Numerical Investigation of Failure Mechanisms of Cast Iron Watermains

NUMERICAL INVESTIGATION OF
FAILURE MECHANISMS OF CAST IRON
WATERMAINS
Kasuni Liyanage
Master of Engineering Candidate
Faculty of Engineering and Applied Science
Memorial University of Newfoundland
Advisor: Dr. Ashutosh Dhar
DEPARTMENT OF CIVIL ENGINEERING

Outline
• Introduction
• Background
• Objectives
• Erosion voids with rigid localized support
• Partially supported bedding with flexible localized
supports
• Effect of a corrosion pit
• Conclusion
• Recommendations for future work

Water main break sends water shooting 30 feet high in
Brooklyn (2015)

Water main break sends debris flying in Toronto (2011)

A underground pipe ruptured and sent water
gushing out of the ground like a geyser in Toronto’s
east end on Saturday, March 23, 2013.
Toronto’s aging infrastructure costing
millions(2012)
City water leaks wasting millions of
tax dollars in Toronto (2011)

Introduction
• Rapid growth in water distribution networks in
1890s
• During 1870 to 1920 use of cast iron for water mains
was 100%
• Current age is 96 to 146 years to date
• Average estimated service life of cast iron pipelines
is 105 to 135 years
• “Dawn of the replacement era”

Water main break
• 240,000 water main breaks per year in the USA
• Estimated water loss: 1.7 trillion gallons per year
• Financial cost: $2.6 billion per year

Failure modes
Circumferential break Longitudinal splitting Blowout holes
Spiral cracking Bell shearing

• Most common failure mode - circumferential cracking
• Circumferential failure mechanism of the pipe is not well understood
• Conventional methods of pipe analysis predict higher circumferential
stresses
50.4%
27.9%
15.2%
53.6%
2.6%
49.9%
17.4%
11.9%
47.4%
0.0%
3.4%
11.2%
17.0%
43.5%
4.7%
21.4%
79.5%
11.0%
15.2%
16.7%
32.8%
25.0%
14.6%
28.6%
0.0% 20.0% 40.0% 60.0% 80.0% 100.0%
CI
DI
PVC
CPP
Steel
AC
% failure for each material
Other or Unknown
Pitting Corrosion
Longitudinal crack
Circumferential crack

Failure causes
• Pitting corrosion
• Loss of Bedding Support
• Localized Concentrated forces
• Other types of corrosion: Uniform, Tuberculation,
Galvanic, crevice
• Manufacturing Defects
• Human Error

Pitting corrosion
One of the most dangerous forms of corrosion due to
difficulty in detecting, predicting and designing against

Research with corrosion pit
• Corrosion pits influenced the localized strain distribution
and produced significant stress concentrations
• Larger pit sizes are vulnerable to circumferential failure
(Makar et al., 2005)
• Thin wall pipes with pitting corrosion exhibits more
vulnerability to circumferential failure than thick wall pipes
(Makar et al., 2005)
• Pit cast pipes, with a corrosion pit size of greater than
40mm to 60mm , having an unsupported length of 3m
could exceed failure strains (Makar et al., 2005)

Lack of bedding support
Pipe cracks or holes creates leakage of water that
erodes the bedding

Research with lack of bedding support
• Larger diameter watermains with longer unsupported
lengths produced higher stresses. (Rajani and Tesfamariam
2004)
• Peak tension was in circumferential direction for pipes
under lack of bedding support. Stresses increased when
the erosion void length, angle or depth increases, or the
pipe thickness decreases (Balkaya et al. 2012)
• Larger void sizes created higher bending moments in the
pipe wall. Maximum moments occured at the vicinity of
the void (Kamel and Meguid 2013)

Localized concentrated forces
As the surrounding bedding erodes away, the fine soil particles
escape leaving the bulky coarse soil particles. These coarse
particles conglomerate locally and develop a localized
concentrated support to the pipeline

Objectives
1. To study the behaviour of stresses in cast iron water
mains subjected to partially supported bedding
condition using three dimensional finite element
analyses.
2. To examine the stresses in cast iron water mains
subjected to non-uniform bedding and localized
concentrated forces.
3. To investigate the effect of pitting corrosion of
buried cast iron water mains using numerical
modelling.

