http://www.nrc-cnrc.gc.

ca/irc

Fatigue failure of large-diameter cast iron mains

NRCC-53610

Rajani, B.B.; Kleiner, Y.

September 2010

A version of this document is published in / Une version de ce document se trouve dans:

Water Distribution System Analysis 2010 - WDSA2010, Tucson, AZ, USA,
September 12-15, 2010, pp. 1-13

The material in this document is covered by the provisions of the Copyright Act, by Canadian laws, policies, regulations and international
agreements. Such provisions serve to identify the information source and, in specific instances, to prohibit reproduction of materials without
written permission. For more information visit http://laws.justice.gc.ca/en/showtdm/cs/C-42

Les renseignements dans ce document sont protégés par la Loi sur le droit d'auteur, par les lois, les politiques et les règlements du Canada et
des accords internationaux. Ces dispositions permettent d'identifier la source de l'information et, dans certains cas, d'interdire la copie de
documents sans permission écrite. Pour obtenir de plus amples renseignements : http://lois.justice.gc.ca/fr/showtdm/cs/C-42
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
FATIGUE FAILURE OF LARGE-DIAMETER CAST IRON MAINS

Balvant Rajani and Yehuda Kleiner

National Research Council of Canada, Institute for research in Construction,
1200 Montreal Road, Ottawa, Ontario, K1J 6G8 Canada

[email protected]

Abstract

Water utility engineers have reported on large diameter water main failures that occurred suddenly without
warning, no signs of prior leaks and no visual evidence of corrosion on the fracture surfaces. Often these
failed mains had been operating without major problems for over 80 years. A possible explanation may be
attributed to alternating or fluctuating stresses such as those caused by heavy traffic and cyclical operating
water pressure with occasional occurrences of transients. These fluctuating stresses are accentuated if the
pipe impinges on an object with sharp geometry and high stiffness like a rock or a stone.

Fatigue of cast iron has been extensively studied in the context of cast iron bridges, structural elements
such as columns, engine blocks, etc, but not in the context of buried, grey cast iron water pipes. Fatigue
analysis methods developed over the years involve a lot of empiricism and combine engineering principles
with experimental observations. Consequently, it is fair to say that fatigue analysis results should be taken
more as a guide than as precise or accurate answers.

A mechanistic approach to explain fatigue failures of buried cast iron pipes had not been previously
explored. This paper explores the application of the fracture mechanics approach (LEFM) to explain some
failures in cast iron pipes that occur through the fatigue mechanism. It endeavors to provide insight into the
plausibility of fatigue failures in grey cast iron pipes when and if subjected to alternating (also often
referred to as repeated or variable) stresses due to surface traffic loads, operating pressure variations and
transient pressure occurrences. It is important to note that the proposed analysis refers to grey cast iron
pipe type, with carbon in form of flake graphite, which is the predominant material of existing iron trunk
mains in North America and Europe.

Keywords
Cast iron mains, large diameter, alternating stresses, fatigue failures.

1. BACKGROUND
Anecdotes told by engineers from water utilities on some cast iron main failures follow these lines: “ We
had a large diameter water main fail suddenly and visual examination of the fracture surfaces showed no
signs of corrosion. There was no prior warning or a leak that surfaced to indicate that anything imminent
was about to happen. This main has been operating without major problems for over 80 years. Why and
how did this failure occur?” Typical examples of such failures in Cleveland, Ohio (1926) and in London,
England (2004) are shown in Fig. 1.
Besides present day stories like the one cited above, anecdotal speculation(s) (CIPRA, 1927; Handover,
1930) suggest that some cast iron mains may have failed due to alternating or fluctuating stresses such as
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 

(a) Cleveland, 1926 (b) London, 2004

Figure 1. Typical breaks in large diameter cast iron mains with no visual corrosion.

