Generating Asset Health Indices which are both Useful and Auditable

Size: px

Start display at page:

Download "Generating Asset Health Indices which are both Useful and Auditable"

Dustin Cain
6 years ago
Views:

1 MIPSYCON 2016 Generating Asset Health Indices which are both Useful and Auditable Tony McGrail, Jay Garnett Doble Engineering

2 Summary: Asset Health Index (AHI) AHI is an estimate of a generally unknownvariable: the actual transformer health Estimating what healthy enough to do what, exactly? More data, should be a better estimate? The index should be useful what does the letter/number mean: Answer a question An index should have associated action & timescale Timescales should calibrate be consistent How to make it useful? Support decisions: justify actions in an auditable manner

3 Estimates Action? Timescale? Justification? Probability? Cost? Consequence? Risk?

4 What s in ISO 55001: Asset Management Standard? Foreword Introduction 1) Scope 2) Normative References 3) Terms & Definitions 4) Context of the Organization 4.1) Understanding the org & context 4.2) Understanding needs of stakeholders 4.3) Determining scope of AM system 4.4) AM System 5) Leadership 5.1) Leadership & Commitment 5.2) Policy 5.3) Organized roles/responsibilities/authorities 6) Planning 6.1) Actions to address risks/opportunities for AMS 6.2) AM Objectives and planning to achieve them 7) Support 7.1) Resources 7.2) Competence 7.3) Awareness 7.4) Communication 7.5) Information Requirements 7.6) Documented Information 8) Operation 8.1) Operational planning & control 8.2) Management of change 8.3) Outsourcing 9) Performance Evaluation 9.1) Monitoring, measurement, analysis & evaluation 9.2) Internal audit 9.3) Management review 10) Improvement 10.1) Nonconformity & corrective action 10.2) Preventive action 10.3) Continual Improvement

5 Definitions (from ISO 55000) Term Asset Asset Management Asset Portfolio Asset System Critical asset Risk Uncertainty Criticality Impact Working Definition an item, thing or entity which has value or potential value Coordinated activities of an organization to realize value from assets Assets within the scope of an asset management system Set of assets which interact or are interrelated Asset having the potential to significantly impact on the achievement of the organizations objectives Effect of uncertainty on objectives; Uncertainty is the state, even partial, of deficiency of information related to, understanding or knowledge of an event, its consequence, or likelihood. Not defined Not defined used in reference to critical asset Consequence Not defined used in risk definition and refers to ISO Guide 73,

When will your asset fail? Why will it fail?

deterioration? Mechanical damage? Thermal performance?

Time scale for failure mode to progress Failure mode parameters

6 When will your asset fail? Why will it fail? Need population statistics Identify failure modes: Insulation deterioration? Mechanical damage? Thermal performance? ROOT CAUSE ANALYSIS! Time scale for failure mode to progress Failure mode parameters which can indicate an issue Mitigate: Reduce probability of occurrence Reduce consequence of occurrence AHI: ISO55000 section 9.1: Monitor/measure/evaluate/analyze 6

Quotes - Asset Management Which maintenance, monitoring or asset replacement tasks are really worthwhile and when? If engineers do not talk the language of business, they do not deserve to be heard.

7 Quotes - Asset Management Which maintenance, monitoring or asset replacement tasks are really worthwhile and when? If engineers do not talk the language of business, they do not deserve to be heard. CEO of the Institute of Asset Management If top management don t talk the language of risk and criticality and understand why maintenance can be an investment they do not deserve to be leaders. We need a way to translate between the two groups. Asset management is all about the money, in the same way football is all about the ball.

