myths and realities in projecting the trend of future urd cable failures
DESCRIPTION
Myths and Realities in Projecting the Trend of Future URD Cable Failures. A Comparison Study Presented by Dan O’Neill To the EEI TD&M Meeting On April 4, 2006 In Houston, Texas. Agenda. The typical problem A comparison of URD programs The myth of the ‘tsunami’ dispelled - PowerPoint PPT PresentationTRANSCRIPT
Myths and Realities in Projecting the Trend of Future URD Cable FailuresMyths and Realities in Projecting the Trend of Future URD Cable Failures
A Comparison Study
Presented by Dan O’Neill
To the EEI TD&M Meeting
On April 4, 2006
In Houston, Texas
2
AgendaAgenda
The typical problem
A comparison of URD programs
The myth of the ‘tsunami’ dispelled
How to address the problem
Observations and key questions
3
Left alone, URD cable failures can be an accelerating problemLeft alone, URD cable failures can be an accelerating problem
For example, in Company A:
• Outages were increasing at 5-6% per year, which means doubling every 12-14 years
• Repair costs were averaging tens of million$ per year, and growing at the same rate
• Customers were experiencing multiple interruptions, with some averaging 3-4 per year on their half-loop, not counting upstream outages from feeder lockouts, trees, etc.
• Replacement spending had been very little, and needed to ramp up to sufficient levels to arrest the growth in outages, and would have to grow to keep up with deterioration
Company A needed to fund a URD cable replacement program that would
arrest the growth of outages and maintain the level like Company B did
Company A needed to fund a URD cable replacement program that would
arrest the growth of outages and maintain the level like Company B did
* URD – Underground residential distribution – typical way of serving a post-1960’s residential subdivision, i.e., 300-5000 feet of usually single-phase, 12-34kV primary voltage cable, direct-buried (not in conduit), connecting 1-30 padmount transformers per half-loop, with 2-10 customers per transformer, so about 50 customers per half-loop (from riser to ‘normally open’ point)
URD Cable Failures
0
1,000
2,000
3,000
4,000
5,000
6,000
19
82
19
85
19
88
19
91
19
94
19
97
20
00
20
03
Fa
ilu
res
Pe
r Y
ea
r
Company A
Company B
4
The solution involves four key questions about the replacement programThe solution involves four key questions about the replacement program
Measuring the right data?
Predicting the right future?
Predicting the right future?
Funded at the right level?
Funded at the right level?
Replacing the right cable?
Replacing the right cable?
• Does the utility know:– What causes failures?– How to avoid them?– How much it costs?
• Does the utility know:– What will happen if programs stay as they are today?– Whether there will be a ‘crashing wave’ or a ‘long swell’?– How the future could be changed?
• Does the utility know:– Which cable segments or half-loops are most cost-effective to address?– Whether and when to inject, outsource, directionally bore, etc.?– How to ensure the field replaces what the model assumed they would?
• Does the utility know:– What level of funding would at least stabilize URD outages?– What funding would be needed to achieve customer satisfaction?– What funding is needed to at least break-even on repair costs?
