine residual, which is directly
dependent on water age (i.e. due to rst-order decay
in the bulk water).
Overall, the ndings of this study suggest that
short-term hydraulic variations within a distribution
network can have signicant eects on the concentra-
tions of DBPs which subsequently occur in the
distribution network. These short-term DBP variations
are likely not of importance to health in most cases,
even if the variability causes DBP concentrations to
briey exceed regulated limits by a small margin, since
DBP toxicity is only associated with chronic exposure
to elevated concentrations over extended periods.
However, this DBP variability is likely to be of
importance to water utility companies who must
collect water samples and report DBP levels
for regulatory purposes. This study suggests that
DBPs should be sampled from networks during periods
of worst-case DBP formation (i.e. peak DBP concen-
trations); in the context of the modelling that was
considered in this study, this would mean during
periods of longest water age e.g., storage tanks at
fullest level and pumping at the lowest normal
operating rate.
Conclusions
This modelling study showed that the concentrations
of disinfection by-products (e.g., trihalomethanes and
haloacetic acids) in drinking water distribution net-
works can vary signicantly due to changes in the
hydraulic operation of the network, such as dierent
water storage tank operating levels and pumping
conditions. For some of the investigated scenarios,
variations in DBP concentrations were as high as 40%
at some network nodes. The smaller of the two studied
networks was more susceptible to these variations.
Water utilities should therefore consider the distribu-
tion network hydraulic conditions that are likely to
lead to worst-case DBP concentrations in their specic
network context; normally this will be linked to
operating conditions which lead to extended water
age in the network (e.g., tanks at their fullest level,
lowest normal pumping rate).
Acknowledgements
The authors thank Julie Hart of Severn Trent Water and Sian
Taylor of United Utilities for supplying the case study
network information and data. The authors also acknowl-
edge Vijay Das, Joel Jardine, and Dr Yanping Zhang, who
helped with some of the initial modelling and data collection.
References
Al-Omari, A.S. and Hanif Chaudhry, M., 2001. Unsteady-
state inverse chlorine modelling in pipe networks. Journal
of Hydraulic Engineering, 127 (8), 669677.
Bargiela, A. and Hainsworth, G.D., 1989. Pressure and ow
uncertainty in water systems. Journal of Water Resources
Planning and Management, 115 (2), 212229.
Bayless, W. and Andrews, R.C., 2008. Biodegradation of six
haloacetic acids in drinking water. Journal of Water and
Health, 6 (1), 1522.
Biswas, P., Lu, C., and Clark, R.M., 1993. A model for
chlorine concentration decay in pipes. Water Research,
27 (12), 17151724.
Clark, R.M., Lykins, B.W., Block, J.C., Wymer, L.J., and
Reasoner, D.J., 1994. Water quality changes in a
simulated distribution system. Journal of Water Supply
Research and Technology AQUA, 43 (6), 263277.
Clark, R.M. and Sivaganesan, M., 1998. Predicting chlorine
residuals and formation of trihalomethanes in drinking
water. ASCE Journal of Environmental Engineering, 124
(12), 12031210.
Clark, R.M., Yang, Y.J., Impelliteri, C.A., Haught, R.C.,
Schupp, D.A., Panguluri, S., and Krishnan, E.R., 2010.
Chlorine fate and transport in distribution systems:
experimental and modelling studies. Journal of the
American Water Works Association, 102 (5), 144155.
Courtis, B.J., West, J.R., and Bridgeman, J., 2009a. Chlorine
demand-based predictive modelling of THM formation
in water distribution networks. Urban Water Journal, 6
(6), 407415.
 Urban Water Journal
Courtis, B.J., West, J.R., and Bridgeman, J., 2009b.
Temporal and spatial variations in bulk chlorine decay
within a water supply system. ASCE Journal of Environ-
mental Engineering, 135 (3), 147152.
Gang, D.D., Segar, R.L. Jr., Clevenger, T.E., and Banerji,
S.K., 2002. Using chlorine demand to predict THM and
HAA9 formation. Journal of the American Water Works
Association, 94 (10), 7686.
Nieuwenhuijsen, M.J., Toledano, M.B., Eaton, N.E., Fawell,
J., and Elliott, P., 2000. Chlorination disinfection
byproducts in water and their association with adverse
reproductive outcomes: A review. Occupational and
Environmental Medicine, 57 (2), 7385.
Nieuwenhuijsen, M.J., Toledano, M.B., Bennett, J., Best, N.,
Hambly, P., de Hoogh, C., Wellesley, D., Boyd, P.A.,
Abramsky, L., Dattani, N., Fawell, J., Briggs, D., Jarup,
L. and Elliott, P., 2008. Chlorination disinfection by-
products and risk of congenital anomalies in England
and Wales. Environmental Health Perspectives, 116 (2),
216222.
Downloaded by [41.71.155.183] at 01:59 04 January 2013
307
Powell, J.C., Hallam, N.B., and West, J.R., 2000. Perfor-
mance of various kinetic models for chlorine decay.
Journal of Water Resources Planning and Management,
Jan/Feb 2000, 1320.
Rossman, L., 2000. EPANET-Users Manual. Cincinnati,
OH: US Environmental Protection Agency (EPA).
Sadiq, R. and Rodriguez, M., 2004. Disinfection by-products
(DBPs) in drinking water and predictive models of their
occurrences: A review. Science of the Total Environment,
321 (13), 2146.
Serodes, J.B., Rodriguez, M.J., Li, H., and Bouchard, C.,
2003. Occurrence of THMs and HAAs in experimental
chlorinated waters of the Quebec City area (Canada).
Chemosphere, 51 (4), 253263.
Sohn, J., Gatel, D., and G, Amy., 2001. Monitoring and
modelling of disinfection by-products (DBPs). Environ-
mental Monitoring and Assessment, 70, 211222.
USEPA, 2006. National Primary Drinking Water Regula-
tions: Stage 2 Disinfectants and Disinfection Byproducts
Rule. Federal Register, 71 (2), 388493.