United States
Environmental Protection
Agency
Municipal Environmental Research^ ^ 0
Laboratory ''
Cincinnati OH 45268
Research and Development
EPA-600/S2-82-033 August 1982
Project Summary
Determinants and Options for
Water Distribution System
Management: A Cost
Evaluation
Robert M. Clark, Cheryl L. Stafford, Michael G. Laugle, and James A. Goodrich
The report summarized here deals
with the problems associated with
maintaining and replacing water
supply distribution systems. Some of
these problems are associated with
public health, economic and spatial
development of the community, and
costs of repair and replacement of sys-
tem components. Statistical models
are developed that demonstrate the
relationship between population
growth and development and growth
of the water supply service network. A
repair frequency analysis has been
completed for distribution system
maintenance events (leaks and
breaks). The economic implication of
various replacement strategies and
the effect of water quality (corrosivity)
on water loss and system cost are
examined. This analysis is based on
the data acquired from one large (260
MGD; 11.39 mVsec) and one smaller
(20 MGD; 0.88 mVsec) water utility.
The capital facilities that make up
urban service networks such as water
supply delivery systems, sewage col-
lection networks etc., are often called
the urban infrastructure. The water
system infrastructure represents a
major investment of a municipality.
Because of the potential public health
and safety implications of an inade-
quate water distribution system,
maintaining this system in good con-
dition is an extremely important
responsibility for water utility man-
agement. As this study shows, once a
length of a pipe begins to require
maintenance, its maintenance rate
increases exponentially. Maintenance
costs soon exceed the costs of
replacement. Therefore establishing a
timely maintenance and replacement
program is extremely important from
an economic and public health
viewpoint.
This Project Summary was devel-
oped by EPA's Municipal Environ-
mental Research Laboratory.
Cincinnati. OH. to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Water supply service can be divided
into a series of functions: support ser-
vices, acquisition, treatment, and deliv-
ery. The treatment and deliveryf unctions
of a water utility represent large econom-
ic investments, but the bulk of the ex-
penditures are in the delivery system.
The absolute magnitude of this expendi-
ture can be illustrated from data taken
at a large midwestern water utility [ap-
proximately 260 MGD (11.39 mVsec)
capacity], which is examined in detail.
The replacement value of the delivery
system ( not including treatment, acqui-
sition and support services) is estimated
at $917,814,700 based on 1978 dollar/
foot rates. Maintenance of the delivery
system in 1978 cost approximately
$2,600,000 per year.
-------
Not only does the water utility delivery
system represent a large and important
portion of the water utilities budget but
it also plays a significant role in com-
munity public health and can become a
determinant of the communities growth
path. The degree and direction of urban
development is heavily dependent on
the availability of this portion of the
infrastructure.
With growing concern over available
resources and the cost of energy and
increasing societal awareness over the
role of urban service systems in urban
development, there is also greater
awareness of the role water systems
play in population growth. Such ques-
tions as, "Does population growth force
growth in water supply systems?" or
"Does availability of water service affect
the quantity and direction of population
growth in an urban area?" must be
asked. Because of the health, social and
economic functions served by water
utilities, examining the economics of
the repair, replacement, and mainte-
nance of delivery systems is worth-
while. In this report the following issues
will be examined: (1) spatial, demogra-
phic, and developmental implications of
water system expansion; (2) an analysis
of main break patterns and their eco-
nomic consequences; and (3) a general
investigation of the effect of water qual-
ity on water loss.
The first two issues are studied in the
framework of a case study. The last
issue is examined in terms of a cross-
sectional study.
Case Study Utilities
Two utilities were used as the source
of data for this report. The larger utility
serves a population of nearly three
quarters of a million and, until recently,
derived all of its water from one source.
Now both plants have a maximum
capacity of nearly 260 MGD (11.39
mVsec) and, on a yearly average, pump
approximately 150 MGD (6.51mVsec).
The distribution system is made up of
3,900 miles (6,275 km) of mains, 97.5%
of which are cast iron, 2.1% are rein-
forced concrete pipe, and less then 1%
are steel.
The smaller utility located near the
larger utility serves a combination of
rural and urban users and also draws
water from two sources. Treatment in
1979 yielded 6.7 billion gallons of water
(25.4 billion liters). Most of the 360
miles (579.2 km) of pipe are cast iron;
the remainder are reinforced concrete
or steel.
System Development and Pop-
ulation Growth
For purposes of this analysis data was
used from the larger of the two case
study utilities. Variables chosen for
study were Population Density, Pipe
Age, Pipe Volume, and Distance from
the Central Business District (CBD).