8 Translation English-Welsh

9 Aged assets but which ones need attention? Age does not necessarily imply condition 40 Substation Operating Transformers Age Profile National Grid - EDO 100% DxD 80% Quantity Units without age data 26.83% TxD 60% 40% 20% Cumulative % 0 0% Year Condition alone doesn t tell us about risk IEEE C57.91 Loading Guide: expected life of a transformer: 180,000 hours (20.55 years)

10 Waterfall

11 Intervention: Double Feedback Loops Need to identify which is appropriate for a particular unit Environmental effects Environmental effects Target: healthy transformer Long term +/- testing and +/- analysis Short term testing and monitoring Dealing with Anomaly Long term intervention (outer loop) Short term intervention (inner loop)

12 Planning Intervention Collect data: Analyze data: ID anomaly: Diagnose: Prognosis: Intervene: Monitor: Iterate: condition, operation, family/design, industry etc looking for anomaly and outliers based on individual asset data, which may combine condition and operation data or aggregate data across several assets or asset types find possible causes what is likely to happen, with what consequences, when? replace a failed unit; repair/refurbish a unit that is failing gracefully ; plan longer term investment; provide assurance via monitoring analyze, review, and check intervention further data collection IEEE Data and Decisions 2011 Smart Grid Conference, Perth, Australia

Real World Ken Wyper, Graeme MacKay : TransGrid, Australia Tony McGrail:

13 Graceful and Rapid Onset Bushing Failures Graceful over months predictable Rapid minutes less predictable Condition Monitoring in the Real World Ken Wyper, Graeme MacKay : TransGrid, Australia Tony McGrail: Doble Engineering, USA International Conference of Doble Clients, 2013, Boston USA

14 Tire Pressure Data Mining Data processing removes information F L Θ F R Data summing tire pressures B L B R Mileage Dist MPH Revs MPG Ambient Sum Vec sum Noisy sum FL FR BL BR

15 Will an Asset Health Index help? What problem are you trying to solve?

16 Index and Indices What data do we have? Quite possible to start with basic nameplate and DGA data What scale should we use? What are we scoring? If we use 1 10, say: 1 is new/good and 10 is about to fail : what does 6 mean? how does 6 compare to 7 is 6 twice as bad as 3? Twice as likely to fail? How accurate or precise is the number? What does 3.9 mean? And what if it changes to 4.2???? Can we relate the health index to a likelihood of failure? Only if we have identified the failure modes Key to extracting value: action and timescale Key to keeping your job: justifiable audit trail

17 Combine? Calibrate timescales Code Code Description 1 No known problems 3 Slightly unusual dissolved gas signature 10 Possible arcing/sparking or partial discharge fault 30 Severe arcing/sparking Description or partial discharge fault 100 Very severe arcing/sparking or partial discharge fault, transformer at high risk of failure 1 transformer is expected to last for the foreseeable future, and at least 15 years 2 transformer is expected to last up to 15 years but may need to be replaced in 5-15 years 3 transformer is expected to last up to 5 years and may need to be replaced in 2-5 years 4 transformer is on active list for replacement within 2 years

18 Uniform weighting 1 good, 5 bad Factor Score Range Trf 1 Trf 2 Trf 3 DGA Main Tank Score Dielectric Score Thermal Score Mechanical Score Oil Score DGA LTC Tank Score Operational Score Design/manufacturer Score Subject Matter Expert Score Sum Normalized Sum (%) units with similar weighted score Which unit needs our attention? Which is most urgent? Which unit is most likely to fail?

19 Logarithmic scaling? 1 good, 100 bad Scale calibration Linear Log Factor Score Range Trf 1 Trf 2 Trf 3 DGA Main Tank Score 1,3,10,30, Dielectric Score 1,3,10,30, Thermal Score 1,3,10,30, Mechanical Score 1,3,10,30, Oil Score 1,3,10,30, DGA LTC Tank Score 1,3,10,30, Operational Score 1,3,10,30, Design/manufacturer Score 1,3,10,30, Subject Matter Expert Score 1,3,10,30, Sum Normalized Sum %) Same units as before now on log scales Which unit needs our attention? Which is most urgent? Which unit is most likely to fail?

