wrf webcast condition assessment & pipe failure prediction project corrosion … · much new...
TRANSCRIPT
© 2017 Water Research Foundation. ALL RIGHTS RESERVED. No part of this presentation may be copied, reproduced, or otherwise utilized without permission.
WRF Webcast
Condition Assessment & Pipe Failure Prediction Project
Corrosion and Failure Prediction
November 30, 2017
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Advanced Condition Assessment and Failure Prediction Technologies for
Optimal Management of Critical Pipes (#4326)
• Partnership project funded by Sydney Water, WRF, Hunter Water Corporation, City West Water, Melbourne Water, South East Water Limited, Water Corporation of West Australia, South Australia Water, and UKWIR.
• $6M+ of committed funding spread over 5 years, plus over $16M of committed cost-share and in-kind support.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
http://www.waterrf.org/Pages/Projects.aspx?PID=4326
Research Team:
• Monash University
• University of Technology Sydney
• The University of Newcastle
WRF Technical Advisory Committee:
– David Hughes
– Jeff Leighton
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Activities
Activity 1: How, when and where will pipes fail within the entire network?
Activity 2: How do we assess the condition of the pipe cost effectively?
Activity 3: How do we calculate pipe deterioration rates accurately with respect to the pipe environment?
Activity 4: What is the time-dependent probability of
failure along the pipeline?
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Reasons to Listen to These Webcasts
Much new field data and supporting theory/models:
• 37 detailed corrosion analyses from field pipe
• Finite element modeling of pipes
• Traffic loading tests
• Pressure loading tests
• Condition assessment testing with ground-truthing of the results on field pipe
• Key Outcomes:– Large patch corrosion leads to catastrophic failure
– Pipes leak before they break
– Manage internal pressure to extend pipe life
– “Monash Tool” developed - a model for corrosion progression, failure prediction
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Critical Pipe Webcast Series
1st webcast
Corrosion and Failure
Prediction
November 30, 2017
2nd webcast
Condition Assessment for Failure Prediction
December 5, 2017
3rd webcast
Key Outcomes and Sydney Water Case
Study
January 11, 2018
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Today’s presenters
• Robert E. Melchers, Professor of Civil Engineering at The University of Newcastle, Australia
• Jayantha Kodikara, Professor of Civil Engineering at Monash University, Australia
© 2017 Water Research Foundation. ALL RIGHTS RESERVED. No part of this presentation may be copied, reproduced, or otherwise utilized without permission.
Predicting Long-Term ExternalCorrosion of Cast Iron Pipes
Rob MelchersThe University of Newcastle,
Australia
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
9
Outline
• Aims & background
• Factors for corrosion of ferrous metals in soils
• Development of corrosion of cast iron in soils with time
• Practical observations
• Corrosion between first leak and eventual pipe fracture
• Drivers for CI pipe corrosion
• Take-home messages
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
10
Aims:
• Cast iron pipes
• internally cement-lined
• In practice need to know amount of
corrosion of pipe now and the future rate
• Approach:
• Model to predict development of external
corrosion as a function of time and soil
environment
• Model based on science
• Considering also field practices
• Pit depth = main interest
• Calibrate model to actual field data
time
corrosion
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
11
Background 1
• Severe corrosion observed sometimes for steels and cast iron pipes
buried in soils
• Field observations back to 1930s
• Various German and UK studies - reviewed elsewhere
• Numerous model and theoretical studies …. (Cole and Marney 2012)
• Also major field experiments:
• US – NBS (Romanoff 1957) – up to 17 years
• New Zealand study (Penhale 1984) - up to 20 years
• Swedish study (Norin and Vinka 2004) – 4 years
• Canada (NRC, Kleiner et al. 2013) – exhumed pipes, limited soil data
• Currently no satisfactory model for prediction
• i.e. with acceptable margin of error (see Ricker 2010)
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
12
Background 2
• Most field studies of very limited extent
• New Zealand study up to 20 years, but few observations in between
• US NBS study (Romanoff 1957)
pieces of cast iron, small scale model pipes, not operational pipes
considered: soil air pore space, specific gravity, shrinkage, internal
drainage, moisture content, soil resistivity, annual precipitation
soil pH, Na+K, Ca, Mg, CO3, HCO2, Cl-, SO4-
• Did not measure: free water at pipe, organic carbon, nitrates (DIN) ….