Answering these four key questions will allow the utility to
optimally manage its URD cable replacement programs
Answering these four key questions will allow the utility to
optimally manage its URD cable replacement programs
5
AgendaAgenda
The typical problem
A comparison of URD programs
The myth of the ‘tsunami’ dispelled
How to address the problem
Observations and key questions
6
Companies’ URD cable installation histories differ by year…Companies’ URD cable installation histories differ by year…
• Most companies started ramping
up in the 1960’s
• Some were responding to local
ordinances requiring URD for
residential developments of any
significant size
– E.g., NY in 1967: URD for
developments with 5 or more
• Housing growth is the key driver
– Often a good correlation between
feet installed and customer growth
– Recession in early 1980’s shows
– In 1990’s some were affected by
local or regional limits to growth
• There can be problems with
getting this data and trusting it
– Plant accounting is a source
– But may not be consistent with
engineering data
URD Cable Installation History Company B
0.0
0.5
1.0
1.5
2.0
2.5
1960
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
Cab
le F
t. I
nst
alle
d P
er Y
ear
(Mil
lio
ns)
URD Cable Installation History Company A
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
Cab
le F
t. I
nst
alle
d P
er Y
ear
(Mil
lio
ns)
7
…and by type, wherein the problem originated…and by type, wherein the problem originated
Unjacketed 3-phase cable with worn concentric neutral around insulated conductors
• Typical progression of URD types:
– HMW unjacketed (1960’s)
– XLPE unjacketed (1970’s or later)
– XLPE jacketed (1970’s or later)
– TRXLPE jacketed (1980’s to now)
– EPR jacketed (1980’s to now)
• Virtually all is concentric neutral
• Original HMW was un-stranded
• Some went from DB to in-conduit
– Especially in rocky soil
– And some had a period of C-in-C
• Typical insulation by voltage:
– 12kV – 170 mil
– 34kV – 240 mil
* Glossary:
HMW – High Molecular Weight Polyethylene
XLPE – Cross-Link Polyethylene
TRXLPE – Tree-retardant XLPE
EPR – Ethylene Propylene Rubber
DB – Direct Buried
C-in-C – Cable in Conduit (cable pre-inserted)
For most companies, the problem is their 34 kV-class or, for their
15kV-class, the 170-mil HMW unjacketed cable installed in the
1960’s and 1970’s, if they have not already replaced it
For most companies, the problem is their 34 kV-class or, for their
15kV-class, the 170-mil HMW unjacketed cable installed in the
1960’s and 1970’s, if they have not already replaced it
8
It is important to quantify failure rates in order to fix the problemIt is important to quantify failure rates in order to fix the problem
Utility
Annual URD
Failures
Primary URD Miles
Failure Rate Per
Mile
Avg. Annual Increase in
Failures
Year Fixed
Problem
Explanation For Situation
A 20 6,000 .003 3 1965 Always used ‘thick’ insulation
B 170 3,000 .057 3 1983 Switched to thicker cable, in conduit
C 400 6,000 .067 40 1981 Switched to jacketed TR cable
Replaced segments on n-th fail
D 300 4,000 .075 11 1987 Switched to jacketed TR cable
Replaced segments on n-th fail
E 1,800 18,600 .100 80 1981 Switched to jacketed, thicker cable
F 500 2,400 .208 0 1981 Switched to jacketed, TR cable
Replaced most old cable
H 5,000 16,000 .313 125 1984 Switched to jacketed, but
did not replace segments much
There was/is a solution: start installing good cable (early), and/or
replace (or inject) the failure-prone cable as you identify it
There was/is a solution: start installing good cable (early), and/or
replace (or inject) the failure-prone cable as you identify it
9
AgendaAgenda
The typical problem
A comparison of URD programs
The myth of the ‘tsunami’ dispelled
How to address the problem
Observations and key questions
10
Exponential density functionMean = 15
0%
1%
2%
3%
4%
5%
6%
7%
0 5 10 15 20 25 30 35 40 45 50
Cycles to failure
Pro
bab
ility
of
failu
re
Many components’ failures fit the exponential process modelWhich means they are ‘memory-less’ and independent of ‘age’ or ‘cycles’Many components’ failures fit the exponential process modelWhich means they are ‘memory-less’ and independent of ‘age’ or ‘cycles’
So, for an exponential
process, preventive
replacement will not work at all, e.g., electronic
chips
For the exponential curve, the slope at the origin
points to the mean
At any point on the curve, the mean ‘cycles to failure’
is the same
11
This is This is typical for typical for
devices like devices like circuit circuit
