Maintenance and Reliability Theory

John E. Skog P.E. WGA3-06 Tutorial June 2006 Rio de Janeiro

Todays Agenda
Evolution of Maintenance and Driving Theory Traditional Bimodal Maintenance Reliability Centered Maintenance Condition Based Maintenance Performance Focused Three Case Studies Cables SF6 Breakers Transformer On-line Monitoring
2

Maintenance Evolution Traditional

Overhaul
Tra diti on al
PM

Time

CM

Asset

Characteristics of Traditional Maintenance

Conservative/ Excessive Belief More Maint. Increased Reliability

Intrusive

Generalized

Traditional Maintenance

Benefits
Periodic inspection servicing is necessary Acknowledgement that full equipment operating life is only possible if worn parts are replaced.

Drawbacks
Time is poor predictor of wear Overhauls create more problems than they solve

High cost Manufacturers did not understand the operating environment Reliability and availability were not being met
5

Maintenance Evolution RCM

Criticality Safety Overhaul
Tra diti on al
PM

Function
RC M
Event Failure Finding

Condition

Failure Cause Effects

Time

CM

Asset

Run to Failure

RCM
A structured process that identifies the effects of
failures and defines the appropriate maintenance path for managing their impacts. RCM identifies both the most technically and economic effective approach to maintenance .

RCM History
Airlines

1965 MSG-1 (Maint. Steering Group) 1970 MSG-2

Experience
DC 8 339 Scheduled Removal Tasks 7 Scheduled Removal Tasks 747 8 Scheduled Removal Tasks

RCM History (cont.)

Airline Observations: Maintenance needs to focus on system that have significant impact on safety or economics. Hard time overhaul policies were ineffective. Management of maintenance was crucial.

RCM History (cont.)

US Navy 1978 Contracted United Airlines

US Electric Utilities - 80s EPRI Sponsored Nuclear 1985-1987 Fossil Fuel Plants US Electric Utilities - 90s Substations 1990 T&D EDF and Others Nuclear Plants Transmission Substations

10

RCM Task Selection

The RCM task selection approach used to ensure that only

applicable and cost effective tasks are selected to address the causes of critical equipment failure modes

RCM Task Categories Inspection-Condition Monitoring-Predictive Maintenance Periodic

Rework-Time Directed Discard-Time Directed

Failure Finding Run to Failure

11

Simplified Task Selection Logic

Mode & Cause Selected Review Effects Safety Consequences Task Must Reduce Risk of Failure Operational Non-Operational Is Condition Monitoring Effective?

No-Periodic Yes PreservedFailure PDM PM NoIs Condition Monitoring Effective? Can the failure be tolerated? No Yes Finding Task PM
No Is the Failure Evident? No Can the failure be tolerated? Is a time directed task effective?

Is a periodic task effective?

Consequences Task must cost Less than the consequences

Consequences Task must cost less than repair

Failed

Is the Failure Evident? Corrective Maintenance Yes Run to failure

12

No-Design Yes Change

Design Change

No

Hidden Failure Finding

Characteristics of Reliability Centered Maintenance

Focused on Dominant Causes

Preservation of Function

Specialized Tasks

13

Reliability Centered Maintenance

Benefits
Critical Functions Equipment and application specific Greater insight into failure process Eliminated ineffective tasks

Drawbacks
Viewed as difficult and not applicable to power industry 99.999% (1 hour of outage per year) reliability is difficult to understand Living program forgotten Did not set maintenance intervals

14

Maintenance Evolution CBM

Criticality Safety Overhaul
Tra diti on al
PM

Function
RC M
Event Failure Finding

Condition

Failure Cause Effects

Time

CM Predictive

Asset

Run to Failure Diagnostics

Proactive

IM

Sensors

Predictions
CBM

Aging Models
15

Real Time Data

Condition Based Maintenance

Condition Based Maintenance accentuates the value of

RCM task selection logic and emphasizes that more intrusive replacement and overhaul tasks only need to take place when measurable wear or aging occurs.

Condition Directed Tasks are initiated when deterioration

has gone beyond a prescribed limit

16

Characteristics of CBM
Greater Reliance on Measurable Conditions

Unobtrusive

Aging Mechanism Understood

Data Driven

17

Condition Based Maintenance

Benefits
Increased availability Reduced costs More frequent analysis of asset condition

Drawbacks
Data systems were may not be adequate. Process management overlooked. No methodology for justifying increased monitoring Increased back-office analysis

18

Maintenance Evolution PFM

Criticality Safety Overhaul
RC M

Function

e oc Pr

Tra iza diti tio ona n l Tra diti ona l

Or ga

Time

Performance Focused Maintenance

Sensors Predictions
CBM

Condition R CM Condition Event PM Event PM Failure Failure Finding Finding CM Asset Run to Asset CM Run Failure to Predictive Failure Diagnostics Predictive Proactive IM Diagnostics Proactive CBM IM

Failure Cause Effects

ss

Data
Real Time Data
19

Aging Models

What is PFM?

