utility operations

S teven F o l k m a n , Jo h n R ic e , A m m o n So re n s o n ,

and N ath a n B ra it h w a it e

Survey of water main

failures in the United
States and Canada

D
uring 2011, Utah State University conducted a basic survey
RESULTS OF A 2011 SURVEY
of utilities across the United States and Canada to obtain data
OF UTILiTIES IN THE UNITED on water main failures of municipal and private water supply
STATES AND CANADA systems. Surveys were mailed to a total of 1,051 US and
Canadian water utilities in May and June of 2011. Those that
PROVIDES INFORMATION ON
responded to the basic survey were also invited to participate in a more
WATER MAIN FAILURES IN detailed survey. A total of 188 utilities responded and completed the basic
MUNICIPAL AND PRIVATE survey, with 47 also responding to a detailed survey. A total of 117,603 mi
of pipe were represented in the basic survey and 32,130 mi in the detailed
WATER SUPPLY SYSTEMS.
survey. This article documents some of the results from both surveys.
One of the primary goals of this survey was to obtain data on current
pipe inventories and failure rates of different pipe materials during a
previous 12-month period. Most of the responses to the survey were
received in the middle of 2011; thus, the survey estimates behavior during
the 2010 and 2011 calendar years. The detailed survey allowed collection
of a wider variety of data relating to operating parameters, installation
practices, and pipe material preferences.

70 OCT OBE R 2 0 1 2   |   J O U R N A L AW WA   •  1 0 4 :1 0   |   F O L K M A N ET A L

2012 © American Water Works Association

SURVEY REGIONS •  Region 8—20 (4) water main pipe represented by
To show participation as a func- •  Region 9—12 (5) those utilities. The trend line and
tion of geography, the areas covered No responses were obtained from equation are a best fit to the data
by the basic survey were broken the Canadian provinces of New with a zero intercept. The slope of
down into nine regions as shown in Brunswick, Newfoundland, Nova this line indicates that on average
the map on page 71. The number of Scotia, Prince Edward Island, or there are 264 people served for each
respondents to the basic survey are Quebec. Figure 1 shows the miles of mile of water main installed.
listed, followed by the number of water main pipe that were reported
respondents to the detailed survey in in the basic and detailed surveys SURVEY SAMPLE SIZE
parentheses. according to region. The total length of water main
•  Region 1—19 (7) pipe reported by the 188 survey par-
•  Region 2—20 (4) MILES OF PIPE VERSUS ticipants was 117,603 mi (the survey
•  Region 3—26 (5) POPULATION did not include sewer or force
•  Region 4—22 (6) Figure 2 shows the relationship mains). In 2007 the US Environmen-
•  Region 5—18 (3) between the population served by tal Protection Agency (USEPA)
•  Region 6—28 (11) the utilities participating in the basic reported that there are approxi-
•  Region 7—23 (2) survey and the number of miles of mately 880,000 mi of distribution

BC
AB
MB
SK
QC NL
Region 9
AK ON PEI

WA NB
NS
Region 1 MT ND ME
VT
OR
MN NH
ID WI
SD NY MA
WY MI
RI
NE IA PA
NV Region 3 Region 6 CT
Region 2 OH Region 8
UT Region 4 IL IN NJ
CO WV
HI KS VA DE
CA MO KY
NC MD
TN
AZ OK
NM AR SC DC
Region 5 AL
MS GA
LA Region 7
TX

FL

This map illustrates the regions as they were defined to report the basic and detailed survey results.