breakers, breakers, where the where the
‘cycles’ are ‘cycles’ are fault fault
operations, operations, and for and for
URD cable URD cable with ‘cycles’ with ‘cycles’
as yearsas years
The Weibull curve assumes ‘wearout’ caused by cyclesWith a failure rate that increases with ‘age’ or ‘cycles’ The Weibull curve assumes ‘wearout’ caused by cyclesWith a failure rate that increases with ‘age’ or ‘cycles’
Weibull conditional failure ratemean = 15
0%
10%
20%
30%
40%
50%
60%
70%
80%
0 5 10 15 20 25 30 35 40 45 50
Cycles at failure
Fa
ilu
re r
ate
(fa
ilu
res
/ su
rviv
ors
)
shape = 2.5
shape = 1.0
12
Three scenarios paint the picture…Three scenarios paint the picture…
Scenario 1: “The egg thru the snake”
When the Weibull distribution has a shape value of 30 and a scale value of 25 years,
– the assumed rate of cable failures are tightly bunched around the 25-year point, and
– the profile of predicted cable failures follows the distribution of installations,
– with the peak failures shifted about 25 years in the future (the dispersion adds about three years: the 1973 peak in installations corresponds to a 2001 peak in failures)
Shape = 30Scale = 25 yrs
Predicted Cable Failures -Illustrative-
0
20
40
60
80
100
120
19
75
19
79
19
83
19
87
19
91
19
95
19
99
20
03
20
07
20
11
20
15
20
19
20
23
20
27
20
31
Fa
ilure
s (0
00
s)
Assumed Rate of Cable Failures-Illustrative-
0%
10%
20%
30%
40%
50%
1 5 9
13
17
21
25
29
33 37
41
45
49
Years Since Installation
Fai
lure
s
Cable Sections Left By Year Installed-Illustrative-
0
5001,000
1,5002,000
2,5003,0003,500
4,000
4,5005,000
19
50
19
54
19
58
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
20
02
Sec
tio
ns
13
…Wider failure distribution smoothes the installation profile… …Wider failure distribution smoothes the installation profile…
Scenario 2: “Smoothing the profile”
When the Weibull shape value is reduced to 10,– the assumed rate of failures are more
dispersed around the 25-year point and– the profile of the predicted failures is a
smoothed version of the distribution of installations,
– with its peak failures shifted about 34 years into the future (from 1973 to 2007)
Predicted Cable Failures-Illustrative-
0
10
20
30
40
50
60
19
75
19
79
19
83
19
87
19
91
19
95
19
99
20
03
20
07
20
11
20
15
20
19
20
23
20
27
20
31
Fa
ilure
s (0
00
s)
Cable Sections Left By Year Installed-Illustrative-
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
19
50
19
54
19
58
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
20
02
Sec
tio
nsShape = 10
Scale = 25 yrs
Assumed Rate of Cable Failures-Illustrative-
0%
2%
4%
6%
8%
10%
12%
14%
16%
1 5 9
13
17
21
25
29
33 37
41
45
49
Years Since Installation
Fai
lure
s
14
…With a wide distribution erasing the installation profile…With a wide distribution erasing the installation profile
Cable Sections Left By Year Installed-Illustrative-
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
19
50
19
54
19
58
19
62
19
66
19
70
19
74
19
78
19
82
19
86
19
90
19
94
19
98
20
02
Sec
tio
ns
Assumed Rate of Cable Failures-Illustrative-
0%1%1%2%2%3%3%4%4%5%
1 5 9
13
17
21
25
29
33 37
41
45
49
Years Since Installation
Fai
lure
s
Shape = 2.5Scale = 25 yrs
Peak in 2072 >>
Predicted Cable Failures-Illustrative-
0
2
4
6
8
10
12
14
19
75
19
79
19
83
19
87
19
91
19
95
19
99
20
03
20
07
20
11
20
15
20
19
20
23
20
27
20
31
Fa
ilure
s (0
00
s)
Scenario 3: “The egg disappears”
When the Weibull shape value is reduced to 2.5, – the assumed rate of failures are widely
dispersed around the 25-year point and– the profile of predicted failures is virtually a
straight line after the first 25 years, – with its peak failures shifted almost 100 years
into the future
15
The third scenario is born out by existing evidenceThe third scenario is born out by existing evidence
• For companies that have done little URD cable replacement, the trend is much like what is pictured in the third scenario:
– Failures increasing at a steady annual rate of about 5 percent, which, with compounding, means a doubling in about 14 years
• With an active replacement program, no such increase will occur, but:
– The replacement itself might need to grow at about 5 percent per year to keep up, until the failure-prone cable is substantially replaced
• The replacement program’s impact can be increased or diminished by how the cable to be replaced is selected:
– It needs to be, as much as possible, worst first
The myth that cable installed in the 1960-70’s had a ‘thirty-year’ life and
so will come ‘crashing down on us’ in the next ten years is just not true.