• Did not report burial processes, backfill properties ….
• Still the most comprehensive data source….
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
13
US-NBS study 120+ soils US-wide, many irons and steels, up to 17 years exposure,
many soil parameters measured – a truly major, costly exercise….
Sources: Romanoff 1957 Ricker 2010
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
14
Background 3• Tomashev (1966): expect similarities with atmospheric corrosion
- wetter longer => more corrosion
• Gupta & Gupta (1974): wetness of metal surface ≠ soil moisture
• von Wolzogen Khur & van der Vlugt (1934): microbiologically
influenced corrosion (MIC) in soil corrosion
• Melchers & Jeffrey (2013) importance of nutrients for MIC
• Heyn & Braun (1908), Brasher (1967), Mercer & Lumbard (1995):
- dilute salt solutions have no significant influence on corrosion
in (near-) stagnant conditions
- soil moisture is almost stagnant -> soil chemistry little relevance
• Backfill is contact with pipe: usually ≠ the native soil profile
• Corrosion under non-uniform deposits / metal contact can be severe
• Different interpretations to earlier efforts – new issues
• Most data in the literature is insufficient….
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
15
Our study: • Develop prediction model for amount of corrosion and future rate
• Based on sound theory + calibrated to data from real pipes
• Developed a new soil data collection protocol with Hunter Water,
Sydney Water
• Exhumed some 37 pipes
• Pipe surfaces digitally scanned
• Data interpreted (see publications)
• Soil data collected and correlated
Outcomes:
• First-pass model developed ….
• New insights from discussions with field staff, from data
interpretations, cross-disciplinary ideas … .
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
16
Modelling approach
• Corrosion development = bi-modal trend
• Steels, cast iron, aluminium alloys, copper alloys
in a wide variety of exposure conditions (see Corrn Sci, Corr J, etc.)
including soils (see Aust Corr confs 2014-2016)
• Long-term trend – linear but not through origin
• Calibrate: using field data from present project ( + Romanoff )
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
Observations:
Figure 2. Observed corrosion penetrations p0 versus exposure period t.
These are all clay soils – light to heavy
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
First cut interpretations
Figure 2. Observed corrosion penetrations p0 versus exposure period t.
wet
wetwetwet
poor
backfill
Rocks
in fill
These are all clay soils – light to heavy
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
19
Time of wetness in soils
• Corrosion can occur only when metal is sufficiently wet
Sands, loams: borrow ‘time of wetness’ concept - atmospheric corrosion
• Effective wet time: Gupta & Gupta (1979):
• Low moisture content (mc) => no corrosion (moisture held in soil)
• Sharp rise in corrosion as soil mc reaches
65% water holding capacity
• Subsequent drop = lack of O2 in
tests - ignore - long-term corrosion
requires little O2.
Clays: usually wet at depth
• Currently: expts for Gupta effect
• Appears similar to loams
Effect: substitute effective time of wetness for real time ….
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
Mean trend and uncertainty bounds
Figure 6.Corrosion penetrations p0 with allowances for soil nitrate content and soil
chloride content versus time of wetness tw. Mean value and percentile trend lines are
shown for tw>20 years.
tw
p
rs
cs
SD2
SD1
Model
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
21
Influences: 1. Microbiological effects
• Microbiological Influenced Corrosion increases pit depths
• No information for soils: but seawater: nutrients critical
• Most nutrients present in soils except Dissolved Inorganic
Nitrogen (DIN)
• Affects mainly long-term corrosion
• Fertilizers, sewage effluent
• Adjustment to data ….
pit depth
adjustment
Pit depth
(mm)
Time (yrs)
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
Imperfection (pit) 1-2 mm depth
Imperfections (pits) < 1 mm depth
Influences: 2. Imperfections in cast iron •Observations show imperfections always present
•Measured as ‘pits’ even though they are not
•May affect corrosion behaviour
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
23
Influences: 3. Physical damage Rock and hard materials
Could damage casting layer - increase corrosion
Requires further investigation
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
24
Influences 4. Soil
inhomogeneities
Causes localized corrosion under
contact points.