those due to surface traffic and unsteady operating water pressure with occasional occurrences of pressure
transients. The fluctuating stresses are accentuated if the pipe happens to impinge on an object with sharp
geometry and high stiffness like a rock or a stone.
No documented descriptions of fracture surfaces that could be attributed to fatigue in cast iron are found in
the literature, except of mention in the report commission by Cleveland Water Department (2001) on trunk
mains failures. This report includes a document prepared by Professional Service Industries (1994) on the
metallurgical examination of failure in a 30 in cast iron main. The report states that fatigue failure surface
exhibits a smooth, more uniform fracture surface perpendicular to the wall thickness and is accompanied by
concentric clamshell like striations. These striations were not found in the specific pipe failure they
examined. It therefore seems that this description was inferred from evidence found in ductile materials
(Brooks and Choudhury, 1993) but has not been documented for brittle materials like cast iron.
Fatigue of cast iron has been extensively studied in the context of cast iron bridges, structural elements such
as columns, engine blocks, etc, but not in the context of grey cast iron water pipes. The fatigue analysis
methods developed over the years involve a lot of empiricism and combine engineering principles with
experimental observations. Consequently, it is fair to say that fatigue analysis results should be taken more
as a guide and not as being able to provide precise or accurate answers.
Fatigue failure occurs when a material is subjected to repeated or varying load(s), never reaching a high
enough level to cause failure in a single realization. The fatigue process that typically takes place in metals
can be divided into two phases or lives (Fig. 2), namely, the initiation phase and the propagation phase. A
crack initiates and grows into a small crack in the initiation phase, while the small crack grows bigger and
leads to failure in the propagation phase. The duration of these two phases is typically expressed in number
of cycle reversals of stress or strain and is dependent on many variables such material type, geometry of
component, etc.
A mechanistic approach to explain fatigue failures of cast iron pipes has not been explored previously
except for a very preliminary and limited assessment by Bonacuse (2001). This paper explores the
application of the fracture mechanics approach (LEFM) to explain some failures in cast iron pipes that
occur through the fatigue mechanism. Further, it endeavors to provide insight into the plausibility of fatigue
failures in grey cast iron pipes when and if subjected to alternating stresses due to surface traffic loads,
operating pressure variations and transient pressure occurrences. It is important to note that the discussion
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
in this paper refers to grey cast iron pipe type, with carbon in the form of flake graphite, which is the
predominant material of existing iron trunk mains

100

Alternating stress
Crack propagation
period

10

Crack initiation
period

Endurance limit, σ e
1
1 100 10,000 1,000,000 100,000,000

No of cycles to failure, N

Figure 2. Initiation and propagation periods of fatigue crack growth.

2. BRIEF REVIEW OF FRACTURE MECHANICS METHOD FOR FATIGUE

ANALYSIS
The approach based on fracture mechanics method is suitable to predict crack propagation, which means that
initial crack size or defect size is known or can be reasonably assumed. As stated earlier, presence of graphite
flakes in grey cast iron introduces a large population of internal flaws even before any loads (stresses) are
applied. New cracks can develop or existing cracks can grow if a cast iron pipe is mishandled during delivery
or during caulking of bell-spigot joints. Of course, the only means to ascertain the existence of crack(s) of
significant size is to conduct inspection of the pipes to uncover if and where these cracks exist.
In addition to cracks initiating at delivery or caulking, it is well known that cast iron pipes that were
manufacture by early (probably between 1850s and 1930s) techniques were fraught with voids and
inclusions in the iron matrix. Some of these voids in small and large diameter cast iron mains could have a
significant size in comparison to pipe wall thickness, as seen in Fig. 3. Again, it is difficult to know a priori
where these voids are except through the use of an appropriate inspection technology, if available. It is
therefore postulated that an assumption about the existence of cracks or voids in cast iron pipes is not
far-fetched but realistic. In the following discussion, the term crack is used to mean a fracture or casting
void or defect as the case may be.
Some important elements of the fracture mechanics approach are:
• Initial crack size is known or can be estimated.
• Plastic zone at the crack tip is assumed to be small compared to crack length and size of cracked
component.
• Local stresses and strains are related to the remote or nominal stresses and strains by the stress intensity
factor, K, which is defined in total or incremental form as,

[1] K = f ( g )σ πa or ΔK = f ( g ) Δσ πa
where σ is the remote or nominal stress, a is the crack length and f(g) is the correction factor that
depends on specimen and crack geometry. The stress intensity factor value that leads to the
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
development of unstable fracture is known as the toughness factor, Kc. Fracture toughness is typically
measured (ASTM E 399-90) in plane strain condition since this state offers the maximum constraint
conditions. This fracture toughness is designated, KIc, since the test corresponds to opening or tensile
mode, or Mode I, loading condition.