FO LK M A N ET A L   |  104: 10  •   JO U R NA L AWWA   |   O C TO B ER 2012 71

2012 © American Water Works Association

pipe in the United States (USEPA, survey would then represent approx- quantified the size of a utility based
2007). The following year, a second imately 10% of the installed water on the number of miles of water
report (Royer, 2008) estimated the main pipe in the United States. Thus main pipe installed. Four categories
were established according to miles
of pipe: small (up to 500 mi),
medium (500–1,000 mi), large
A primary goal of this survey was to obtain data (1,000–3,000 mi), and very large
(more than 3,000 mi). Each survey
on current pipe inventories and failure rates of different
participant was allocated to one of
pipe materials during a previous 12-month period. the categories. Figure 3 shows the
distribution of total miles according
to these categories.

amount of installed water main pipe survey sample size is significant and PIPE MATERIALS
in the United States to be more than should provide reliable results. The Many pipe products have evolved
1 million mi. The current population Canadian provinces comprised over time, and most of these could
of the United States is 312 million 8,423 mi of pipe in this survey, or be broken down into subcategories
(US Census, 2011). Using the esti- 7.4% of the total. on the basis of processing and sur-
mate from the previous section of face treatments. These changes,
264 people served per mile of water PARTICIPANTS CLASSIFIED along with new installation tech-
main, the length of water mains can IN FOUR CATEGORIES niques, should affect the life expec-
be estimated to be 1.18 million mi The average utility surveyed had tancy of the pipe. The basic survey
([312 million people]/[264 people/ 626 mi of water main, with the was intended to be relatively simple
mile]). With the use of this last esti- largest having 4,468 mi and the to complete in order to encourage
mate, the total length of pipe in the smallest having 2 mi. This survey wide participation. Most utilities
have limited records as to which spe-
cific pipe materials were installed
decades ago. Therefore, subcatego-
ries of material types were not
tracked in the survey. The results
reported in this article represent
FIGURE 1 Total miles of pipe by region responding to the generic pipe material performance,
basic and detailed surveys but may not represent a specific
product on the market today.
Basic survey
25,000 The distribution of pipe materials
based on miles of pipe is shown in
20,000
Figure 4. Pipe material distribution
Pipe—mi

15,000 as a percentage of the total length

10,000 for both the basic and detailed sur-
veys is shown in Figure 5. The
5,000
detailed survey had a smaller num-
0 ber of respondents but still gave a
1 2 3 4 5 6 7 8 9
similar material distribution. The
Region
“other” category in Figures 4 and
Detailed survey
5 includes materials such as high-
8,000 density polyethylene, galvanized
steel, and copper. Eighty percent of
6,000 the installed water mains are a
Pipe—mi

combination of cast-iron at 28%,

4,000
ductile-iron at 28%, and polyvinyl
2,000
chloride (PVC) pipe at 23%. The
amount of concrete pressure pipe
0 (CPP), steel, and asbestos–cement
1 2 3 4 5 6 7 8 9
(AC) material types were also
Region
reported. The length of high-den-
sity polyethylene pipe reported rep-

72 OCT OBE R 2 0 1 2   |   J O U R N A L AW WA   •  1 0 4 :1 0   |   F O L K M A N ET A L

2012 © American Water Works Association

resented only 0.14% of the total
and thus was not tracked as a sepa-
rate group.
Figure 6 shows the regional distri-
bution of pipe material type as a
percentage of the total length of pip-
ing in that region. Significant differ-
ences can be seen in the type of pipe
material used in the regions. Cast
and ductile iron represent approxi-
mately 90% of the pipe in regions 6
and 8. PVC has a dominant role in
region 9 and is a significant player in
regions 2–5, and 7.
Pipe age and diameter. The detailed
survey asked respondents to provide
the age distribution of installed pipe.
Four age groups were provided:
0–10 years, 10–20 years, 20–50
years, and more than 50 years. Fig-
ure 7, which lists the percentage of
water main length that fits in each
age category, indicates that approxi-
mately 43% of installed pipes are in
the 20–50-year age category and
22% are more than 50 years old.
Respondents were also asked to
break down the fraction of total
installed pipe length by five pipe-diam-
eter categories. Figure 8, which shows
the percentage of water mains that fit
into each size range, indicates that
The total miles of pipe covered by this survey is almost
10% of that in use in the United States.
approximately 66% of the installed
pipe is 8 in. or less in diameter.
Delivery pressure. On the basis of
results received in the detailed survey,
the average delivery pressure was
determined to be 77 psi. The range of
values reported for average delivery
pressure was 45–150 psi. Respon-
dents were asked to select one of
three ranges provided in the detailed
survey to indicate how much the
water pressure fluctuates at any given
point in their systems during a typical
day. The survey indicated that
approximately 83% of respondents
are able to limit pressure fluctuations
to less than 20 psi, 11% had pressure
fluctuations between 20 and 40 psi,
and 6% have pressure fluctuations in
excess of 40 psi.
PIPE FAILURE
Most common failure mode, mate-
rial, and age. The detailed survey
asked the respondents to identify the
most common water main failure
mode by selecting from one of the
following categories: circumferential
crack, corrosion (pits or holes), lon-
gitudinal crack, leakage at joints,