Data for the variables were arranged by
census tracts beginning in 1940. This
date was chosed because (1) utility data
are most complete as of this date, and
(2) the great surge of suburbanization
occurred between 1940 and the
present.
Population density figures per tract
were computed from census informa-
tion between 1940 and 1970. The date
the pipe was first installed subtracted
from 1980 yielded the present age of the
pipe. To relate age to population density,
a weighted average age was computed
per census tract by multiplying the age
of each pipe by its length; adding
together the product to get a sum per
tract; and then dividing by the total feet
of pipe in each tract. Pipes 6 inches in
diameter or greater, which represent
the major transmission of water as
opposed to local distribution, were used
in this analysis. These pipes represent
approximately 7.2% of the total miles of
the pipe in the system.
Pipe volume, essentially pipe density,
was calculated by dividing the total
volume of a pipe in a census tract by the
acreage of the tract. This provided a
measure of volume per acre of pipe in a
census tract comparable to population
density. Distance from the CBD was cal-
culated by measuring the distance
between the geographic centers of each
census tract and the CBD.
Development of quantitative mea-
sures among these variables is difficult,
but a combination of these variables in
conjunction with graphic techniques
can be used to develop insight into the
relationships under study. Suburbani-
zation surged from the mid 1950's to the
present, and the inner city experienced
a severe decline in population density.
As population grew on the periphery of
the city, increases in pipe volume
became necessary near the CBD to
supply the outlying areas.
Two equations were developed in an
attempt to relate the change in popula-
tion density (PD) and pipe volume/acre
(PVA) versus distance from the CBD.
From the equations, PD and PVA exhibit
similar distance decay relationships for
a constant AY. Because of the relative
values of the constants in the two equa-
tions, however, it can be seen that for a
constant distance, PD decreases with
time but PVA increases. These relation-
ships suggest that PVA tends to precede
population.
Results of the study showthat there is
a relationship between population, pop-
ulation distribution, and water supply.
Water is not simply distributed in a
haphazard fashion and despite the ten-
dency of the water supply profession to
think in technical terms, there are sig-
nificant socio-economic implications to
their work. As society enters a period of
growing concern over resource availa-
bility, allocation, and urban develop-
ment, this important link should not be
ignored. More research needs to be con-
ducted in this important area.
Analysis of System Reliability
Facilities used for supplying water
service, although predominantly of a
more permanent character than those
of other public utilities, are nevertheless
subject to mortality and replacement.
Because facility life is long, great diffi-
culties arise in securing factual data
relating to actual life and mortality expe-
rience. Even before a pipe reaches the
point of ultimate replacement, as it
ages, its carrying capacity is severely
reduced. Many cities are experiencing
high maintenance rates indicating that
their distribution systems are failing.
Water main breaks disrupt service,
reduce fire fighting capacity, may dam-
age property, and pose a public health
threat while incurring substantial repair
and replacement costs. When a pipe
breaks, the leak has to be located, the
pipe excavated, and the leak fixed or a
section replaced. A section of pipe expe-
riencing a significant number of breaks
or leaks may be replaced entirely with a
new pipe.
Pipes break because of the:
1. quality and age of the pipe itself,
including connectors and other
equipment;
2. type of environment in which the
pipe is laid, e.g., the corrosive-
ness of the soil, frost and heav-
ing, external loads;
3. quality of the workmanship used
in laying the pipe; and
4. service conditions, such as pres-
sure and water hammer.
An analysis of water main breaks can
provide insight into the reasons why
breaks are occurring in a given area of
-------
the network or in a specific pipe.
Insights from such an analysis can
change pipeline design and construc-
tion policies and provide information as
to whether or not a pipe should be re-
paired or replaced. In deciding whether
to replace a pipe, the replacement cost
and future costs associated with the
new line should be compared with the
cost of repairing the existing line and
incurring possible future costs of repair
and disruption of service.
During the course of this analysis, the
investigators found it difficult to define
"break." Examination of many years of
data revealed that few actual "breaks"
occurred; a "break" in this context
means a rupture of the line causing a
cessation of service. A more subtle and
insidious occurrence was continuous
leakage from certain pipes causing
maintenance crews to take remedial
action. Therefore, the analysis con-
tained in this report is based on "main-
tenance events" or repairs but not on
actual ruptures. A repair is defined as
any event in which water was leaking
and which a crew was sent to fix. These
events do not include leaks from valves
or clamps, but only joint or main line
leaks. Valves and clamps are considered
to be either internal or external fixtures
but not part of the pipe itself.