20 Aspects of practical systems Are they useful? Do they help make decisions? Do they provide justification and an audit trail? Mode of failure Time to failure Short term v longer term

21 Delphic Approach DGA used as a base Review available data as a roundtable exercise Include design type, DGA, off line test, local knowledge Code Score the members of the population Review based on known failure rates Description 1 transformer is expected to last for the foreseeable future, and at least 15 years 2 transformer is expected to last up to 15 years but may need to be replaced in 5-15 years 3 transformer is expected to last up to 5 years and may need to be replaced in 2-5 years 4 transformer is on active list for replacement within 2 years Just how many code 3 and code 4 s should there be? Add more data and review as data becomes available At 0.5% per year failure rate Code 4: about 1% of population Code 3: about 1.5% of population

22 System : Reviewing a weighting system Practical system in use at present Weighted scores for several components Combined sum interpreted as a Probability of Failure (PoF) Data displayed on a risk chart Iso-risk lines used as reference predetermined by utility Track variation in one transformer S5 with variation in single key factor

23 Reality: Weighted system % Final Sum Weighting Substation Bushing Transformer Bushing Power Age Failure TYPE TYPE Power FQ Faults Load TYPE Power Factor Factor Score Rate LTC Bushing Factor AHI Consequence RISK S S S S S S S S S S Max ASset Health Index Weighted AHI Scores & Risk Weighted AHI Scores & Risk S6 S7 S2 S9 S10 80 Data point (AHI,Consequence) S4 70 S6 S7 S1 S3 60 S8 S2 S S10 Isorisk lines at: S4 S1 S3 20 risk = 225 S8 S S5 1 risk = ASset Health Index Consequence of Failure Consequence of Failure

24 Variations Bushing goes to worst possible condition 100 Weighted AHI Scores & Risk 100 Weighted AHI Scores & Risk ASset Health Index S6 S7 60 S2 S9 50 S S4 S1 S3 20 S S Consequence of Failure ASset Health Index S6 S7 S2 S9 S10 S4 S1 S3 S8 10 S Consequence of Failure 100 Weighted AHI Scores & Risk 100 Weighted AHI Scores & Risk ASset Health Index S6 S7 60 S2 S9 50 S S4 S1 S3 20 S8 10 S Consequence of Failure ASset Health Index S6 S7 S2 S9 S10 S4 S1 S3 S8 S Consequence of Failure System is good for longer term investment justifications - short term operational response covered elsewhere - mode of failure, time to failure

25 Overview Transformer Analysis Methodology was originally developed within National Grid UK, and has been since developed by Doble Engineering. Methodology has been developed through almost twenty years experience. Methodology has been applied, in whole or in part, to various other transformer operators, including transmission and distribution companies, generating companies, and large industrial companies. Goals of the Analysis? Determine the health of our critical transformers Improve our Transformer Fleet Reliability Performance Consistent updating of information for strategic planning and maintenance Have a documented assessment to drive planning and maintenance

26 Process Component scores for several factors each scored logarithmically: Core/windings Dielectric Thermal Mechanical Oil Ageing Contamination OLTC Exterior condition And other factors Combined into an overall score using addition of log scores Poor scores stand out Overall score is given in three ways: Now based on latest assessment Mitigated a result of intervention Possible improvement as a result of intervention

27 Outcome League table for transmission operator sorted by worst overall condition score Current and Mitigated Condition Design/Manufacturer Possible improvement in score Component score based on sub-components Overall Condition Core and Windings Oil OLTC Exterior T-No Ratio Rated PoweManufactureDesign PowerManufacturerDesign Famsign Family Design Score Year Now Mitigated Possible Improvement Dielectric Thermal Mechanical Ageing Contamination T /132 kv 240 MVA AEI WythenA04a Wythenshawe T /132 kv 120 MVA EEC E11b T /275 kv 1000 MVA GEC G02b T /132 kv 120 MVA EEC E11b T /66 kv 180 MVA CP D T /132 kv 120 MVA MVE M T /275 kv 750 MVA HHE H T /33 kv 100 MVA PPT P T /132 kv 240 MVA CAP C T /132 kv 240 MVA HHE H07a T /132 kv 240 MVA AEI WythenA04b Wythenshawe T /132 kv 220 MVA PPT P06a T /132 kv 240 MVA HHE H07a T /132 kv 120 MVA EEC E11a T /132 kv 240 MVA HHE H07a T /132 kv 120 MVA EEC E11b T /132 kv 120 MVA FER F T /132 kv 180 MVA FUL L T /132 kv 240 MVA AEI WythenA04b Wythenshawe T /66 kv 120 MVA AEI Rugby A