Figure 3. Typical casting voids and defects in small and large diameter grey cast iron pipes.

Fatigue analysis based on principles of linear elastic fracture mechanics (LEFM), essentially relates crack
growth rate with stress intensity. This relationship is expressed in the form of power equations such as those
proposed by Paris, Walker and Forman. The application of linear elastic fracture mechanics to conduct
fatigue analyses are documented in several textbooks, most prominent of which are those by Broek (1988)
and Ewalds and Wanhill (1984). Therefore, reference is made to the theory only when required in the
context of fatigue failure of cast iron mains.
The essential steps in this method are: (1) conduct tests to obtain threshold and plane strain fracture
toughness, (2) conduct tests at different stress intensities to obtain crack growth data (3) generate load
spectrum based on realistic load occurrences for the pipe under consideration, (4) integrate the power
equation to determine crack size as a function of number of stress cycles for a given initial crack size and
data generated in steps 1, 2 and 3. The fatigue crack propagation analysis (step 4) in this study is conducted
using a software application, AFGROW, developed at the US Air Force Wright Aeronautical Laboratories
(Harter, 2008). All of the above steps are discussed in detail below, as appropriate.
2.1 Fracture toughness
Rajani et al. (2000) conducted fracture toughness measurements on pit and spun cast iron pipes of different
vintages, manufactured in US and Canada. The specimens for these tests came from pipes installed between
1881 and 1969 in 16 different water utilities across the United States and Canada. The range of fracture
toughness for these pit and spun cast iron pipes were 9.9 ± 2.3 and 13.3 ± 1.5 MPa√m, respectively.
Fracture toughness of spun cast iron pipes is on average higher than of pit cast iron pipes. Also there is a
general indication that facture toughness of pit cast iron improved as improvements were introduced in the
manufacturing process to combat the brittleness of cast iron. Nonetheless, fracture toughness for both pit
and spun cast iron varies widely.
Deb et al. (2002) also reported fracture toughness tests on spun cast iron pipes from US and Canada but
only test results for pipes from Canada fall within acceptable range for cast iron. Reported results from US
utilities were in the 17 to 100 MPa√m range which seems to be rather high compared with other measured
values.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
2.2 Fatigue growth parameters
In the linear elastic fracture mechanics approach to fatigue analysis, fatigue cracking is assumed to take place
in three stages as shown in Fig. 4. In stage I, crystals slip and extend through contiguous grain boundaries,
voids and surface imperfections and are not visible to the observer. In stage II, the crack grows monotonically
and is observable on micrographs taken by means of electron microscope. In stage III, final fracture occurs in
an uncontrolled or unstable crack growth. Only stage II crack growth is discussed in this paper.
 

Stage II
Crack
propagation
Log (da/dN)

Stage I (steady state) Stage III

Crack Unstable
initiation crack (fast
fracture)

(ΔK)th Kc

Log ΔK
Figure 4. Crack growth development as a function of stress intensity factor.

Fatigue crack growth will not occur if the stress intensity factor in stage I is below a certain threshold value,
ΔKth. Crack growth (da/dN vs. ΔK) in stage II is represented by the Paris expression, as follows,

= A(ΔK )
da
[2]
m

dN
where N is number of cycles, A and m are material constants and ΔK is the stress intensity range (= Kmax –
Kmin). Constants A and m are obtained by fitting the above expression to experimental test results. It is
important to note that units of constant A are specific to the exponent m, i.e., [Stress-m Length(1 – m/2)]. Two
important factors that affect the constants A and m to describe crack growth are stress ratio, R, and
environmental effects. Appropriate test conditions have to be used so that constants A and m reflect the
desired and expected environment.
The number of cycles to failure, Nf, may be calculate from [2] by integrating with respect to crack size,
af da
[3] Nf = ∫
A(ΔK )
ai m