There is no ‘crashing wave’, only a ‘long swell’ until the worst is replaced
The myth that cable installed in the 1960-70’s had a ‘thirty-year’ life and
so will come ‘crashing down on us’ in the next ten years is just not true.
There is no ‘crashing wave’, only a ‘long swell’ until the worst is replaced
URD Cable Failures
0
1,000
2,000
3,000
4,000
5,000
6,000
19
82
19
85
19
88
19
91
19
94
19
97
20
00
20
03
Fa
ilu
res
Pe
r Y
ea
r
Company A
Company B
16
HL&P addressed its URD problem effectivelyHL&P addressed its URD problem effectively
HL&P 15kV Failures
HL&P 35kV FailuresHL&P (CenterPoint Energy) had the same type of problem as many and addressed it with a combination of Lightning Arrestor (L/A) upgrade and a cable replacement program:
• Key program items– L/A change out program to limit “let-through current”
starting in the mid 1980’s– From 1981 to 2001 all replacement cable was jacketed
cable in conduit, but due to cost and no observed increase in reliability, the installation practice was shifted back to direct buried after 2001
• 35kV Cable– Adopted an aggressive 5 year replacement program (1987
to 1992), funded at $10 million per year, that replaced 95% of the original installed cable
• 15kV Cable– Active policy for past the 10 years of replacing half loops
with 2 or greater failures in a rolling 12 month period. 2005 funding level was $2.4million or a 19 mile replacement program (at $24/ft)
HL&P has leveled the exponential growth of failures, and stabilized failures at an acceptable level, where it has stayed for over a decade. Others can do the same.
HL&P has leveled the exponential growth of failures, and stabilized failures at an acceptable level, where it has stayed for over a decade. Others can do the same.
17
AgendaAgenda
The typical problem
A comparison of URD programs
The myth of the ‘tsunami’ dispelled
How to address the problem
Observations and key questions
18
Cable Failure
Improper Installation
Mechanical Damage
Mark-outs
Enforce Penalties
Temperature of cable
Manufacturer Defect
Insulation Breakdown
Dig In
Treeing
Rocky Soil
Enforce Trench
StandardsImproper Training
Insulation Thickness/Type
Cable Injection
Cable Replacement
Lightning
One Call
Upgrade to MOV
Arresters around open
New Construction
Jacket/casing missing/broke
Moisture in cable/joint
Rock Bruising
Corrosion Strength
Steam (Ducted)
LoadingVentilation(Ducted)
AddCapacity
Maintain Manholes
Capacity Planning
Thumping
Cathodic Protection
Maintain Cath Prot
Street Crossing
It is important to investigate causes of failure…It is important to investigate causes of failure…
Thermal Instability
Wet Manhole
…in order to know how to fix the problem…in order to know how to fix the problem
19
URD Failure Rates Per Mile -Illustrative-
-
0.5
1.0
1.5
2.0
2.5
2 7 12 17 22 27 32
Years Since Installation
Fa
ilure
s P
er
Mile Actual Data
Weibull Fit
Experience with many utilities’ URD data confirms certain patternsExperience with many utilities’ URD data confirms certain patterns
2004 - Primary and Secondary Cable Failures excluding exit cables
2/6/04 had 2” rain
0
10
20
30
40
50
60
0 10 20 30 40 50 60 70 80
Monthly Average Temperature (Islip)
UG
Out
ages
2002-2004 Primary Cable Failures excluding exit cables
0
50
100
150
200
Jan Feb MarchApril May June July Aug Sept Oct Nov Dec
UG
out
ages
Most utilities URD data show that URD failures are correlated with age,
temperature, moisture, and, of course, cable type
Most utilities URD data show that URD failures are correlated with age,
temperature, moisture, and, of course, cable type
20
0.000
0.100
0.200
0.300
0.400
0.500
0.600
0 5 10 15 20 25 30 35 40 45 50
Age of cable (years)
Ca
ble
fa
ilu
res
pe
r m
ile
Typically, age-based failure rates are still low (~ 50% above average)So age-based programs must replace a lot of good cable to get the badTypically, age-based failure rates are still low (~ 50% above average)So age-based programs must replace a lot of good cable to get the bad
Although a Although a failure rate failure rate difference difference
of 3x is of 3x is significant,ssignificant,s
ome ome conditions conditions
provide 10x, provide 10x, e.g., 350’ e.g., 350’
cable cable sections sections that have that have failed 3 failed 3
times in the times in the last 5 years last 5 years
fail at 9.0 fail at 9.0 per mile per mile
Note: System average failure rates noted earlier are affected by average age of the system
21
Minutes per outage