Many factors were found to influence
the number of maintenance events
associated with a given pipe. The fol-
lowing sections contain an analysis of
some of these factors and an economic
evaluation of the optimal time for pipe
replacement.
Analysis of Maintenance Event
Data
Common sense and experience
would indicate that there are many vari-
ables that might influence repair
events. The basis for this study is a data
set from the two case study water utili-
ties consisting mostly of feeder and
transmission mains. The data set
includes the pipe lengths considered in
the analysis, associated physical design
and demographic data, and cost data.
These mains have been categorized into
457 separate pipes. Separation into
pipe links was based on a junction
between pipes or a change in pipe
diameter. No pipes laid before 1940
were used in the analysis. Break data for
smaller pipes were virtually non-
existent.
The following data were collected for
each pipe section: diameter, material,
age, pressure differential, absolute
pressure, cleaning and lining (if per-
formed), average amount of traffic tra-
versing pipe in a 24-hour period,
percent of length in low, moderate, or
highly corrosive soil, and number of
freezes and thaws since installation.
In addition census tract data were col-
lected to analyze the effect of surface
development and land use on pipe
breakage: percent in transportation,
percent in industry, percent in com-
merce, percent in residences, and popu-
lation density.
Soil data were obtainedf rom U.S. Soil
Conservation Service maps, and pipe
locations were plotted to determine sur-
rounding soil type. The Soil Conserva-
tion Service provided the criteria for
evaluating soil corrosivity, and determi-
nation was made as to whether or not
the pipes lay in high, moderate, or low
corrosive soil.
Most of the water works pipes are
beneath city streets; only a few are
installed beneath sidewalks. Traffic
data were collected from both county
and a city data sources. Because most of
the street pavement in the utility service
area is uniform, stress on the mains is
due primarily to overhead traffic not to
differences in road surfaces.
Weather information obtained from
the U.S. Weather Bureau was complete.
Data from appropriate regional planning
commissions was the source of land use
data for transportation, residential,
commercial, and industrial activities for
each census tract in the large utility's
service area.
With the use of these data, a series of
analyses were made incorporating: sur-
vival analysis, probability of mainte-
nance event, maintenance event
equations, economic analysis of
replacement, and the impact of water
quality on failure rate.
Survival Analysis
A study was made of repairs to all
pipes in the data base from the first
through the tenth repair. Repair mortal-
ity curves (Figure 1) show that over a
period of 40 years, 52.5% of the pipes
studied have had one or less mainte-
nance events, 30% had two or less
maintenance events, etc. These data
indicate that a minority of pipes are
responsible for a majority of the mainte-
nance events. As will be seen in the
following section, those pipes that had
maintenance events, and them at an
increasing frequeny over time.
It was also possible to develop the life
expectancy of pipes based on their age.
Five year old mains with no mainte-
nance events can expect to have an
additional 11.2 years without an event,
whereas 30-year old mains have 5.7
years, 40-year old mains have 1-year
remaining (Figure 2), etc.
Probability of Failure
Of the pipes studied, only a relatively
small number experience maintenance
events, even after long periods of time.
For those that did experience such
events, the time between one event and
the next became increasingly short (Fig-
ures 2 and 3). To study this pheno-
menon, the interarrival time between
repairs was formulated as an exponen-
tial function. The relative slopes of the
curves indicates the time between a
failure becomes increasingly short as
the number of maintenance events
increases (Figure 4). For example, given
that a pipe has three events, the proba-
bility of having another everjtin a very
short time is high. v
Event Estimating Equations
Repair records, were available after
1940 on 307 pipes considered in the
original data set. Because the first
maintenance event did not usually
occur until 15 years after the pipe has
been laid, the analysis could begin at
1930 instead of 1940 on the assump-
tion that no breaks occurred in the first
10 years. Of the 307 pipes laid between
1930 and 1980 only 108 have been
repajred.
Examination of the data revealed that
two underlying mechanisms seemed to
be occurring with those pipes that expe-
rienced maintenance events. A lag
period occurred between the time the
pipe was laid and the first maintenance
event. After the first event, the number
of events seemed to increase exponen-
tially. Therefore, two equations were
developed, the first to estimate the time
to the first event and the second, to esti-
mate the number of events occurring
after the first event.
The predicted events can be com-
pared with actual events as estimated
by the two equations (Figure 5). Each of
the variables considered in the analysis
is discussed in detail in the final report.