28 System: Which Transformers to Replace? Ranking system developed by utility Initial Risk = Health X Criticality Health = (Initial Health + Dynamic Health) Final Risk = (Initial Risk X Age Factor) + LTC Type Factor

29 Ranking Ranking is Done Using the Final Risk NAME Position Equip. MFG_NAME Health Criticality Initial Risk Final Risk Ranking Class SubstationA TR01 13KV MFG A SubstationB TR05 35KV MFG B SubstationC TR03 13KV MFG C SubstationD TR02 13KV MFG A SubstationE TR01 13KV MFG B SubstationE TR02 13KV MFG D SubstationF TR06 13KV MFG A SubstationG TR01 35KV MFG B SubstationH TR03 13KV MFG B SubstationI TR03 35KV MFG E SubstationJ TR04 13KV MFG D SubstationK TR02 13KV MFG D SubstationL TR04 13KV MFG E SubstationM TR01 13KV MFG F Failure mode?

30 How do you know which mode? How do you know the timescale? Revisiting the Waterfall

31 Revisit Planning Intervention Collect data: Analyze data: ID anomaly: Diagnose: Prognosis: Intervene: Monitor: Iterate: condition, operation, family/design, industry etc looking for anomaly and outliers based on individual asset data, which may combine condition and operation data or aggregate data across several assets or asset types find possible causes what is likely to happen, with what consequences, when? replace a failed unit; repair/refurbish a unit that is failing gracefully ; plan longer term investment; provide assurance via monitoring analyze, review, and check intervention further data collection IEEE Data and Decisions 2011 Smart Grid Conference, Perth, Australia How do you know your actions are working?

32 Are you using Raw data or Derived data Possible definitions: Raw data is measured Derived data is calculated What does this mean??? Who can check?

33 Working on a Probability of Failure Using: Population statistics Modes of failure in operation and timescale Identify individual probabilities and sum? Identify exposure of each risk (the expected annual value) and sum? Acceptability of failure this is asset risk management Need good statistics or good estimates Where did our original timescales for action come from? 33

34 Probability and Uncertainty Probability can be calculated precisely Uncertainty must be estimated Risk management combines probability and consequence Good decisions may still have bad results Let s roll a dice: if it comes up as any of 1-5 you win $1000 If it comes up as a 6, you lose $10 The expected value is you win >$800 but you can still lose! 2015 Doble Engineering Company. All Rights Reserved 34

35 Asset Management: Condition ranking, risk ranking Mode of failure time to failure what does the raw data say?

36 Asset Management: Condition ranking, risk ranking Need to see the details

37 Summary: Asset Health Index (AHI) AHI is an estimate of a generally unknownvariable: the actual transformer health Estimating what healthy enough to do what, exactly? More data, should be a better estimate? The index should be useful what does the letter/number mean: Answer a question An index should have associated action & timescale Timescales should calibrate be consistent How to make it useful? Support decisions: justify actions in an auditable manner

38 Any Questions? Tony McGrail Doble Engineering Company Thank you!

MANAGING THE RISKS ON TRANSFORMER ASSETS

MANAGING THE RISKS ON TRANSFORMER ASSETS by ARUN D YARGOLE & TONY McGRAIL DOBLE ENGINEERING COMPANY ayargole@doble.com What is an Asset? In commercial terms- It is an Entity that gives the owner the revenue