Erosion Voids with Rigid Localized
Support
900 450 22.50
Invert
Springline

Finite Element model
• Performed using ANSYS v15.0
• Nonlinear 3D coupled soil-pipeline structure
• Validated using the analytical solution (Rajani &
Tesfamariam 2004)
SOLID186
SOLID65
CONTA174
TARGE170
FE Modelling

Material Properties
Item Pipe – Cast Iron Soil – Medium sand
Behaviour Linear elastic Isotropic elasto-plastic
Young’s modulus 206 GPa 20 Mpa
Poisson’s ratio 0.26 0.25
Density 7850 kg/m3 2344 kg/m3
Friction angle - 320
Dilatancy angle - 250
Cohesion - 0.5 kPa

Boundary conditions

Effect of void geometry without local
supports – Longitudinal stresses
• Void at invert Void at springline
Longitudinal
stress (MPa)
22.5
45
90
90
22.5
45 22.5
45
90
Longitudinal
stress (MPa)
Symmetric Unsymmetric Symmetric Unsymmetric
Void at Invert Void at Springline

Effect of void geometry without local
supports – Circumferential stresses
• Void at invert Void at springline
Symmetris Unsymmetric Symmetric Unsymmetric
Void at Invert Void at Springline
Circumferential
stress (MPa)
22.5
45
90
Circumferential
stress (MPa)
22.5
45
90

Comparison between longitudinal and
circumferential stresses
Longitudinal stress (Max) Hoop stress (Max)
Void at
invert
Void at
springline

Effect of void geometry with rigid
local supports
Longitudinal
stress (MPa)
22.5, 45 &
90
overlapped
22.5
45
90
Circumferential
stress (MPa)
22.5, 45 &
90
overlapped
22.5
45
90
Symmetric Unsymmetric Symmetric Unsymmetric
Circumferential stress Longitudinal stress

Partially Supported Bedding with
Flexible Localized Supports

Material Properties
Item Cast iron Pipe Medium Sand Soil
Unit weight 77 kN/m3 23 kN/m3
Modulus of elasticity 206 GPa/138GPa 24 MPa
Poisson’s ratio 0.26 0.25
Friction angle - 38o
Cohesion - 0.5 kPa
Dilatancy angle - 15o

Evaluation of Analytical Solution
Analytical solution developed using Winkler pipe-soil
interaction model (Rajani and Tasfamariam 2004)
Unsupported
region
Plastic
region
Elastic
region

Foundation modulus
• The reciprocal of the flexural characteristic length and is
defined as,
• Where k’s is the elastic foundation modulus,
• The factor ‘0.65’ (called herein as ‘’) in the equation is
varied to provide a better match of the results with those
obtained using 3D finite element analysis.

Comparison of flexural stresses from
analytical model and 2D FE analysis
2D FE modelled using BEAM188 elements for pipe and
COMBIN14 elements for soil
Unsupported
region
Supported
region
-20
-10
0
10
20
30
40
50
0 1000 2000 3000 4000 5000
Longitudinalstress(MPa)
Distance along the pipeline length from midspan (mm)
Rajani and Tesfamariam 2004
2D FE

Comparison of flexural stresses from
analytical model and 3D FE analysis
3D FE modelled using SOLID186 elements for pipe and
SOLID65 elements for soil
-20
-10
0
10
20
30
0 1000 2000 3000 4000 5000
Distance along the pipeline length from midspan
(mm)
3D FE
Analytical (a=0.65)
Analytical (a=1)
Analytical (a=1.5)
Analytical (a=2)
-20
-10
0
10
20
30
40
50
60
0 1000 2000 3000 4000 5000
(mm)
3D FE
Analytical (a=0.65)
Analytical (a=1)
Analytical (a=1.5)
Analytical (a=2)
3D FE
Analytical (=0.65)
Analytical (=1)
Analytical (=2)
3D FE
Analytical (=1)
Analytical (=2)
Thick wall pipe Thin wall pipe
Void depth of 200mm