Timing of Replacement
Economic Analysis
According to previous analysis, the
number of maintenance events in a
given section of a pipe can be developed
from an equation. As the number of
events per year increases, so does the
-------
cost of responding to them. Equations
were developed to estimate the number
of years from installation to the first
maintenance event; a constant evolved
from the number of repairs, the type of
pipe, the pressure differential, the age
of pipe from the first break, the percent
of land over pipe in low and moderately
corrosive soil, and the surface area of
pipe in highly corrosive soil; and the
growth rate coefficient.
Given the predictive equations, it is
possible to project the number of times a
pipe might break. Such an analysis can
aid in making the decision between con-
tinued repair or replacement. If it can be
shown that a main will encounter an
increasing number of repairs, the main
should be replaced before the dollars
spent on repair exceed the amortized
100
90
80
70
60
50
40
30
20
JO
value of the main in the ground. A cost
trade-off can be calculated by taking the
actual historic cost of laying a main,
updating it to present value by use of the
construction cost index, amortizing the
cost by a formula, and comparing it to
the predicted cumulative dollars spent
on repair. Data from the large utility for
1971 to 1978 was used to develop the
average repair cost per break. During
this period, repair costs have fluctuated
from $1,170/break to $1,760/break,
with $1,430 the overall mean. There-
fore, for the purpose of this analysis, a
repair was assumed to cost $1,430.
In this example, a 16-inch, 1,680-foot
section of a steel main laid in 1937 was
replaced with a 12-inch ductile iron
main in 1978 at a cost of $138,122. This
section had experienced 32 breaks in 41
1st
I
05/0/5 20 25 30
Years
Figure 1. Percent having one or less repair events
4
35 40
years. With the use of equations, the
predicted repair costs can be compared
with the actual repair costs, and for this
steel pipe, the optimal time of replace-
ment occurred around 1969 instead of
the actual replacement date in 1978.
Figure 6 shows the various repair and
replacement cost curves. In time, utility
requirements may change, and problem
pipes may be replaced by entirely differ-
ent materials to avoid future problems;
this must be taken into account in a
utility's replacement strategy.Through-
out the analyses, steel mains had an
unusually high number of repairs, but
unfortunately, not enough steel mains
exist in the data set to allow individual
regression analysis for steel pipes
alone. From these data, it is possible to
predict generally when pipes should be
replaced. Applying these kinds of ana-
lyses to a specific pipe with precise
accuracy may, however, be difficult.
Influence of Water Quality
Water quality may also affect repair
and replacement costs in water distri-
bution systems, e.g., corrosive water
may increase the number of breaks in
water systems. Analyzing the effects of
water quality within a single utility is
difficult because water quality is gener-
ally uniform throughout the system.
The corrosivity of drinking water is a
parameter that has health and eco-
nomic significance as well as aesthetic
significance. Corrosion in a distribution
system may add contaminants to fin-
ished water before it reaches the con-
sumer. Some of these contaminants,
such as lead and cadmium, at suffi-
ciently high concentration levels in
drinking water, may constitute a health
hazard.
The annual loss from water corrosive-
ness has been estimated at about $700
million. In addition to corrosion deterio-
rating the pipe used to convey water,
water leakage from deteriorated distri-
bution systems can be substantial. In
some instances, as much as 25% of the
water leaving a treatment plant is lost
before reaching the consumer.
To analyze the effects of corrosion on
water loss and cost of water supply, the
hardness or softness of the water of 60
water utilities throughout the United
States was determined. For the purpose
of this analysis, if the raw water con-
tained less than 60 mg/Lof hardness as
CaC03, it was considered soft. Utilities
that altered their source water by treat-
ment were placed in the appropriate
category (hard or soft). The analysis
-------
so
40
1 30
.<& 20
10
0 6 10 16 20 25 30
Age of pipe without any breaks
Figure 2. Life expectancy of pipes.
36 40
25
20
15
1O
1 23456 789
Number of repair events
Figure 3. Average number of yeart to subsequent failure.
showed utilities with soft water had sig-
nificantly higher (31%) total unit costs
than those with less aggressive water.
Another factor associated with corro-
sion is water loss. An equation was
developed based on the loss of revenue
producing water as compared with the
total water treated.