FIGURE 2 Population served relative to total pipeline miles from the basic survey

1,400,000 y = 263.78x
R 2 = 0.8014

1,200,000

1,000,000

800,000
Population

600,000

400,000

200,000

0
0 1,000 2,000 3,000 4,000 5,000

Pipe—mi

FO LK M A N ET A L   |  104: 10  •   JO U R NA L AWWA   |   O C TO B ER 2012 73

2012 © American Water Works Association

 fatigue, or other. Figure 9 shows that
TABLE 1 Summary of failure data over a 12-month period circumferential cracks are the most
common mode of failure, followed
Number Failure Rate by corrosion. The detailed survey
Pipe Material Length—mi of Failures Number/(100 mi)/(year) also asked respondents to identify
Cast iron  33,611.0  8,204 24.4 which pipe material failed most
Ductile iron  33,238.7  1,620  4.9 often. Figure 10 shows that slightly
more than 55% of respondents iden-
Polyvinyl chloride  26,840.3   689  2.6
tified cast iron as the pipe material
Concrete pressure pipe   2,355.3   128  5.4 that failed most often (likely because
Steel   4,300.1   581 13.5 it is one of the two most commonly
Asbestos–cement  13,502.8   954  7.1 used pipe materials and on average
is the oldest pipe material in use),
Other   3,755.3   787 21.0
followed by AC at 17%. Respon-
 Total 117,603.4 12,963 11.0 dents also reported that the typical
age of a failing water main was 47
years, with a range of values from
20–86 years. The survey also asked
what the respondents believed the
expected life should be for newly
FIGURE 3 Total miles of pipe in each size group
installed pipe. Respondents expected
new pipe to have an average life span
35,000 of 79 years, with responses ranging
30,000 from 30 to 200 years.
Total Length of Pipe—mi

25,000
Computing failure rates. The basic
survey asked respondents to consider
20,000
a water main failure as one in which
15,000 leakage was detected and repairs
10,000 were made. However, they were
asked not to report failures caused
5,000
by joint leakage, construction dam-
0 age, or tapping of service lines. The
0–500 500–1,500 1,500–3,000 3,000–5,000
goal of the survey was only to exam-
Utility Size Group (in miles of pipe)
ine the performance of properly
installed pipe.
Utilities reported the number of
FIGURE 4 Total length of pipe classified according to material failures over a recent 12-month
type from the basic survey period according to pipe material
and the installed length of pipe mate-
rial. The failure rate was computed
40,000
by dividing the total number of fail-
35,000 Other includes ures from all utilities for a particular
HDPE, galvanized
30,000 steel, and copper pipe material by the total length of
25,000 that pipe material. For example, the
Pipe—mi

survey reported a total of 12,963

20,000
failures of water mains during a
15,000 recent 12-month period for all pipe
10,000 materials. The total installed water
5,000 main length from the survey was
117, 603 miles (or 1,176.03 hun-
0
CI DI PVC CPP Steel AC Other dreds of miles). Thus the overall fail-
ure rate is 12,963/1,176.03 = 11.0
failures/(100 miles)/year. Figure 11
AC—asbestos–cement, CI—cast iron, CPP—concrete pressure pipe, shows the failure rate at each utility
DI—ductile iron, HDPE—high-density polyethylene, PVC—polyvinyl chloride
for all pipe materials and indicates
that utilities can experience widely