Growth rate of outages
Past outages per mile
Cost permile
The key to optimal replacement is high failure rate…The key to optimal replacement is high failure rate…
Future minutes per year avoided
Dollars spent
Future minutesavoided per yearFuture outages
avoided per year
Future outagesavoided per year
Past outagesper year
Bang per buck
Past outages per year Miles of line
to be replaced
1 $90,000
1.25 45001 min.$2.00
Miles of line to be replaced
Dollars spent
8x x x =
Where:
• $90,000 per mile = 5280 feet/mile x $17 per foot to replace• 8 outages/mile/year = 13 spans/mile x (3 outages per 400ft span in past 5 years)• 25% growth rate = 3 outages in past 5 years becomes 3 outages in next 4 years• 4500 minutes per outage = 50 customers per outage x 90 minutes per outage
The higher the failure rate... …the higher the bang per buck
22
…As well as ways to reduce the unit cost…As well as ways to reduce the unit cost
• Injection is sometimes a cost-effective option
– Guaranteed by some vendors for many years
– Typically half the cost per mile when used on the right
cable
– Not effective with blocking splices
– Does not solve problem of corroded neutral
– Not really an option for replacing individual segments, but
good for half-loops
• Volume can reduce unit cost
– Half-loops get better cost than individual segments
– But not worth it if failure rate of replaced cable drops faster
than unit cost when volume increases
• Use trenchless technology where possible
– Tunneling under driveways, through tree root systems,
etc.
• Take credit for saving O&M, if appropriate
– Repairing future failures can be made easier, e.g., conduit
23
The distribution of failures by half-loop is a key part of the programThe distribution of failures by half-loop is a key part of the program
• Most of utilities’ URD cable sections and half-loops has not failed in the last five years. Replacement of that cable would be unnecessary at this time
• A customer satisfaction-driven program would target those half-loops that experienced a high rate of failure, because every segment that fails in the half-loop causes outages to all customers behind that device (the fuse on the riser)
• The replacement program should then be further refined by replacing only those half-loops or sections in the half-loop that have not already been replaced, or that fit certain criteria, e.g., corroded neutral, voltage, etc. The goal is not to replace all the cable, but to
replace enough of the right cable at the right time to affect the trend of failures
The goal is not to replace all the cable, but to replace enough of the right cable at the right
time to affect the trend of failures
49%
37%
7%3% 3%*
0%
10%
20%
30%
40%
50%
60%
De
vice
s0 1 2 3 4 or
greater
Failures
URD Failures
Distribution of Failure by Half-Loop
(2001-2005)
Possible target of replacement
program
24
With the right approach, an optimal program can solve the problemWith the right approach, an optimal program can solve the problem
Based on the number of miles of cable that fit the criteria of the half-loop program:
• A program of 2x miles of URD cable replacement, beginning in 2007 and rising by y% per year, would stabilize failures at a normal 2007 level
• 2005 was a hot year, like 1999, so a normal 2006 would be less
• An x-mile program would leave failures rising, although at half the rate
Without a replacement (or injection) program, or with
a minimal program, Company A’s URD failures would
continue to double every 12-25 years
Without a replacement (or injection) program, or with
a minimal program, Company A’s URD failures would
continue to double every 12-25 years
Company A URD Cable Failures
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
20
02
20
04
20
06
20
08
20
10
20
12
20
14
20
16
Fai
lure
s P
er Y
ear
Actual '82-'05
Projected '82-'16
Repl x mi +y% per yr
Repl 2x mi +y% per yr
25
Elsewhere*, we have shown how to choose the right level of replacementElsewhere*, we have shown how to choose the right level of replacement
From the viewpoint of a prudent company (and its regulator), there are three tests of a replacement program of this type:
1) Trending – What are the trends in spending and outages of this type?• If spending is down while outages are up, more spending is needed
• If spending is level and outages are level, spending may be adequate (but see below)
• Obviously, there are issues in adjustment for weather, costs, productivity, etc.