Based on this equation, utilities with
small differences in elevation suffer sig-
nificantly greater water loss. This loss
could be because of a lack of pressure
zones. Pressure zones are essential to
ensure adequate water service in sys-
tems with hills; they allow pipes within
these zones to have similar internal
pressures. Systems in generally flat ter-
rains having only one pressure zone,
with the pipe subjected to varying pres-
sure have increased breakage and a sig-
nificantly greater percent of loss.
Surface supplies also have a lower loss
rate than do ground water supplies. This
may be because most ground water
supplies pump directly to the customer,
and therefore, have higher pressure dif-
ferentials than do systems that incorpo-
rate a large number of tanks and
standpipes.
The cross-sectional analysis in this
study indicates that aggressive water is
a factor, along with many others, in the
cost of water supply. Although not con-
clusive, the results seem to justify more
detailed case control studies of systems
supplying either aggressive or nonag-
gressive water.
Summary and Conclusions
This report has dealt with problems
associated with maintaining and replac-
ing water supply distribution systems.
Statistical models as well as graphic
displays have been developed to exam-
ine the relationships between water
supply infrastructure development and
population distribution and growth. A
technical economic analysis of the fac-
tors influencing the reliability of a water
distribution system and associated
costs for repair and replacement was
made. The effects of water quality (cor-
rosivity) on water Iqss and system cost
was also examined.
The results of this study indicate that
there is a relationship between popula-
tion distribution and water supply sys-
tem development. Infrastructure
development typically precedes even
moderate population growth, and as
society enters a time of growing con-
cern over resource availability and qual-
ity and urban development, control of
infrastructure development can be an
-------
1.00 -
0.80 -
.^0.60 -
0.40 -
0.20
1st Break
10 15 20 25
Years
30 35
Figure 4, Probability of pipe failure.
875
750
675
500
250
125
Actual
Predicted
1930 1940
1950 1960
Years
1970 1980
important tool in urban morphology
This analysis implies that decisions b
water supply planners may have signif i
cant socio-economic implications.
From the development of the equa
tions for maintenance events severa
conclusions can be drawn:
1. Metallic pipes take nearly 13 year
more to experience maintenance
problems than do reinforced con
crete pipe. Metallic pipes accumu
lateVnore maintenance events thai
do reinforced concrete pipes over i
period of time. * * .-.
2. Large diameter pipes tend to have i
longer period before the first main
tenance event than do smalle
diameter pipes.
3. Large percentages of industria
=-- devetoj3merrt decrease the timi
until the first maintenance event.
. 4. Th§ -amount of developmen
increases repeat breaks.
The equations should not be used fo
predictive analysis but can be used ti
indicate some of the variables tha
accelerate or retard maintenanci
events. Using these equations it wa:
possible to suggest a scenario for thi
time of optimal repair and replacement
For the data used in this analysis, thi
optimal repair period was slightly ove
30 years.
Water quality may have an advers<
impact on the maintenance event fre
quency for water delivery system pipes
Analysis revealed that utilities witt
aggressive water might expect up t<
31% higher unit costs.
Throughout the various analyses, dif
ficulties were encountered in the dat;
collection. In many cases, the format fo
recording data was left up to variou:
individuals throughout the years, am
was, therefore, subject to much individ
ual discretion. One conclusion to b<
drawn from this study is the need fo
water utility managers to institute care
ful record keeping procedures for track
ing pipe repair and replacement costs
Most technical data obtained frorr
agencies such as the National Weathe
Service or Soil Conservation Servic<
were very good; however, data from th<
utilities and planning agencies some
times lacked;consistency and complete
ness, Because significant savings car
be achieved.by replacing transmissior
and distribution pipes at the prope
time, the issue of a system's deteriora
tion will no doubt become much mon
significant in the future.
Figure 5. Predicted vs. actual breaks for combined data set.
6
-------
Optional
replacement Actual
time t*r replacement
(19691 (1978)
1982 Date
45 Years
Figure 6. Repair vs. replacement costs.
The EPA authors Robert M. Clark (also the EPA Project Officer, see below).
Cheryl L. Stafford, Michael G. Laugle. and Jamas A. Goodrich are with the
Municipal Environmental Research Laboratory. Cincinnati, OH 45268.
The complete report, entitled "Determinants and Options for Water Distribution
System Management: A Cost Evaluation," (Order No. PB 82-227 745; Cost:
$9.00, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati. OH 45268
ftUSGPO: 1988-559-092/0447
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
RETURN POSTAGE GUARANTEED
PS 0000329
U S ENVIH PROTtCTIUN AGENCY
REGION 5 LIBRARY
230 S UEARbOKN STREET
CHICAGO IL 60604
------- |