74 OCT OBE R 2 0 1 2   |   J O U R N A L AW WA   •  1 0 4 :1 0   |   F O L K M A N ET A L

2012 © American Water Works Association

different failure rates for the same
pipe material. This finding should
not be surprising. Several significant
variables affect pipe performance,
including age, soil types (corrosive
or noncorrosive), corrosion preven-
tion techniques, installation prac-
tices, and climate. Thus a utility may
have a significantly different failure
rate from those reported here. To
compute an average failure rate for
a given pipe material, a large num-
ber of respondents are needed.
Failure rates for each pipe material.
The basic survey measured pipe fail-
ures over a recent 12-month period,
broken down by material type.
Table 1 lists the total length of pipe
by material type, the number of fail-
ures over a recent 12-month period,
and the failure rate for each pipe
material. Figure 12 shows the failure
rates as a function of material type,
and Figure 13 classifies the failure
rates by material type and by US and
Canadian respondents. When the
failure rates per 100 mi of pipe per
year were compared for cast-iron,
ductile-iron, PVC, concrete, steel,
and AC pipes, PVC was shown to
have the lowest overall failure rate.
Plans for replacing water mains. The
detailed survey asked whether
respondents had a plan for regular
replacement of water mains that are
nearing their end of useful life. More
than 77% reported they do have a
replacement plan in place. The survey
also asked what percentage of water
mains are beyond their useful life but
It is imperative that utilities make wise choices when
repairing and replacing pipe, and benchmarking
can provide guidance in making those decisions.
have yet to be repaired/replaced
because of a lack of funds. Respon-
dents reported that from zero to 75%
of their pipe was beyond its useful
life. The average of all respondents
showed that 8.4% of the pipe is
beyond its useful life.
Contractor experience. Proper
installation is important to getting
the maximum life out of a water
main. The survey asked whether the
respondents’ utilities used contractor
experience as a weighting factor dur-
ing the selection process. Sixty-six
percent of respondents do consider
contractor experience. Many of
those respondents who do not con-
sider contractor experience reported
that state or local laws prevented
them from doing so.
Allowed pipe materials. Results
from the survey indicated that
87.2% of utilities would allow
installation of ductile iron, 59.6% of
PVC, 38.3% of concrete pressure,
and 36.2% of steel pipe. Typical
comments regarding pipe exclusion
included “corrosion issues for duc-
tile iron and steel” and “strength,

FIGURE 5 Percentage of total length of pipe classified by material type from the surveys

Basic
Detailed
35

30 29 28
28 27
25
25
23
Total Length—%

20

15
12
11
10

5 3 4 3
2 3
2
0
CI DI PVC CPP Steel AC Other

AC—asbestos–cement, CI—cast iron, CPP—concrete pressure pipe, DI—ductile iron, PVC—polyvinyl chloride

FO LK M A N ET A L   |  104: 10  •   JO U R NA L AWWA   |   O C TO B ER 2012 75

2012 © American Water Works Association

 FIGURE 6 Regional percentage of length of pipe classified by material type from the basic survey

Other Steel PVC CI

AC CPP DI

6
Region

0 10 20 30 40 50 60

Installed Length—%

AC—asbestos–cement, CI—cast iron, CPP—concrete pressure pipe, DI–ductile iron, PVC—polyvinyl chloride

76 OCT OBE R 2 0 1 2   |   J O U R N A L AW WA   •  1 0 4 :1 0   |   F O L K M A N ET A L

2012 © American Water Works Association

tapping difficulties, and bedding
concerns for PVC.” Comments FIGURE 7 Percent of pipe length classified according
regarding CPP included “difficult to to age from the detailed survey
install, tap, and repair” and “has
corrosion issues.” 50
Leak-detection methods. The 43.3
40

