DigitalClone® Wind Turbine Validations

DigitalClone

DigitalClone Wind Turbine Validations

2016 Sentient Science Corporation Confidential & Proprietary

1

DigitalClone is a Small Data ApproachA small data approach to determine component and system failure ratesDigitalClone Wind Validations2

When bearings enter failure, it happens very fast... minutes, hours, daysThat fast failure causes failure in other expensive componentsSave the bearing and you get life extension and a larger pay back than Big Data

Pedigree: This is the same process the military did with us. We were built from the ground up to work with the Technologies who manage the Death process.

30M investment from DoD, DoEBuilt for the worlds largest OEMs10 years design and validationIntroduced into wind18000 assets under contarctFocused on making the operators sucessfulWorking with suppliers to match operator demand with supply

2

Small Data ApproachOperators and suppliers will know the earliest point in time when critical components will begin to fail and have options to extend lifeDigitalClone Wind Validations3

Bearings are unique to other components in that they have a realitvely long initiation of failure (What Sentient Sees) and a extremely short propagation of failure (What sensors see)

When bearings propagate failure, unusable failure, it happens very fast, minutes, hours, days - if you are lucky. That fast failure causes failure in other expensive components. Save the bearing and you get life extension and a larger payback than big data.

Therefore Big Data from Sensors like GE Predix and the plethora of Big Data offers, answers are too late for Life extension.

3

Small Data ApproachThis means you now have much more information on asset life, life extension, financial impact

DigitalClone Wind Validations

Earliest Failure Reporting & ActionsKnown ComponentFailureAccurate FinancialPlanning18 month Rolling O&M ForecastingCBM OptimizationRisk Management4

4

Aligning with Our Customers Business GoalsDigitalClone Wind Validations5

OperationsManagementAssetManagementRiskManagementNewDevelopment

Reduce Energy Costs by 13% or $10/MWHSupplyChainBusinessStrategy6 Areas of Business Impact, Value and Focus


Overview of Sentients 6-step Technology ProcessDigitalClone Wind Validations6Calculate Component Stresses (bulk)Build Material Microstructure ModelBuild Surface ModelCalculate Probabilistic LifePredict Failure Mode & Component LifeSimulate Material Microstructure Response

Subsurface crack network

Contact SurfaceSub-surface

Ground Finish Superfinish2.1.3.4.5.6.

6

3 Life Extension Services DigitalClone Materials-Based Computational Testing (SaaS)

DigitalClone Live for OperatorsLife Extension of Gearboxes, Main bearings, Pitch Bearings

DigitalClone Live for SuppliersInterfacing Suppliers with OperatorsDigitalClone Wind Validations7

7

Sentient Predicted Failed and Damaged Gearboxes Correlated well with Field Data Wind Turbine Gearbox ValidationSentient Predicted Failed and Damaged Gearboxes Correlated well with Field Data DigitalClone Wind Validations8Turbine IDSentient PredictionsActual/Site InfoSentient RecommendationNotesAMech. failure Nov 2012Replaced March 2013N/APredictions line up perfectly with actual site dataBMech. failure next 6 monthsOn watch listDerate to 1.75MWSentient recommends turbine be inspected for main bearing pitting/spalling, derate to push back failure 1-2 monthsCMech. failure Aug 2012Replaced Sept. 2012N/APredictions line up perfectly with actual site dataDMech. failure any momentOn watch listDerate to 1.5MWSentient recommends turbine be inspected for main bearing pitting/spalling, derate to push back failure 1-2 months

Sentient Science Solution Reduce the cost of maintenance by transforming CM to PM through Optimization.$50-$55 MM of Maintenance Cost reduction potential in 5 years.

Business ProblemCM & PM mix as of today is 55:45 respectively. It will only get worse with aging.CM costs 30X more than PM. CM cost to go up with aging.


Wind Turbine Gearbox ValidationPredicted 38 Gearboxes with Damage, Confirmed by Site InspectionsDigitalClone Wind Validations9DigitalCloneFleet Failure DataCustomer knew of 18 critical assets

DigitalClone predicted an additional 20

Customer was able to plan maintenance with little to no unscheduled downtime over next two yearsParts on hand for replacements on site exactly when needed, no delaysNo failures caused immediate downtimeUnplanned failures were prevented, customer saved over $4.7M in first year


Gearbox Life Extension ValidationUptower and Derate RecommendationsDigitalClone Wind Validations10Sentient provided Watch List and Life Extension Action reports to customer during Dec. 2014Long term and short term visibility to failure, uptower repair was made during scheduled maintenance activity, and zero unplanned downtime was incurred.Traditional data analytics did not detect this failure

Butterfly WingsTurbine IDDigitalClone Predicted MTODDamage Noticed DateAction Taken for Life ExtensionLife Extension, MonthsASept, 2015Oct, 2015Uptower Replacement33BOct, 2016Aug, 2016Derate14


1.5 MW Fleet Failure Rate Predictions ValidationGBX Supplier Differentiation and ValidationDigitalClone prognostic modeling applied to the 1.5 MW gearboxesDigitalClone predicted cumulative fleet failure rate correlated well with customers failure dataIndividual years fleet failure rate and O&M costs predicted in the next 1 to 5 yearsOperators take actions to reduce this O&M cost increase with dynamic models to enable turbine life extension from:Optimizing Uptower Component ReplacementsGearbox Upgrades and Operational Changes (Lubricants, Derate/Uprate)Major Component Long Lead Time PlanningScheduled versus Unscheduled MaintenanceEvaluate New Gearbox and Components Life Extension before ProcurementDigitalClone Wind Validations11

Supplier 1Supplier 2Supplier 3Supplier 45 years10 years15 years20 yearsCustomers Cumulative Fleet Failure Rate as of 2016


DigitalClone SimulationBearing SupplierHardness (Ra) Range 0 to 4500 um depthMicrostructure InclusionsMicrostructure QualityD6058 HRCA60-53 HRC (Design requirement >58 HRC)

1.5 MW Gearbox Bearing Supplier Quality Prediction and Validation Differences in Bearing Suppliers Identified and Validated with Site FailuresDigitalClone Wind Validations12HSS Bearing DigitalClone ComparisonStress / Life Curve

Sentient predicted Supplier A HSS bearing failures within 8 years of operation. 100% of sites HSS bearing failures are with Supplier A, confirming DigitalClone predictions.Bearing material microstructure, surface finish, and material quality impact lifedepending upon location and stresses experienced.