The integration of the equation [3] is typically carried out numerically since ΔK is function of the crack
length, a, and of the correction factor f(g) as shown in [1].
Higher A values in [3] will lead to shorter fatigue lives (in terms of cycles to failure). It is generally
observed (Stephens et al., 2000) that fatigue life estimation is strongly dependent on the initial crack length,
ai, and not very sensitive to the final crack length, af,.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
Over the past several years, values for constants Ap and mp that correspond to the Paris crack growth
expression have been published for different cast irons. These values for different stress ratios, R, are shown
in Table 1. In general, caution should be exercised when comparing constants from different cast irons since
they all have their own metallography, which can significantly influence fatigue properties.

Table 1. Fatigue constants for Paris fatigue expression.

Paris†
Stress exponent, coefficient, Ap
Reference
ratio, R mp −mp (1−m p / 2 )
MPa m
Baicchi et al. (2007) 0.1 6.95 5 x 10-15
0.5 7.53 1 x 10-14
Bulloch (1995) 0.05 6.7 6.12 x 10-16
0.3 6.2 1.35 x 10-14
0.7 6.5 2.59 x 10-12
James and Wenfong (1999)§ 0.1 10.8 2.50 x 10-20
0.2 11.2 1.75 x 10-20
0.3 11.8 7.50 x 10-21
0.4 13 8.50 x 10-22
0.5 13.7 2.10 x 10-22
This study
Paris
Lower bound 7.8 1.29 x 10-16
Upper bound 9.2 5.60 x 10-16
† Unit for stress intensity factor is MPa√m and for length dimension is m.
§ Constants estimated from graphical plots.

2.3 Load spectra and stress histories for cast iron mains
Implicit in the applications of equations [1] (in range form) and [3] that describe fatigue crack propagation,
is the assumption that alternating stress has a constant amplitude. Cast iron water mains can experience
stresses due to operational water pressure, earth loads (loads due to swelling soils and frost heave, if
applicable), pipe self-weight, fluid content, traffic loads and transient (surge or water hammer) pressures.
Earth loads, pipe self-weight and fluid content can be safely assumed to be static while all other types of
loads are random events that occur at different frequencies. In addition, frost loads will be imposed during
winter on pipes buried in soils susceptible to frost heave. Traffic loads and transient pressures are
man-induced and depend largely on usage and operational practices, respectively.
As discussed above, load history and consequent stresses need to be determined to conduct a realistic
fatigue crack propagation analysis. The load history of random events is usually generated from measured
load spectra, i.e., measurement of the frequency with which each load is expected to occur. The load history
of random events can be complex and needs to be reduced to a number of constant amplitude events, which
involves a process called cycle counting. Many different cycle counting methods have been proposed and
Bannantine et al. (1990) provide details on these methods. Broek (1988) commented that it was more
important to get the maxima and minima right than the specific method to count cycles. ASTM E 1049-85
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
(2005) standard that describes the rainflow method for cycle counting was used here together with
AFGROW.
Table 2 shows the six principal different loads or stresses that a buried pipe might experience, of which
earth load, pipe self-weight and fluid content are static while the others have expected frequencies that vary
from several events an hour to several events a month. A truck will have two or more axels and the loads
imposed by each axel on the buried pipe will occur within seconds. For fatigue analysis, it is sufficient to
consider the loads imposed by the two axels as one event since the change in stress in the pipe when the
second axel passes is insignificant. The transient pressure wave event will typically reach peak pressure and
attenuate over a few seconds. As for traffic loads, only the peak pressure is treated as an event. Frost and
swelling loads will raise the mean stress during part of the year if the pipes are geographically located in
areas affected by cold season/frost susceptible soils or rainy season/swelling clays conditions. Contrary to
traffic and transient pressure, seasonal loads will develop over a number of weeks. For simplicity, the frost
and swelling loads are assumed to begin and die out instantaneously, at the start and end of the respective
seasons.
The different loads experienced by the pipe can be combined to simulate loading history. In lieu of
measured load history, a partial list of a total of 24 possible load combinations is shown in Table 3.
Additional load combinations that reflect water pressure fluctuations below and above the mean operating
pressure are not shown here. Frost load, if any, is introduced in the load combination through a factor, fFL,
between 0 and 1 applied to the earth load. The fluctuations in the operating pressure (Δpmo) are represented
through a factor, fmo, between 0 and 1, which is applied to the mean operating pressure (pmo), i.e., Δpmo= pmo
(1 + fmo).
Figure 5 shows a typical stress history that a buried main would experience if earth, frost, traffic and
transients loads are considered. The stresses are exaggerated so that stress variations are visually noticeable.
It is important to note that mean stresses in the cold season are higher than in the warm season only if the
ground freezes and their impact will depend on the properties of native and backfill soils, and on the period
of sustained temperatures below freezing point.