Total Length—%
detailed survey asked respondents
whether regular leak-detection 30
methods were used at their utility, 22.2
20 18.4
and 57% indicated that they were. 14.9

The methods of leak detection used 10

1.3
included acoustic leak detectors, 0
visual inspection of lines, digital 0–10 10–20 20–50 > 50 Unknown
correlation sensors, and eddy cur-
Pipe Age—years
rent detectors.
Corrosive soils and corrosion pre-
vention treatments. The detailed sur-
vey asked respondents whether FIGURE 8 Percentage of pipe length classified according to pipe diameter
there are one or more regions in from the detailed survey
their service area with soils that are
corrosive. A total of 75% of the
respondents reported that they have
at least one area with corrosive 70 65.9
soils. Thus, corrosion is a signifi- 60
cant problem for most of the re­­ 50
Total Length—%

spondents. The survey also asked

what kind of corrosion treatments 40

they are currently using. Typical 30

corrosion prevention treatments for 20 18.4
DI pipe included polywrap and the 10.1
10
installation of magnesium anodes. 3.4 2.2
Covered bedding improvements, 0
8 10–14 16–24 30–36  36
type 50 concrete, coatings, and the
use of anodes were common treat- Pipe Diameter—in.
ments for CPP. Steel pipe corrosion
treatments listed impressed current,
anodes, and protective coatings.
Effect of ambient temperature on fail-
FIGURE 9 Percent of respondents selecting a most common
ures. The detailed survey asked
failure mode
whether the utility observed an
increase in water main failures with 60
extreme ambient temperatures, either 50
50
warm or cold. The results indicated
that 72% of the respondents did note
Respondents—%

40
an increase in pipe failures with
extreme cold ambient temperatures 30 28.3
and only 13% reported a correlation
with warm temperatures. Clearly this 20
13.0
result would be dependent on the cli-
10
mate at each utility. 4.3
2.2 2.2
0
CONCLUSION
Corrosion

Fatigue

Other
Longitudinal
Crack

Leak
Circumferential
Crack

To capture statistically signifi-

cant water main break data that
can provide an accurate portrayal
of current pipe behavior and water

FO LK M A N ET A L   |  104: 10  •   JO U R NA L AWWA   |   O C TO B ER 2012 77

2012 © American Water Works Association

utility practices requires a large individual utilities as possible •  The primary water main pipe
number of participants. The total internal benchmarks. materials in use today are cast iron
miles of pipe covered by this sur- •  There are approximately 264 (28% of the installed base), ductile
vey is almost 10% of that in use in people who are served per mile of iron (28%), and PVC (23%). The
the United States. From these data water main pipe, which is an average survey results indicate that the type
many observations can be made at number representing both rural and of pipe material installed varies sig-
a national level and applied to urban populations. nificantly across the regions; approx-
imately 90% of the pipe in the
northeastern United States is either
ductile or cast iron, whereas PVC is
FIGURE 10 Percent of respondents selecting a most common the dominant pipe in Canada.
failure material •  Water main pipe characteristics
uncovered by the survey indicate
60
55.3 that more than 22% of currently
installed pipe is more than 50 years
50 old and that about 66% of water
mains are 8 in. or less in diameter.
40 The average age of a failing water
Respondents—%

main was only 47 years, with 8.4%

of pipes classified as being beyond
30
their useful life. These results are in
sharp contrast to respondents’
20 17.0 beliefs that new pipe should last 79
years on average.
10
•  The average delivery pressure is
6.4 6.4
4.3 4.3 77 psi and 83% of utilities keep
2.1
water pressure fluctuations at less
0
than 20 psi.
CI DI PVC CPP Steel AC Other
•  The survey also found various
limitations placed by utilities on pipe
AC—asbestos–cement, CI—cast iron, CPP—concrete pressure pipe, DI–ductile iron,
PVC—polyvinyl chloride
materials. The percentage of utilities
allowing installation of specific pipe
materials is as follows: ductile-iron
pipe—87%, PVC pipe—60%, con-
crete pressure pipe—38%, and steel
pipe—36%.
FIGURE 11 Total failure rates at each utility relative to miles of pipe for all pipe •  Utilities exhibit a large variation
materials in water main failure rates.
•  Approximately 75% of utilities
70 have at least one region in their ser-
vice area with corrosive soils.
60
(Number of Failures)/(100 mi)/year