Example: Rope prints =>
Soils: Lumpy clay in back-fill:
- would explain patterns seen on
pipes
- never previously considered
- on-going research activity
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
25
Soil inhomogeneity (e.g. stiff clays)
• Romanoff data re-
analysed and re-
interpreted
• Worst pitting corrosion
for stiff clays, etc.
• Stiff clays as backfill
produce
inhomogeneities at
surface.
• Sands = good!
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
26
Soil inhomogeneity (e.g. stiff clays)
• Stiff clays backfilled in trench -> air voids, also at cast iron surfaces
• When wet, produces differential aeration = localized corrosion
• Occurs under clay-iron interface (or rope-iron interface)
• Mechanism =
• Consistent with field observations
• Consistent with corrosion theory
• Not previously investigated
• Corrosion decreases for finer materials, lower wetness time
• Current research effort: prediction / measurement of inhomogeneity
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
27
From pitting to pipe failure
• Fracture requires sufficient thinning over sizeable areas of pipe wall
• Corrosion mostly produces pitting, some deep pits
• Pitting occurs in progressive steps
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
28
Field observations
• Most corrosion along bottom of pipes (i.e. along bedding)
• Larger areas of thin pipe wall – what causes that? – not pitting …
• Observations of failed regions of pipes …
• Corroded surface topography differs around failure area (i.e. near
fracture crack)
• ‘wide and deep’ corrosion (pitting?) only where cement lining is visible
• Elsewhere no sign of wide scale corrosion
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
29
New explanation of progression to failure
• In time, an extreme pit perforates pipe wall from the outside
• Small area of cement lining exposed = permeable
• Inside pipe: fresh oxygenated water under pressure – water forced out
• Diffuses through cement lining, behind graphitized layer, outwards
• Fresh water, O2 rich, corrodes pipe wall area in the neighbourhood
• About r0 = 0.2mm/y (e.g. 1 mm every 5 yrs >> rs long term rate)
• Details now in the literature.
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
30
Drivers for CI pipe corrosion: Moisture
•sands and loams - better drainage of trenches,
•clays – tend to be ‘impermeable’ - nothing can be done, except well-
placed and drained sand surround,
•avoid depressions in alignment, location near water courses, etc …
Cast Iron - material surface defects could contribute to pitting
Nutrients - for MIC – caused by fertilizers, sewage plant effluent
•not always able to access pipe at depth
Backfill homogeneity and compaction
•Size of voids in backfill causes increased localized corrosion
•Corrosion usually most severe at 6 o’clock position – pipe bedding
issues, bedding tends to hold water, compaction less effective
•Poor placement higher up – 10 and 2 o’clock positions – “soil jam”!?
Pipe leakage could be a warning sign of future pipe fracture
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
31
Take-home messages
• Corrosion is not a linear function through the origin => a corrosion
‘rate’ is highly misleading
• (native) soil properties and soil chemistry are of little importance
• Microbiological corrosion - not always easy to achieve at normal pipe
depths (nutrient supply issue)
• Back-fill physical properties are important:
homogeneity, local damage, bedding, soil around pipe,
quality of workmanship
• Critical issues: prediction of what is going on now, and what is likely to
happen in the future
• Sound prediction requires sound models based on sound science and
calibrated to real data – the alternative is hocus pocus….
• We are aiming for sound prediction.
The University of Newcastle, Australia - Centre for Infrastructure Performance and Reliability
32
Estimation of ‘time of wetness’
• Time of wetness not recorded by any investigation
• Need to estimate it …
• Approach adopted:
• Simple bucket model for climate soil moisture – averaged over 1 y.