Warm season

Transient
Pipe stress

Operating pressure Traffic

fluctuations
Cold season

Time (hours)

Figure 5. Variations (qualitative) of pipe stress as a function of operational and seasonal loads.

2.4 Crack growth prediction using AFGROW

The software application, AFGROW, developed at the US Air Force Wright Aeronautical Laboratories
(Harter, 2008) is used in this study to conduct fatigue crack propagation analysis. AFGROW has many
features to conduct a comprehensive fatigue analysis but only a few selected features judged to be
applicable to the fatigue analyses of cast iron pipes are discussed here. These features are (i) specifications
of crack size and its location, (ii) fatigue material properties discussed in sections 2.1 and 2.2, and (iii) load
spectrum discussed in section 2.3.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
Table 2 Combinations to generate load spectrum for fatigue analysis of a buried main.
Expected frequency of
Load or stress Comment
event(s)
Earth load, pipe Static Defines mean stress in the pipe
self-weight, fluid
content
Frost load One event/year Assumed to starts as soon as ground begins to
freeze and peaks in mid-winter and reduces to zero
soon after ground thaws. Frost load effectively
raises the mean stress in the pipe during the cold
season.
Swelling load One events /year Starts as soon as the moisture in the surrounding
soil the pipe increases which is typically expected
to happen during the rainy season. Swelling load
effectively raises the mean stress in the pipe during
the rainy season.
Traffic load Several events/hour Frequency will be high during peak traffic hours.
Operational Several events /day Typically operational pressure will fluctuate
pressure several times an hour around the mean operational
pressure.
Transient Several events/ Occurrence of transient pressure in trunk mains is
pressures month largely dependent on hydraulic design and on the
operational issues such as opening and closing of
valves, power-out or failure of pumping station.

Table 3 Expected frequency of loading events for a buried main.

No Load combination Comment
1&2 Earth load, etc, (We (1 + fFL)) + mean Mean (normal) operating conditions.
operational pressure (pmo)
3&6 Earth load (We (1 + fFL)) + mean Operating conditions where water
operational pressure (pmo) ± fluctuations of pressure fluctuates around the mean
operational pressure (Δpmo) operating pressure.

7&8 Earth load, etc, (We (1 + fFL)) + mean Conditions when transient pressure
operational pressure (pmo) + transient occurs with the mains operating at
pressures (pt) normal conditions.
9 & 10 Earth load (We (1 + fFL)) + mean Conditions when traffic passes
operational pressure (pmo) + traffic load overhead with the mains operating at
(Wt) normal conditions.