PFM is a comprehensive maintenance strategy

emphasizing the understanding of Failure Mechanisms, Measurement, Interval Optimization, Task Prioritization, Feedback and the use of Data. PFM recognizes the need for process control

20

What Maintenance is Included in PFM?

Maintenance includes all activities associated with

preserving or restoring critical functions. Typical maintenance activities include:
Preventive Maintenance Condition Monitoring/Inspections Diagnostic Testing Integrated Monitoring Predictive Activities Hidden Failure Finding Tasks Condition Directed Corrective and Renewal Tasks Corrective Maintenance Pre-Emptive Replacement

21

PFM 12 Step Methodology

Failure Rate

Time

Identify System Boundaries and Critical Functions Steps 1-2

Perform Failure Mode and Effect Analysis (FMEA) Steps 3-5

Aging Mechanisms Step 6 Task Selection and Interval Optimization Step 7

Measures, Metrics &KPIs Reconciliation and Program Development Step 8-9

22

Step 10

Implementation Documentation Step 11- 12

Bridging Business Issues and Technical Requirements

Understanding: The Aging Process Failure Initiation Mechanisms

23

General Age Reliability Patterns Renewal

Failure Rate Bathtub
Failure Rate

Strategies are ineffective

Pronounced Wear out

Time Failure Rate

Failure Rate

Time Decreasing Failure Rate

Gradual Wear out

Time Failure Rate Failure Rate

Time Low Initial failure rate with quick increase

Constant Failure Rate

Time
24

Time

Task Interval Optimization-Weibull Age Modeling

100%

F(t) =

-[(t-t0)/] 1-e

80% 60% 40% 20% 0%

With Some Good Probability Cum ulative Failure Data, Failures Can be Predicted Mathematically

Infant M ortality Random W ear Pronounced W ear

t0 = Guarantee Period = Characteristic Life .. MTBF = Shape factor

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49

Failure Probability Distribution

18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49

Infant M ortality Random W ear Pronounced W ear

25

Failure Initiation Patterns

2

Binary Pattern

Some Failures 2 Provide no Warning, Some Give Adequate Warning Failure Pattern Intermittant
OK

OK

Failed
0 0 5 Increasing A 10 ge 15 20 25

Failed

0 0 5 Increasing A 10 ge 15 20 25

Fast W Pattern ear

OK

Slow W Pattern ear

OK

F ailed

0 0
Age 5 Increasing 10

15

20

25

Failed

0 0 5 Increasing A 10 ge 15 20 25

26

Characteristics of PFM
Life Cycle Approach to Data Management Process Measurement And Feedback Integrates Best Business and Technology Practices

Optimization

Focused Maintenance

27

Performance Focused Maintenance

Benefits
Increased availability and reliability Reduced life-cycle costs Data collection fundamental part of the process Integrated business and technology approach to maintenance

Drawbacks
Requires quality data collection and storage processes Must be understood and supported by the highest management levels

28

Performance Focused Maintenance Case Studies (3)

29

Case I-15kV Distribution Cables

Issues: Aging population-(four insulation systems) Inspection program that did not affect failure rates Complicated and time consuming replacement ranking system Ineffective condition assessment tasks Poor asset data 0.4% replacement rate Drivers Performance Based Rates High replacement costs Key Considerations Design Improvements Mostly in conduit
30

Equipment Group: Population by Age and Insulation Type

U G P rim a ry C a b le s
2 ,0 0 0 1 ,8 0 0

1 ,6 0 0

1 ,4 0 0

Conductor Miles

1 ,2 0 0

HM W
1 ,0 0 0

800

600

400

XLPE P IL C T R -X L P E

200

0
96 98 90 86 74 76 72 68 62 58 60 64 66 70 78 80 82 84 88 92 94 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 00

Ye ar

Identify System Boundaries and Critical Functions Steps 1-2

Type of Insulation PILC HMW-PE XLPE TR-XLPE

Expected Service Life 45 - 50 years 15-20 years 25-30 years 45 years +

31

Circuit Miles 4,900 1,500 30,200 2,400

FMEA: Cables

Perform Failure Mode and Affect Analysis (FMEA) Steps 3-5

Dominant Failure modes are Failure to Insulate and Failure to provide a ground plane

32

Tree Examples - Failure to Insulate

Cable Insulation Water Trees

Aging Mechanism Step 6

:Cable Insulation gy nt a te e tr em tS lac an ep u lt es eR R tiv mp -e re P

#2 AWG XLPE cable from Fullerton. Water treeing more than 40% through the insulation.