Flexural stress (Cont.)
-20
-10
0
10
20
30
40
50
60
0 1000 2000 3000 4000 5000
(mm)
3D FE
Analytical (a=0.65)
Analytical (a=2)
Analytical (a=5)
Analytical (a=10)
-15
-5
5
15
25
0 1000 2000 3000 4000 5000
(mm)
3D FE
Analytical (a=0.65)
Analytical (a=2)
Analytical (a=10)
Thick wall pipe Thin wall pipe
Void depth of 50mm
3D FE
Analytical (=2)
Analytical (=5)
Analytical (=10)
3D FE
Analytical (=2)
Analytical (=10)

Flexural stresses with elastic and
elastio-plastic soil models
-10
-5
0
5
10
15
20
0 1000 2000 3000 4000 5000
(mm)
Elasto-plastic soil
Elastic soil

Effect of Void
Thin and thick wall pipes with two different void thicknesses
(50 mm and 200 mm, respectively) and three different void
configurations are investigated for different pipe material
moduli
180o void at invert 90o void at invert 90o void at haunch

Wall
thickness
Elastic
modulus of
pipe
Void
angle
Void
location
Void
thickness
(mm)
Maximum
circumferential
stress (MPa)
Maximum
longitudinal
stress (MPa)
10 mm 206 GPa 180o Invert 200 6.4 -0.1 13.2 -13.1
10 mm 206 GPa 180o Invert 50 5.8 0.7 9.3 -8.8
10 mm 138 GPa 180o Invert 200 6.2 0.1 11.4 -11.2
10 mm 138 GPa 180o Invert 50 5.7 0.9 8.1 -7.5
10 mm 138 GPa 90o Invert 50 5.1 1.0 5.6 -4.4
10 mm 138 GPa 90o Haunch 50 12.0 -5.4 4.6 -3.3
5 mm 206 GPa 180o Invert 200 17.0 -4.6 22.3 -19.6
5 mm 206 GPa 180o Invert 50 15.3 -1.9 16.4 -12.1
5 mm 138 GPa 180o Invert 200 15.9 -3.4 19.3 -16.3
5 mm 138 GPa 180o Invert 50 14.2 -0.9 14.4 -10.1
Comparison

Effect of flexible localized
concentrated supports
Localized support was modelled as elastic springs using
COMBIN14 elements. A 90o void with a void thickness of 50
mm at the invert of the pipe is considered. Elastic moduli of
both the pipe and soil are taken as 138 GPa and 24 MPa,
respectively.

Stresses due to localized supports
• A spring constant of 1500 N/mm caused an increase in
stress of about 30 to 40 times.
• However, the peak stress is in circumferential direction
-300
-100
100
300
Longitudinal stress (MPa)
Circumferential stress (MPa)

Effect of a Corrosion Pit
FE mesh is refined near the corrosion pit

Material Properties
Pipe material (cast iron) properties Soil Properties
Density 7850 kg/m3 Density 2344 kg/m3
Elastic modulus 206GPa, 138GPa,
70GPa
Elastic modulus 24MPa
Poisson’s ratio 0.26 Poison ratio 0.25
Friction angle 38o
Dilatancy angle 15o
Cohesion 0.5kPa

Pipe stresses
Circumferential direction
Longitudinaldirection
Uniform bedding Non-uniform bedding
CircumferentialstressLongitudinalstress