• Count number of days mc > critical point
• Calculate % time of wetness
• Multiply by exposure period
• => ‘time of wetness’ over life of pipe
• Model needs:
- soil properties
- rainfall
- evapotranspiration
- drainage factors …
© 2017 Water Research Foundation. ALL RIGHTS RESERVED. No part of this presentation may be copied, reproduced, or otherwise utilized without permission.
Condition Assessment & Pipe Failure Prediction Project
Failure Prediction
Prof. Jayantha KodikaraMonash University, Australia
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Deterioration and failure process of CI pipes
375 mm (14.7”) diameter, 80 year old cast iron pipe
Rajani et al. (2014)
www.theage.com.au (December 16, 2014)
Considered onlyLarge diameter pipes
Diameter ≥ 300mm (12”)
Factor of safety reduces with time.
Failure starts as a leak inmost cases
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Failure locations and modes
What is a pipe failure? A major break (or burst) leading to uncontrollable loss of water that needs immediate attention.
Mode 2: Fracture initiated from the joint, not necessarily associated with corrosion
Mode 1: Fracture / Failure in corroded patches, mostly in pipe barrel
Sydney Water Harris Street failed pipe
Rathnayaka (2016)
Rathnayaka (2016)
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
External loads (traffic and soil) – Field-scale tests
4570
450
65
0
4800
9203660 220
Pipe depth at 1080 mm to pipe crown
17
60
90
0
1140
Location 3 (Displacement transducers)Location 2
(Strain gauges)
Wheel path
2140
Location 1(Displacement transducers)
Road pit
13
60
3200
Foot path
65
018001400
Pipe depth at 1000 mm to pipe crown
Pipe depth at 870 mm to pipe crown
17
00
Moisture probes
16
001400
60
0Logger cabinet
N
Nature strip
Location 4 (Strain gauges)
Road
Road kerb
Monitoring locations at test bed
Pipe exposure
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
External loads – Field-scale tests
Sydney Water test bed site
ExcavationCondition assessment
Instrumentation –Joint rotation
Instrumentation – Pressure pads
Truck testing
Instrumentation – strain sensors
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
➢ Different trucks with different axle
configurations were used.
➢ Three different road surface types
were tested (road base, bitumen
and concrete).
➢ Different types of tests (passing
and braking at different speeds,
cornering, passing over a speed
hump and stop each axle on pipe).
➢ Pipe was pressurised to a
maximum pressure of 670 kPa
(97psi).
Video 1 – The truck passing
over a speed hump
Sydney Water Test-bed truck testing
External loads – Field-scale tests
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Strain measurements (for all tests) Overburden pressure (for all tests)
Maximum strain due to truck testing was ~20 microstrain road and maximumground pressure on pipe crown was 27kPa (3.9psi) for bitumen surface
Sydney Water Test bed truck testing
External loads – Field-scale tests
10 kPa = 1.45 psi
1 kN = 224 lb
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
➢ Pressurised from 0 to 670 kPa (97 psi) in approximately 100 kPa increment
➢ Depressurised from 55 to 30 m then from 30m to 14 m in approximately 5
m decrement
0
100
200
300
400
500
600
700
0 20 40 60 80 100 120 140
-900
00
450
900
1200
Microstrain
Pre
ssu
re (
kP
a)
Maximum strain due to internal pressure was 126 microstrain
15°30°
SG01
SG03
SG04
SG05 SG06SG07
SG08
SG09
SG10
SG11
SG12
SG13
SG02
10 kPa = 1.45 psi
Test bed pipe pressurisation – Strain measurements
External loads – Field-scale tests
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Summary of external load tests on cast iron pipe
➢ The change in pipe hoop strain due to internal water pressure showed that undernormal operational pressure, the pipe strain is within the elastic region with amaximum strain of about 126 microstrain.
➢ The change in strain from truck passing and braking tests are similar withdifferences less than 20 microstrain.
➢ The changes in strain from truck tests with 0 and 50 m (164 ft) internal pressureare very similar with differences of few microstrain only.
➢ Truck test results showed small differences in hoop strain measured fromdifferent tests (i.e., passing, braking, stopping).
➢ Pressure differences between braking and other tests are less significantat the depth of 860 mm (34”) (less than 5 kPa).