BESTBESTBEST

Cycles to Damage (L66)Max Contact Pressure (MPa)

Stress Life CurveABetter LifeBetter Stress PerformanceOuter RaceInner Race

Outer RaceInner RaceD

BEST


ComponentSupplierPredicted L10 LifePlanet BearingSupplier A8 yrsSupplier C15 yrsSupplier D> 20 yrsLow Speed Intermediate Shaft Bearing UWSupplier A6.1 yrsLow Speed Intermediate Shaft Bearing DWSupplier A8.1 yrsHigh Speed Intermediate Shaft Bearing UWSupplier E> 20 yrsSupplier C14 yrsHigh Speed Intermediate Shaft Bearing DWSupplier E> 20 yrsSupplier B12.2 yrsHigh Speed Shaft Bearing UWSupplier E> 20 yrsSupplier C> 20 yrsSupplier B9 yrsHigh Speed Shaft Bearing DWSupplier E> 20 yrsSupplier C> 20 yrsSupplier B 7.7 yrsSupplier D 12.5 yrs

Risk ranking of bearing suppliers, confirmed by Customers site failures and teardown reportsSupplier A (highest risk)Supplier BSupplier CSupplier DSupplier E (lowest risk)

Validation - Effect of Bearing Supplier Quality on 1.5 MW Gearbox Life DigitalClone Wind Validations13Primary Component Risk:High speed shaft (HSS) bearingPlanet bearing Low Speed Inter. Shaft (LSIS) bearingHigh Speed Inter. Shaft (HSIS) bearingSecondary Component Risk:High Speed Shaft (HSS) Pinion and Inter. Speed ShaftPlanetary Gear/Sun GearLow Speed Shaft and Inter. GearRing GearMisalignment and bearing pitting / spalling then causes


13

Validation - Effect of Gear Supplier Quality on Gearbox Life DigitalClone Wind Validations14Business Problem

Need for advanced gear materials to meet new gearbox design requirements.It is expensive and time consuming to evaluate gearbox performance at various operating conditions and life accuratelyValidation & Sentient Science Solution

Sentient predicted the effect of gear material quality on performance, validated with physical test data1000x more virtual test points for gears at reduced time and cost over physical test Combining statistically significant virtual test points with limited physical test data provides accurate fatigue life estimates and allowable stress/loadsSimulate the affect on gearbox life from: Overload/over torque conditionsManufacturing ProcessLubricationResidual stressSurface finish


6MW Turbine Main Bearing ValidationSentient conducted RCA and predicted Main Bearing Damage Location, correlating with field failuresDigitalClone Wind Validations15Sentient used DigitalClone to predict the effect of site operating conditions on main bearing fatigue life and conducted RCA for premature bearing failuresDigitalClone predicted adhesive/abrasive wear (lubricant starvation) on the outer race as the primary damage mode leading to main bearing failure, confirmed by site observations

PredictedActual


Drives System Bearing Supplier Comparison and ValidationDigitalClone predicted that bearing Supplier D is superior compared to other suppliers, confirmed by physical test dataDigitalClone Wind Validations16Bearing supplier ranking based on material quality and life. Sentient predicted bearing life and ranking correlated with test dataOEM selected Supplier D for production

Bearing A Version 1Bearing A Version 2Bearing B Poor QualityBearing B Good QualityBearing CBearing DSupplier RankingSupplierTest Data L10 Life, HoursSentient Predictions L10 Life, Hours2Supplier A Ver1 &Ver 2 978.00 (Ver 1&2)622.58 (Ver 1)716.93 (Ver 1&2)764.25 (Ver 2)4Supplier B with & without Carbides308.00 (with and without Carbides)Without carbides - 653.90370.26 (with and without Carbides)With carbides - 130.453Supplier C380.00380.39N/A1Supplier D1056.0835.93N/A


16

Test dataAdvanced Bearing Modeling and Validation Case Hardened vs. Shot Peened vs. NitridedDigitalClone Wind Validations17Bearings OEM and Sentient Science began a computational testing and DigitalClone Supply initiative in 2015 to support bearing material upgrade initiatives. OEM could not test the life of Bearing C (nitrided) to failure, cost and time was prohibitive.Sentient evaluated 3 bearings with different material microstructures, residual stresses, and microgeometries. OEM selected nitrided bearings, validating DigitalCloneValue Assessment: Replaced $650K+ Expense of Equivalent Bearing Fatigue TestsReplaced 534 Days of Equivalent Bearing Fatigue TestsQuantified Life Extension of the Bearing

DigitalClonePhysical Test DataL10 Life, Hrs L50 Life, Hrs L90 Life, Hrs Weibull SlopeL10 Life, Hrs L50 Life, Hrs L90 Life, Hrs Weibull SlopeBaseline8.9133.6578.471.41814.8038.0469.431.64Peened9.03534.61481.471.40315.7647.7096.631.32Nitrided140.56233.66323.041.508----


17

DigitalClone® Wind Turbine Validations

Engineering