(i) Crack size and its location: AFGROW allows the consideration of many different options for the type of
structure to analyse, i.e., plate, pipe, etc as well as the location of the crack (defect or flaw) and its size.
However, it lacks the option for pipe sizes typically encountered in the water industry and moreover
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
considers pipes as thin-wall cylinders. Glinka (2009), who developed the f(g) functions implemented in
AFGROW, recommends that it is adequate to treat the pipe wall as a thin plate for practical purposes.
Glinka’s assertion was confirmed by conducting preliminary analyses using dimensions close enough to
that of a thin pipe for AFGROW to consider it as a thick pipe.
(ii) Fatigue material properties: AFGROW allows the consideration of several crack growth models. Each
of these models has its own constants and parameters. In this study, only Walker’s crack growth model with
λ =1 (Paris’ model) was considered. Other material properties required to conducted crack growth analyses
are plane strain fracture toughness, plane stress fracture toughness, threshold fracture toughness and tensile
failure strength. The shift due to stress ratio, R, is considered approximately since sufficient test data were
not available for rigorous consideration of its effects on cast iron.
(iii) Stress (load) spectrum: AFGROW allows the input of one or more sub-spectra to specify
cycle-by-cycle load history. A normalized stress (load) spectrum is generated (as discussed in section 2.3)
from the specified frequencies of changes in operational pressures, pressure transients and traffic loads. The
facility provided by AFGROW to scale the normalized stress (load) spectrum through the stress
multiplication factor (SMF) is used to conduct fatigue analysis.

3. ANALYSIS OF FATIGUE FAILURE IN CAST IRON MAINS

Tables 4 and 5 list all the input parameters used the analysis. These include pipe properties, operational,
trench and traffic conditions, and mechanical properties (including constants for the different crack growth
models). To reflect aleatory uncertainty, most input parameters are provided in the form of most likely
value plus or minus expected variation. The steps followed to conduct the analysis were:
1. Specify pipe size, burial depth, etc.
2. Specify the appropriate Walker’s (if λ = 1 then Walker’s expression reverts to the Paris’ expression)
constants for cast iron. Alternatively, AFGROW allows for tabular input of da/dN vs. ΔK that would be
typically obtained from test results.
3. Specify mean operating pressure and type of traffic loading. Also specify if ground is expected to freeze
or swell, as appropriate.
4. Assume the expected frequency of events for operating pressure fluctuations, traffic loads, and pressure
transients. If the event frequencies are available from field measurements, then these should be used.
5. Determine maximum stresses experienced by the pipe for load combinations indicated in Table 3.
6. Generate a stress history (referred to as spectrum in AFGROW) with the stresses determined in step 5
and expected frequencies specified in step 4. Typically, the stress history is normalized with respect to
the maximum stress.
7. Execute AFGROW with the above data to obtained crack growth development as a function of number
of cycles.
For practical purposes and to reduce computational effort, fatigue crack growth analysis in cast iron pipes
was terminated if the fatigue life (or equivalent number of cycles) exceeds 300 years, as a pipe with a
fatigue life that long would not be of immediate concern.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
Table 4 Pipe properties, operational, trench and traffic conditions.
Most likely value Expected
Parameters
(MLV) variation (EV)

48” cast iron pipe Bell flange Spigot

Inside diameter, mm 1,241.4 1,21 -
Wall thickness, mm 72.2 28.7 5%
Trench width, mm 1,800 10%
Pipe segment length, mm (ft) 2,438 (8’) -

Trench conditions
Pipe burial depth (to crown of pipe), mm 900 10%
Backfill unit weight, kg/m3 1,800 10%
Laying conditions as per AWWA C101-77 A
Coefficient of sliding friction (trench wall & 0.4 -
backfill)

Operational conditions
Mean operating pressure and its fluctuations (%) 3.45 bars (50 psi), 20%
Period of mean pressure fluctuations, per event 12 hour
Design transient pressure and % of design 6.89 bars (100 psi), 60%
transient pressure
Period of transient pressures, per event 1 day

Traffic conditions
No of passing trucks 2
Traffic type Heavy
Period of traffic (truck loads), per event 1/2 hour