➢ The use of a uniform pressure of about 25 kPa (3.63psi) on the pipe leveldue to traffic load is conservative.
➢ On road base, the pressure may rise to about 40 kPa (5.8psi) on the pipecrown, as applicable to road works.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
3. Internal loads (water pressure and
pressure transients)
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Internal loads – water pressure and pressure transients
• Internal water pressure is the major contributor to pipe stress in large diameter pipes.
• Pressure surge events can significantly increase pipe stress.
• Pressure and transient modelling and monitoring undertaken in selected networks.
• This is to calibrate past failures and show future prediction capability.
• Direct link was established between pressure transients and pipe failure during this study.
Radcom data logger – 25Hz
Some logger installation locations
Fire hydrants
PRV
Pump station
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Internal loads – water pressure and pressure transients
Pressure Monitoring: Cardiff South-Pump start-ups and pipe failure
Pipe
failure
5s
10s 20s30s
5s
30s5s
5s
Significant steady state pressure rise was measured for the pumps runs in night (low demand) and a
pipe failure occurred after 13 pump events.
10 kPa = 1.45 psi
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Internal loads – water pressure and pressure transients
Pressure transient
hot spots
Pressure transient maps developed from numerical modelling
Pressure transient maps can be valuable tool to identify pipe failure hot spots.
Imported pressure transient data into
pipe asset database (units kPa)
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Pressure transient amplification in reticulation
Internal loads – water pressure and pressure transients
Numerical model predicted pressure transients could amplify when they enter
reticulation pipelines near the origin of the transient event (e.g. pump station).
10 kPa = 1.45 psi
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Summary of the study on internal loading
➢ Severe pressure transients from multiple start-ups, shut-downs or automated controls are a major concern for pipe bursts.
➢ Routine pressure transients may not lead to immediate pipe failure, but can contribute to damage accumulation and eventual failure.
➢ Non-routine pressure transients may also lead to pipe bursts (e.g. pressure transient generated due to a pipe burst at a different location).
➢ Generated pressure transients can dissipate rapidly with distance away from the pressure transient generation site (within 2-3 km in many cases).
➢ Reasonable accuracy between monitoring data and models was found.
➢ Transient modelling data can be used for pipe failure predictions and pipe asset management.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Pipe materials and strength
• Key material properties are:– Tensile strength – determines the initiation of
failure or leak– Fracture toughness – determines the break (burst)
possibility
• Pipe material properties depends on:– Mode of manufacture (pit cast, spun cast etc.)– Place of manufacture– Manufacturing defects
• Pipes can be classified into cohorts based on
above. And cohort properties can be used in the
failure analysis. This will also provide strategic
directions for where to put effort in material data
collection.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Evidence of failure mechanisms
1. Evidence from failed pipes – forensic analysis
2. Evidence from unfailed exhumed pipes
3. Evidence from theoretical analysis
4. Evidence from pipe burst testing
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Initial fracture – aftermath of small holes
• When corrosion progresses, basal failure is inevitable since stress increases exponentially.
• However, for small pits, this may happen almost at through-wall corrosion.
• These are typically up to about 20 to 30 mm corrosion pits. So small pits not a problem?
• But how about larger patches?
5
10
20
30
40
50
60
70
80
90
100
% of metal loss
Scale: 1 grid = 50mm
Example of a large
patch (3D scan data)
Example of a large patch
(Actual patch)
Patch size:100mm
x 120mm )
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Initial fracture – aftermath for large patches
▪ When corrosion progresses, basal failure of large patches creates a
fracture.
▪ If the patch is large enough this facture would allow water leakage,
which could also accelerate the corrosion laterally.
▪ If the patch is not large enough corrosion would progress normally
until the patch becomes large and then leaks.
▪ Once a fracture is created high stress concentrations occur at the
crack edges and under cyclic loading (transients or traffic), cracks
can propagate further sub critically.
▪ Accelerated corrosion could also weaken the crack front (stress
corrosion).