2) Benchmarking – What are other companies doing in spending and performance?• If other comparable companies are spending more or getting better results, shouldn’t you?
• Obviously, there are reasons why some companies may differ for good reason
3) Modeling – When the process is modeled, what does it indicate the required level of spending should be to maintain performance or improve it to what customers expect?• This is the kind of modeling we have demonstrated above
• There is customer satisfaction data to suggest that the threshold may be around three outages per year – including outages caused by devices upstream of the URD half-loop
• Compared to the other two tests, this one is the most useful if the modeling is done right
A ‘prudent’ replacement program should be designed
with these tests in mind, especially the third
A ‘prudent’ replacement program should be designed
with these tests in mind, especially the third
* See “The Reliability Conundrum – What Is the Right and Prudent Level of Spending on Service?”, Public Utilities Fortnightly, March 2004, by Daniel E. O’Neill
26
AgendaAgenda
The typical problem
A comparison of URD programs
The myth of the ‘tsunami’ dispelled
How to address the problem
Observations and key questions
27
What we have learnedWhat we have learned
Measuring the right data
Predicting the right future
Predicting the right future
Funded at the right level
Funded at the right level
Replacing the right cable
Replacing the right cable
• Capturing the right information during installation and failures, e.g., date installed, insulation type, location and protection device operated, etc. that will enhance the data mining and prioritization process going forward
• Modeled correctly in terms of installation history, failure rate, and replacement/retirement
• No ‘tsunami’, just growth at a compounded rate
• Replacing based upon centralized selection criteria that include failure history, voltage, corroded neutral, restoration time, number of customers, etc.
• Based on trending, benchmarking, and modeling • As a ‘stake in the ground’, at least determine the amount of funding
needed to stabilize failures at current levels, then determine what it would take to achieve customer satisfaction
Answering these four key questions will allow the utility to
optimally manage its URD cable replacement programs
Answering these four key questions will allow the utility to
optimally manage its URD cable replacement programs
28
Observations and Key QuestionsObservations and Key Questions
Observations
• Replacement program – Utilities need to implement an enhanced replacement program for URD cable. Because the new types of cable have much lower failure rates, replacement of older vintages will result in lower failures
• Cable selection – Utilities need to select cable that is failing at a rate of 2 or more per half-loop per year supplemented with a section replacement program
• Without such a replacement program, URD failures will continue to double in 10-15 years, with consequences for repair cost, increased multiple interruptions to the same customers, possible lengthy outages during heat waves, and accumulation of the inevitable replacement cost
• There is no ‘tsunami’, only a long, high swell, in the sense that failures and costs will continue to rise at about 5% per year, doubling every 12-15 years. But a replacement program can stabilize the failures, or even reduce them
Key Questions
• Is the current level of failures acceptable, or should the utility aspire to reduce them further?
• Can the utility execute an enhanced replacement program effectively with its current processes?
• Are their other opportunities to reduce and refine the selection process e.g.: cable injection, lightning arrestor upgrades, partial discharge condition testing, etc.?
29
Questions?Questions?