#2 AWG HMW vintage cable which failed in Ridgecrest. Water treeing more than 50% through the insulation

Source: DAE, Improving the performance of underground cable. Sept 14, 2001

33

Cable Insulation Failure Model

Cable Unreliability 1 0.8
Unreliabilit

0.6 0.4 0.2 0 yr(s) 10 yr(s) 20 yr(s) 30 yr(s) 40 yr(s) 50 yr(s) 60 yr(s)
Years after installation

PILC HM W XLPE

10% Limit

TR-XLPE

34

Pre-emptive Replacement Strategy Age Limit (10% failure)

SAIDI + SAIFI IMPACTS
Millions $70 $60 $50 $40 $30 $20 $10 $0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Year No Age Limit Age Limit

35

Case II SF6 Breaker Maintenance

Issues: Extension of Oil Breaker Maintenance Philosophy Declining Reliability Increasing Maintenance Costs Increased Availability Required No CMMS Maintenance Behind Schedule

36

10 Year Results
Effective Knowledge Transfer Improve Data Collection Extended Maintenance Intervals (double) with Defendable Basis Additional PM Triggers-Age Exploration Reduction in Rework ActivitiesImproved PM Effectiveness Increased Availability Improved Reliability Elimination of PM Backlog

37

SF6 PFM Implementation Results

SF6 Breakers

(69% of the total population)

10 Year Analysis Period

More than 10,000 Inspections and Maintenance Tasks

38

Improved Maintenance Plan Resulted in a Decreased Failure Rate

RCM Implementation
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1991 1992 1993 1994 1995 1996 1997 1998 1999

Stability
200 180

SF6 CIRCUIT BREAKERS TYPE OF PROBLEMS DETECTED DURING MAINTENANCE ACTIVITIES

Decreasing Number of Inspections

160 140

Catch-up Period
Decreasing Failure Rate

120 100 80 60 40 20 0

2000

2001

2002

YEAR

No problems found Major Problems Detected Major Failures (% / Installed)

39

Minor Problems Detected SF6 Circuit Breakers Measured Minor Failures (% / Installed)

Improved Effectiveness

S F 6 C IR C U IT B R E A K E R S F IN D IN G S D U R IN G R E V IS IO N S - R E S O L U T IO N O F M IN O R F A IL U R E S

100%

Improved Maint.
80%

Techniques

60%

40%

20%

Improvement in Maintenance Effectiveness

0% 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

R e s o lv e d d u r in g w o r k s

R e s o lv e d in s h o r t te r m

S c h e d u le fo r n e x t in s p e c tio n

40

Case III-Application of On-Line Monitors

Issues: Aging Asset Population Recent Cascading Failure Push to Install New Monitoring Technology Poor Experience with Hydrogen Monitors Increased Insurance Rates

41

Fleet Characteristics
Large Power Transformers 220 KV to 115 or 66KV 120 to 280 MVA Single and Three Phase Average Age = 39 Years Max Age = 76 Years Replacement Costs $3M to $4M (on the pad) Population = 188

42

Failure History (population = 188)

Failure E vents

7 6 Failures .s 5 4 3 2 1 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Year

43

Age Distribution

A-Bank Age Distribution

50 40 30 20 10 0
-8 0 -7 0 -6 0 -4 0 -3 0 -2 0 -5 0 76 66 56 46 36 26 16 610

100% 80% 60% 40% 20% 0%

Age Distribution Accumulated Percentage

Age

Median Age 44 years

44

Failures as a Function of Age

A Bank Failures
6 5 Failures 4 3 2 1 0 0-5 6-10 11-15 16-20 21-25 26-30 31-35 36-40 41-45 46-50 Age at Time of F ailure
45

1.20% 1.00% 0.80% 0.60% 0.40% 0.20% 0.00% Rate

Industry Reported Failure Distribution

Failure Distribution by Im pacted System

Unknown (45%)

Dielectric (38% to 45%)

M echanical (0%4%) Containment (3%6%) Current Carying (7%)

46

Utility Reported Failure and Trouble Distribution

Failures
E xternal Interface 7%

On-line Monitor Target

Voltage R egulating 40%
None 9%

Problem Occurrence Distribution

Can't do Maintenance 2% NLTC 7% High PF 2%

Problems

Leaks 19%

D ielectric 46% E lectrom agnetic 7%

Accessories 12%

Inert 18%

Cooling 13%

Bushings 18%

47

Age Reliability Patterns-Failure Probability

Failure Rate
Failure Rate

Bathtub

Pronounced Wear out

Time Failure Rate

Failure Rate

Time Decreasing Failure Rate

Gradual Wear out

Time Failure Rate

Monitoring can be very effective if..