The following only discusses the analysis of through crack at bell flange (thickness and outermost part of
bell) of 48” pipe. The stress history generated for 48” diameter cast iron main operating under conditions
and parameters specified in Tables 4 and 5 resulted in hoop stresses in the 5.2 to 10.6 MPa range for the
spigot end and 23 to 42.3 MPa range for the bell flange. It is likely that actual stresses may have been even
higher because typically, aggressive compaction of lead caulking was very likely to put the bell flange in
tension (this tension was not considered in the current analysis). The threshold fracture toughness was taken
as 2 MPa√m.
The stresses had to be increased by about 3.5 times (SMF = 3.5) the estimated stresses under normal
operating loads to obtain fatigue lives of less than 300 years. This indicates that the bell has to be severely
stressed if the bell failure is to be explained by a fatigue mechanism. Table 6 exhibits the results of the
analyses conducted for 48” bell flange as discussed below.
(a) Fatigue life of the 48” cast iron pipe decreases as the length of initial crack length increases.
(b) Fatigue analyses were only conducted for the upper bound of the da/dN vs. ΔK curves since these led to
fairly short fatigue lives. Lower bound da/dN vs. ΔK lead to even higher fatigue lives, which indicates
that pipes with very low values of Ap may deemed to be safe from fatigue failure.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
Table 5. Mechanical properties for cast iron pipes for fatigue sensitivity study.
Most likely Expected
Parameters
value (MLV) variation (EV)
Hyperbolic parameters for cast iron
Elastic modulus, Eo, GPa 120 10%
Tensile strain at failure, σfT, % 0.6 10%
Tensile stress at failure, εfT , MPa 185 10%

Compressive strain at failure, σfC , % 0.9 15%

Compressive stress at failure, εfC, MPa 278 10%

Fracture toughness for cast iron

Plane strain, KIc, MPa√m 12 10%
Plane stress, Kc, MPa√m 24 10%
Threshold fracture toughness, ΔKth, MPa√m 2 20%

Fatigue Paris constants† for cast iron

Fatigue exponent and coefficient, mP, AP (lower bound) 7.8, 1.29 x 10-16 5%, 5%
Fatigue exponent and coefficient, mP, AP (upper bound) 9.2, 5.6 x 10-16 5%, 5%

† Unit for stress intensity factor is MPa√m and for length dimension is m.

(c) The final crack length, af, was first determined to be 26.28 mm based on the mechanical properties
specified in Table 4 and the given stress conditions defined above. The range of initial crack size ratio,
ra, was varied between 99% and 99.5%, which corresponds to range of initial crack lengths of between
26.02 and 26.15 mm, respectively.
(d) The bell flange wall thickness is much higher than spigot wall thickness. Consequently, stresses due to
operational loads at the bell end are much lower than induced in the barrel or the spigot end. However,
as discussed above, lead-caulking action can elevate the stresses beyond what would be induced due to
operational loads alone. The effect of the increase in stresses was incorporated in the fatigue analysis
through the stress multiplication factor (SMF) provided in AFGROW. The analyses show that the bell
end needs to be highly stressed to lead to fatigue life of less than 300 years.
Monte Carlo simulations to determine the variation in the estimates of fatigue lives for the 48” bell flange
were conducted, and results are exhibited in Fig. 15 in the form of frequency distribution (histogram) plots.
Analysis indicates that given a crack length of 26.15 mm, fatigue failures are likely to occur (with a
relatively high likelihood of nearly 60%) within the first 50 years. It is important to note that this relatively
short fatigue life results because extreme conditions were imposed especially in regards to the stresses in
the bell flange.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
Table 6 Estimates of fatigue life for bell flange of 48” cast iron pipe.
Crack growth model Fatigue life as function of initial crack size
Stress range, σ (MPa) and SMF 5.2 to 10.6 and SMF = 3.5
Initial crack size ratio, ra 99.00% 99.25% 99.50%
Paris
Upper bound, Nf (cycles) 1,384,896 1,034,256 687,696
Upper bound, Tf (years) 79 59 39

1.0

0.8
Relative frequency

0.6

0.4

0.2

0.0
0 200 400 600

Fatigue life (years)

Figure 6. Histogram of fatigue life of 48” CI pipe bell flange with Paris model (ai = 26.15 mm).