▪ Finally when the crack length is critical spontaneous break would
occur. This mechanism is referred to Leak Before Break.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Examination of failed / unfailed pipes
Failure occurred at large corrosion patches
Small corrosion pits does not cause failures
Hunter Water case study
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Laboratory pipe burst experiments
Monash University pipe testing facility
Capabilities
• 5 MPa (725 psi) static pressurising capability
• 1.2 MPa (174 psi) cyclic pressurising capability (to simulate pressure transients)
• Fully automated
• ~2000 pressure cycles per a day
• Pressure, flow (through a leak), strain measurements
Test pipe piece
Pressuring system
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Laboratory pipe burst experiments A summary of pipe burst tests results
Test A pipe Test B pipe
Test C pipe Test D pipe
▪ In all cases, a large corrosion patch with deep
corrosion was required cause failure.
▪ Natural corrosion patch of 100mm x 120mm
(3.95” x 4.7”) size with 72% wall thickness.
reduction failed at 3.6 MPa (520 psi), Approx.
pressure six times higher than typical operating
pressure for the pipe tested.
▪ In all tests, a crack was initiated at the base of
the critical corrosion patch leading water
leakage (Leak-before-break-LBB).
▪ Flat base corrosion patches tend to provide
reduced LBB window compared to corrosion
patches with curved corrosion profiles.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Laboratory pipe burst experiments
Video 2 – Pipe burst in Test 3
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
LBB concept for CI pipes
• Therefore, LBB is applicable for cast iron pipes.
LBB examples from laboratory tests
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
LBB concept for CI pipes
Estimating LBB time window
Corrosion patch design
Generating a through crack
Crack grow to the critical length
Severe corrosion damage
Leak
Burst
Validation tests Field observation
Window period of LBB
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
LBB concept for CI pipes
➢ Previously leak detection used for reducing water wastage.
0
50
100
150
200
250
300
350
0 500 1000 1500 2000 2500 3000
Aver
age
leak
rat
e (L
/min
)
Cycles, N
Full-scale fatigue tests
Measured
Upper
limit_simulation
Lower
limit_simulation
➢ Leak monitoring may be used to prevent pipe failures, further research is needed.
10 L/min = 0.353 cfm
Detectable leak
Leak-before-break (LBB)- full-scale tests
Video 3 –Fatigue crack growth
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Concluding remarks
1. ACAPFP project mainly focussed on failure mechanisms of pipe barrel and to a lesser extent of joints of large diameter pipes (>300mm).
2. Traffic and soil loads - On the basis of very comprehensive Strathfield traffic load testing on SW test bed pipe, it was found that the influence of traffic load is about 1/6 or less in comparison to that due to internal water pressure for longitudinal failure (or splits). So more effort is needed to be placed on water pressure influence.
3. Water pressure and transients - On the basis of pressure transient measurements and modelling and case studies, it was found that pressure transients can be a major cause for pipe bursts. However, there was direct evidence that bursts may not occur in one transient but after a number of transients (direct evidence from Hunter Water pipe failure during testing). Leakage can occur prior to failure.
4. Material properties - Main material properties for pipe failure assessments are: pipe material tensile strength (initiation of fracture or leak) and fracture toughness (burst). Pipes can be put into to cohorts for assessing these properties and general failure patterns.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED.
Concluding remarks
5. Pipe failure mechanisms – For large diameter cast iron pipe barrels, a large corrosion patch with significantly deep corrosion (typically over 80%) is required to cause failure at operational pressures. The first failure was observed as a fracture at the base of the patch that eventually lead to water leakage. Major longitudinal fracture occurred at higher pressures when crack grew to critical length.
6. LBB Concept - The above Leak Before Break (LBB) concept was developed on the basis of failed and unfailed corroded pipes, burst testing and numerical modelling, but further research is needed.
© 2017 Water Research Foundation. ALL RIGHTS RESERVED. No part of this presentation may be copied, reproduced, or otherwise utilized without permission.
Q&A
© 2017 Water Research Foundation. ALL RIGHTS RESERVED. No part of this presentation may be copied, reproduced, or otherwise utilized without permission.
Thank You
Comments or questions, please contact:
For more information visit:
www.waterrf.org