Maintenance and Reliability – Theory

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

1

Today’s 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

3

Characteristics of Traditional Maintenance
Conservative/ Excessive Belief More Maint. Increased Reliability

Intrusive

Generalized

4

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 6 .

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

Steering Group) z 1970 MSG-2 z Experience – DC 8 • 339 Scheduled Removal Tasks • 7 Scheduled Removal Tasks – 747 • 8 Scheduled Removal Tasks 8 .RCM History • Airlines z 1965 MSG-1 (Maint.

– Management of maintenance was crucial.RCM History (cont. – Hard time overhaul policies were ineffective.) • Airline Observations: – Maintenance needs to focus on system that have significant impact on safety or economics. 9 .

80’s – EPRI Sponsored Nuclear 1985-1987 – Fossil Fuel Plants • US Electric Utilities .90’s – Substations 1990 – T&D • EDF and Others – Nuclear Plants – Transmission Substations 10 .) • US Navy – 1978 Contracted United Airlines • US Electric Utilities .RCM History (cont.

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 Preserved PDM PM No-Failure Is Condition Monitoring Effective? Can the failure be tolerated? No Yes Finding Task PM No Evident? Is the Failure 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 Directed Tasks are initiated when deterioration has gone beyond a prescribed limit” 16 .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.

Characteristics of CBM Greater Reliance on Measurable Conditions Unobtrusive Aging Mechanism Understood Data Driven 17 .

Process management overlooked.Condition Based Maintenance Benefits Increased availability Reduced costs More frequent analysis of asset condition Drawbacks Data systems were may not be adequate. 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 Asset Run to CM Run to Failure Predictive Failure Diagnostics Predictive Proactive IM Diagnostics Proactive CBM IM n Failure Cause Effects ss Data Real Time Data 19 Aging Models .

Interval Optimization. Task Prioritization. PFM recognizes the need for process control 20 . Feedback and the use of Data.What is PFM? PFM is a comprehensive maintenance strategy emphasizing the understanding of Failure Mechanisms. Measurement.

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 .What Maintenance is Included in PFM? Maintenance includes all activities associated with preserving or restoring critical functions.

Metrics &KPIs Reconciliation and Program Development Step 8-9 22 Step 10 Implementation Documentation Step 11.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.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) = β -[(t-t )/ η ] 1-e 0 80% 60% 40% 20% 0% With Some Good Cum ulative Failure Probability Data. MTBF – β = 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 . Failures Can be Predicted Mathematically Infant M ortality Random W ear Pronounced W ear – t0 = Guarantee Period –η= Characteristic Life ..

Failure Initiation Patterns 2 Binary Pattern Some Failures 2 Provide no Warning. Some Give Adequate Warning Intermittant Failure Pattern OK OK Failed 0 0 5 Increasing 10 A ge 15 20 25 Failed 0 0 5 Increasing 10 A ge 15 20 25 2 2 Fast W ear Pattern OK Slow W ear Pattern OK F ailed 0 0 Age 5 Increasing 10 15 20 25 Failed 0 0 5 Increasing 10 A 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 .

900 1.4 0 0 Conductor Miles 1 .50 years 15-20 years 25-30 years 45 years + 31 Circuit Miles 4.0 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 .0 0 0 1 .8 0 0 1 .200 2.2 0 0 HM W 1 .6 0 0 1 .Equipment Group: Population by Age and Insulation Type U G P rim a ry C a b le s 2 .500 30.400 .

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 .

Water treeing more than 50% through the insulation AWG XLPE cable from Fullerton. Source: DAE.Tree Examples .Failure to Insulate Cable Insulation Water Trees Aging Mechanism Step 6 Cable Insulation : y g e t t n a e r t m S e t c n a l a p lt e u s R e e R v i t p m e e r P „#2 #2 AWG HMW vintage cable which failed in Ridgecrest. Water treeing more than 40% through the insulation. Improving the performance of underground cable. Sept 14. 2001 33 .

4 0.8 Unreliabilit 0.6 0.Cable Insulation Failure Model Cable Unreliability 1 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) .

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 .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 .

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 .

80% 0.00% Rate .20% 1.20% 0.40% 0.60% 0.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 1.00% 0.

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 .

Time Failure Rate Low Initial failure rate with quick increase Constant Failure Rate Time 48 Time ..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.

0% 50.5% 1.0% 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 Year Annual Failure Rate 60.00% 0.0% 20.0% 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 H SB Failure M odel Actual Annual Failure Rate-Weibull 1.5% 0.00% Failure Rate-Weibull Actual 52 40 44 24 12 16 49 20 28 Age 32 36 48 56 0 4 8 .50% 1.0% 10.Typical Reliability Predictive Models Annual Failure Rate 1.50% 0.0% Linear Failure Rate Actual 0.0% 40.0% 0.0% 30.

Model Comparison Applied to Existing Fleet 10.00% Annual Failure Ra 8.00% 2.00% 0.00% 4.00% 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Weibull HSB Linear Current Failure Rate 50 .00% 6.

Failure Mechanisms 2 2 Binary Pattern Intermittant Failure Pattern OK OK Failed 0 0 5 Increasing 10 A ge 15 20 25 Failed 0 0 5 Increasing 10 A ge 15 20 25 2 2 Fast W ear Pattern OK Slow W ear Pattern OK F ailed 0 0 Age 5 Increasing 10 15 20 25 Failed 0 0 5 Increasing 10 A ge 15 20 25 51 .

0% 80.0% 0 2 4 6 8 10 12 14 16 18 20 22 24 26 2628 28 30 32 34 36 38 40 Months 52 . Incremental of Failure Probaility Incremental Probaility of Failure Cumulativ e Probability of Failure Cumulativ eof Probability Probability of Failure Successful Detection 12 M onth D G A Sam pling C ycle 0.0% 4.0% 40.0% 3.0% M ean Tim e for an incipient fault = 14 m os.Incipient Failure Model Probability of Failure 100.0% 7.0% 63% P robability of D etection Incipient Failure Pre-cursor Model 8.0% 60.0% 6.0% 5.0% 20.0% 1.0% 2.0% 0.

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 .

000 -$150.Cumulative Cash Flow for Multi-Gas Monitors $150.000 $50.000 -$150.000 -$250.000 -$50. On-line Monitoring 2.000 $100.000 $0 -$50. No Maintenance 2.000 -$200. No Maintenance 5. On-line Monitoring Monitoring 2. Periodic DG A 4. No Maintenance-Periodic DG A 54 . Periodic DG A 1. No Maintenance 3. On-line Monitoring 2.000 -$200.000 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. No Maintenance-On-line 3.000 -$100.000 -$250.000 -$100. No Maintenance 1.

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 e Cash Flow 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 .

Failure Rate • Full Asset Utilization • Risk Reduction • Key Performance Indicators – Asset Family – Maintenance Process – Life Cycle Costs 58 .Future Trends in Maintenance • Sharing of Failure and Trouble data – Mode and Cause Level – Demographics • Application • Type – Population • Age models vs.

NE Olympia WA.skog@mtec2000. 98506 USA 59 .352.9977 2037 Berry St.Open Discussion and Questions: Need More Information? john.com 360.

Sign up to vote on this title
UsefulNot useful