3. FINAL REMARKS
It is important to note that a cast iron main can only undergo fatigue failure if the pipe has existing cracks or
voids. Previous analyses show that if the cast iron mains were free of cast iron defects due to manufacturing,
then cracks could still have been introduced during the loading/unloading delivery process or during the
lead caulking of the bell-spigot joint, with a high likelihood of not being noticed on installation and
commissioning.
Analysis shows that long fatigue life is assured if an existing crack size is below a certain limit and the cast
iron main is operating under conditions that generate low stresses. Therefore, from an inspection viewpoint,
it is crucial to detect cracks that exceed this length limit if sudden failure (uncontrolled crack growth
propagation) is to be avoided.
The fatigue crack model described in this paper allows the consideration of a wide variety of scenarios. The
analysis described here only consider one specific scenario and a thorough sensitivity study needs to be
carried to reach more conclusive understanding of fatigue growth failures in cast iron trunk mains.
The outcome of the application of crack growth models should be judged from a qualitative aspect. The
nature of the fatigue analysis is based on considerable empirical relationships and, wherever possible,
physical evidence should be sought to confirm fatigue failure. It is recognized that this exercise is not easy
and not always possible to pursue rigorously.
 Water Distribution System Analysis 2010 – WDSA2010, Tucson, AZ, USA, Sept. 12‐15, 2010 
 
Acknowledgment
This research project was co-sponsored by the Thames Water Utilities Ltd and the NRC. The authors wish
to acknowledge the invaluable help provided by Dr. Ahmed Abdel-Akher at the NRC to implement the
algorithms described in this paper.
References
American Society for Testing and Materials (ASTM). (1990). Standard test method for
plane-strain fracture toughness of metallic materials, ASTM E 399-90. ASTM International,
West Conshohocken, PA.
American Society for Testing and Materials (ASTM). (2005). Standard practices for cycle
counting in fatigue analysis, ASTM E 1049-85. ASTM International, West Conshohocken, PA.
Baicchi, P., Collini, L. and Riva, E. (2007). “A methodology for the fatigue design of notched
castings in gray cast iron.” Engineering Fracture Mechanics. Vol. 74, pp. 539–548.
Bannantine, J. A. Comer, J.J. and Handrock, J. L. (1990). Fundamentals of metal fatigue analysis.
Prentice Hall, Englewood Cliffs, NJ.
Bonacuse, P. J. (2001). Fracture of cast iron pipes. Supporting document in Cleveland Trunk Main
Task Force Final Report. City of Cleveland, Division of Water, OH.
Broek, D. (1988). The practical use of fracture mechanics. Kluwer Academic Publishers, Boston,
MA. 522 p.
Brooks, C.R. and Choudhury, A. (1993). Metallurgical Failure Analysis. McGraw Hill, New
York. N.Y.
Bullock, J.H. (1995). “Near threshold fatigue behaviour of flake graphite cast irons
microstructures.” Journal of Theoretical and Applied Fracture Mechanics, Vol. 24, pp. 65-78.
Cast Iron Pipe Research Association (CIPRA). (1927). Handbook of cast iron pipe: for water, gas,
steam, air, chemical and abrasives. Chicago. IL.
Cleveland Water Department. (2001). Cleveland Trunk Main Task Force Final Report. City of
Cleveland, Division of Water, OH.
Deb, A. K., Grablutz, F. M., Hasit, Y. J. Snyder, J. K., Loganathan, G.V. and Agbenowsi, N. (2002).
Prioritizing Water Main Replacement and Rehabilitation. AwwaRF and AWWA, Denver.
Ewalds, H. L. and Wanhill, R.J.H. (1984). Fracture mechanics. E. Arnold, London England.
Handover, G., Wilkins, J., Prescott, W. and James, R.W. (1930). Burst water mains. Report
prepared by members of sub-committee for the Works and Stores Committee of the Board,
Metropolitan Water Board. 119 p.
Harter, J.A. (2008). AFGROW Users guide and technical manual, Version 4.0011.14, US Air
Force Research Laboratory. OH.
James, M.N. and Wenfong, L. (1999). “Fatigue crack growth in austempered ductile and grey cast
irons-stress ratio effects in air and mine water.” Materials and Science Engineering. Vol. A
265, pp. 129-139.
Rajani, B. B., Makar, J. M., McDonald, S. E., Zhan, C., Kuraoka, S., Jen, C.-K., and Viens, M.
(2000). Investigation of grey cast iron water mains to develop a methodology for estimating
service life, AwwaRF and AWWA, Denver.