Time Failure Rate Low Initial failure rate with quick increase

Constant Failure Rate

Time
48

Time

Typical Reliability Predictive Models

Annual Failure Rate
1.5%

1.0% Linear Failure Rate Actual 0.5%

0.0% 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 Year

Annual Failure Rate

60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 H Failure M SB odel Actual

Annual Failure Rate-Weibull

1.50% 1.00% 0.50% 0.00%

Failure Rate-Weibull Actual

52

40

44

24

12

16

49

20

28

Age

32

36

48

56

Model Comparison Applied to Existing Fleet

10.00% Annual Failure Ra 8.00% 6.00% 4.00% 2.00% 0.00% 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Weibull HSB Linear

Current Failure Rate

50

Failure Mechanisms
2 2

Binary Pattern Intermittant Failure Pattern

OK OK

Failed
0 0 5 Increasing A 10 ge 15 20 25

Failed

0 0 5 Increasing A 10 ge 15 20 25

Fast W Pattern ear

OK

Slow W Pattern ear

OK

F ailed

0 0
Age 5 Increasing 10

15

20

25

Failed

0 0 5 Increasing A 10 ge 15 20 25

51

Incipient Failure Model

Probability of Failure

100.0%

63% P robability of D etection

Incipient Failure Pre-cursor Model

8.0% 7.0% 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0%

80.0% 60.0% 40.0% 20.0%

0.0%

M ean Tim for an e incipient fault = 14 m os.

Incremental Probaility of Failure Incremental

Probaility of Failure Cumulativ e Probability of Failure Cumulativ e Probability of Probability Successful of Failure Detection

12 M onth D A G Sam pling C ycle

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 2628 Months

52

On-line Monitoring Decision Model

Failure Model Direct Costs Transformer Collateral Damage Fines Indirect Costs Commissions and Ratepayers Insurance Stress on other units Supply impacts True Risk Reduction Fleet Replacement Impacts
53

Cumulative Cash Flow for Multi-Gas Monitors

$150,000 $100,000 $50,000 $0 -$50,000 -$50,000 -$100,000 -$100,000 -$150,000 -$150,000 -$200,000 -$200,000 -$250,000 -$250,000 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years Years 1. On-line Monitoring 2. No Maintenance On-line Monitoring 2. No Maintenance 2. 5. 1. On-line Monitoring 3. Periodic DG A 1. No Maintenance-On-line Monitoring2. No Maintenance No Maintenance Periodic DG 3. A 4. No Maintenance-Periodic DG A

54

Transformer Fleet Risk Exposure Profiles

Total Annual "A Bank" Failure Risk

Millions $12 $10 $8 $6 $4 $2 $No Current PM Current PM + Maintenance Program On-Line DGA Bushing Transformer Main Insulation System LTC Electro-Magnetic

55

Extended Useful Life

Deferred Transformer Replacement

$900 $800 $700 $600 $500 $400 $300 $200 $100 $0 ($100) Year 0 Year 1 Year 2 Year 3 Interest Deferral on Capital Expenditure Cost of On-line Monitor and annual O&M Cumulativ Cash Flow e

Thousand

Extended Life

56

Conclusions from PFM Approach:

Substantial benefit can be obtained from installation of multi-gas monitors across a large fleet of power transformers Improved transformer reliability Reduced failure impacts Realization of full transformer useful life Identification of units in urgent need of repair/replacement. Substantial reduction in overall transformer operating risks

57

Future Trends in Maintenance

Sharing of Failure and Trouble data Mode and Cause Level Demographics
Application Type

Population
Age models vs. Failure Rate

Full Asset Utilization Risk Reduction Key Performance Indicators Asset Family Maintenance Process Life Cycle Costs
58

Open Discussion and Questions:

Need More Information?

[email protected]
360.352.9977 2037 Berry St. NE Olympia WA, 98506 USA

59