Effect of void with a corrosion pit
Wall
thickness
(mm)
Pipe
Modulus
(GPa)
Void
angle
Void
depth
(mm)
FOS in
Longitudinal
direction
FOS in
circumferential
direction
10 138 180 200 9.70 28.53
10 138 180 50 14.53 28.17
10 138 90 200 15.84 36.86
10 138 90 50 25.28 30.21
10 70 90 50 34.27 26.54
5 138 180 50 9.50 7.65
5 138 90 200 9.44 10.39
5 138 90 50 14.17 9.04

Effect of Material Stiffness
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
5 10 15 20 25 30
Ratioofstressesin
Longitudinaland
circumferentialdirections
Relative stiffness (R)
Pit size 40mm
Pit size 50mm
Pit size 60mm

Effect of localized concentrated supports
with a corrosion pit
Wall
thickness
(mm)
Pipe
Modulus
(GPa)
Void
angle
Void
depth
(mm)
FOS in
Longitudinal
direction
FOS in
circumferential
direction
10 138 180 200 0.924 2.934
10 138 180 50 0.957 2.971
10 138 90 200 0.981 3.045
10 138 90 50 0.989 3.048
10 70 90 50 0.994 3.181
5 138 180 50 0.274 1.080
5 138 90 200 0.278 1.108
5 138 90 50 0.278 1.117

Comparison - with and without Pit
-5
0
5
10
15
30
210
60
240
90
270
120
300
150
330
180 0
Pit + Non-uniform bedding
Pit
Non-uniform bedding
Pit with uniform bedding
Non-uniform bedding without pit
Pit with non-uniform bedding
0
10
20
30
30
210
60
240
90
270
120
300
150
330
180 0
Pit
Pit+non
non-unif
Pit with uniform bedding
Non-uniform bedding without pit
Longitudinal stress Circumferential stress

30
130
230
330
30
210
60
240
90
270
120
300
150
330
180 0
data1
data2
Non-uniform bedding with
localized support without pit
and localized support
-40
-20
0
20
40
60
80
100
30
210
60
240
90
270
120
300
150
330
180 0
Pit+Non
Non-un
Non-uniform bedding with
localized support without pit
and localized support
Longitudinal stress Circumferential stress
Comparison - with localized support

Conclusion
• Higher stresses in pipe wall is caused:
– in thinner pipe,
– for higher relative pipe stiffness
– larger erosion voids,
– larger corrosion pits and
– With localized concentrated forces.
• Peak stress in longitudinal direction for:
– non-uniform bedding support with
• Higher void size (width and depth)
• Symmetric void shape
• Higher relative stiffness of pipe w.r.t soil

Conclusion
• Corrosion pit with non-uniform bedding and localized
concentrated support with a 400N/mm of spring constant may
lead to circumferential cracking
• Localized concentrated support can increase the stresses in the
pipe by 25 to 100 times depending on its rigidity.

Recommendations for future work
• Investigate the three dimensional stresses of large
diameter cast iron water mains.
• Conduct field monitoring to obtain more information of
failure mechanisms and causes.
• Integrate experimental investigation of failure of cast
iron water mains with different flexural stiffness of the
pipe.
• Incorporate the effect of seasonal fluctuations as a
temperature induced force.
• Account for long term material behaviour of bedding
soil and asses the influence of concentrated forces on
the pipe wall stresses.

Acknowledgements
• Dr. Ashutosh Dhar – Supervisor
• Research and Development Corporation of
Newfoundland and Labrador – Financial
Support
• Faculty of Engineering and Applied Science
• School of Graduate Studies
• Canadian Geotechnical Society
• MUN Writing Center

THANK YOU!

Numerical Investigation of Failure Mechanisms of Cast Iron Watermains

Recommended

Recommended

More Related Content

What's hot

What's hot (19)

Similar to Numerical Investigation of Failure Mechanisms of Cast Iron Watermains

Similar to Numerical Investigation of Failure Mechanisms of Cast Iron Watermains (20)

Numerical Investigation of Failure Mechanisms of Cast Iron Watermains

Editor's Notes