•  This survey found that on the

50 basis of the number of failures per
100 mi of pipe per year, PVC pipe
40 currently has the lowest overall
failure rate.
30 Information from this survey is
Average
intended to assist utilities in seeing
20
how they compare with national
10 norms. The amount of pipe ap­­
proaching the end of its life is grow-
0 ing. It is imperative that utilities
0 1,000 2,000 3,000 4,000 5,000 make wise choices when repairing
Pipe—mi
and replacing pipe, and benchmark-
ing can provide guidance in making
those decisions.

78 OCT OBE R 2 0 1 2   |   J O U R N A L AW WA   •  1 0 4 :1 0   |   F O L K M A N ET A L

2012 © American Water Works Association

ACKNOWLEDGMENT
This work was completed with
support from Uni-Bell PVC Pipe
Association and the Water Finance FIGURE 12 Failure rates of each pipe material per 100 mi over a one-year period
Research Foundation.
30

ABOUT THE AUTHORS 25

24.4
Steven Folkman is 21.0

Failures/(100 mi)/year
an associate 20
professor at Utah
State University, 15 13.5
4130 Old Main
Hill, Logan, UT 10
84332-4130; 7.1
4.9 5.4
steven.folkman@ 5
2.6
usu.edu. A member of AWWA,
Folkman has oversight of Utah 0
State’s Buried Structures CI DI PVC CPP Steel AC Other
Laboratory, is co-author of Buried
Pipe Design, and has been involved AC—asbestos–cement, CI—cast iron, CPP—concrete pressure pipe, DI–ductile iron,
PVC—polyvinyl chloride
with analysis and testing of buried
structures for more than 20 years.
His work has been published
previously in Journal of Civil
Engineering and Architecture and
Transportation Research Record.
He received his BS, MS, and PhD
degrees from Utah State University.
John Rice is an assistant professor
and Ammon Sorensen and Nathan
Braithwaite are research assistants, FIGURE 13 Failure rates reported in this survey by US and Canadian respondents
all at Utah State University.
US
REFERENCES Canada
40
Royer, M.D., 2008. Condition Assessment of
Drinking Water Transmission and Distri- 35.0
bution Systems. Presented at EPA Sci- 35
ence Forum 2008, Washington, D.C., May
20–22, 2008. www.epa.gov/awi/ 30
pdf/600f09030.pdf (accessed Aug. 31,
Failures/(100 mi)/year

2012). 25 23.9
21.3
US Census, 2011. www.census.gov/
20
(accessed Oct. 5, 2011).
USEPA (US Environmental Protection 15.2
15 13.8 13.4
Agency), 2007. Distribution System
Inventory, Integrity and Water Quality.
10
www.epa.gov/ogwdw/disinfection/tcr/ 7.0 7.5
pdfs/issuepaper_tcr_ds-inventory.pdf 6.2
4.3 3.9
(accessed Aug. 31, 2012). 5 2.9
0.7 0.9
0
http://dx.doi.org/10.5942/jawwa.2012.104.0135 CI DI PVC CPP Steel AC Other

AC—asbestos–cement, CI—cast iron, CPP—concrete pressure pipe, DI–ductile iron,

Journal AWWA welcomes PVC—polyvinyl chloride

comments and feedback

at journal@awwa.org.

FO LK M A N ET A L   |  104: 10  •   JO U R NA L AWWA   |   O C TO B ER 2012 79

2012 © American Water Works Association