Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact...

30
© 2007 Impact Technologies, LLC. All Rights Reserved. Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Carl Byington, P.E. Director, Impact Technologies, LLC Director, Impact Technologies, LLC carl carl . . byington byington @impact @impact - - tek tek .com .com Damage Forecast Time Damage +2σ -2σ +2σ -2σ Present 0 50 100 150 200 250 300 350 400 450 500 100%- 90%- 80%- 70%- 60%- 50%- 40%- 30%- 20%- 10%- 0%- Predicted Measured Measured Relative Health Index 0 0.1 0.2 0.3 0.4 0.5 0.6 1 Failure Good Marginal Mild Wear Critical

Transcript of Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact...

Page 1: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Impact Technologies PHM and Structural Health Management

Carl Byington, P.E.Carl Byington, P.E.Director, Impact Technologies, LLCDirector, Impact Technologies, LLCcarlcarl..byingtonbyington@[email protected]

Damage Forecast

Time

Dam

age

+2σ

-2σ

+2σ

-2σPresent

0 50 100 150 200 250 300 350 400 450 500

100%-

90%-

80%-

70%-

60%-

50%-

40%-

30%-

20%-

10%-

0%-

PredictedMeasured

Measured

Rel

ativ

e H

ealth

Inde

x

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

FailureGood Marginal

Mild Wear Critical

Page 2: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Who is Impact?

Founded in 1999, currently ~90 employees, experienced Mechanical, Electrical and Software Engineers and Scientists dedicated to Predictive Equipment Health Management TechnologiesCustomers include: U.S. Navy, Air Force, Army, DARPA, DOE, EPRI, Boeing, Honeywell Engines, General Dynamics, Northrop Grumman, GE, Rolls Royce, P&W, UTC, NASA, Lockheed Martin, Dresser-Rand, etc.Top DOD Small Business contractor in the U.S. for Automated Health Management technologiesTypical Roles: Technology Research and Development, Software Development and Licensing, System Design and Integration, Engineering Support3 offices: Rochester, NY; State College, PA & Atlanta, GA

Page 3: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Advanced Sensor & Embedded Monitoring

Integrated Vehicle Health Management System Specific

Diagnostics/Prognostics

Electronics/Avionics PHM

Automated Equipment Health Monitoring SW

Decision Aids, Maintenance & Logistics Tools

Impact Portfolio

Larger Indicators & Labels

Mode DetectionIndicator

NetworkStatus Info.

StreamingOperatingConditions

Fault DetectionAnd Isolation

Page 4: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Array of PHM Technologies

Drive TrainVibration MonitoringBearing/Gear PrognosticsShaft MonitoringClutch DiagnosticsLube System Monitoring

PropulsionPerformance Diagnostics & PrognosticsBlade/Disk Usage & Useful Life RemainingEngine Fuel System Monitoring Lube Oil Monitoring

Actuator & ControlsReal-time Control/ Response ModelingEMA/EHS Actuators Model and Feature-Based

Sensors/DataSteady State/ Transient Mode DetectSensor Validation ModuleVirtual Sensors

AvionicsEvidence-Based BIT ReasonerSystem-Level Bayesian NetworkPower Supply MonitoringRadar System Monitoring

StructuresDamage LocalizationSeverity DeterminationContinuous Monitoring and Prediction of Structural Health

Fuel/HydraulicsFuel System ModelsPump and Valve PHM

Page 5: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Some SHM Developments

AFRL Composite Damage Localization

JSF Gearbox Corrosion Modeling

NAVAIR Gear Prognosis

DARPA Structural Integrity Prognosis

Systems

Page 6: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Michael J. Roemer, Jianhua Ge, Alex Liberson

Impact Technologies, LLC

Dr. G. P. Tandon, Dr. R. Y. KimUniversity of Dayton Research

Institute

AFRL Damage Detection, Isolation and Prediction

maximum triangulation numbermaximum triangulation number

Data Features/Models Prediction

Page 7: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Training Processing of the Neural Network for Uni-Directional Composite

Relationship Between Damage Size and RMS of Residuals for Uni-Directional Composite

Neural Network Training Results(Total 24 (4*6) sensors)

321.510.50Truth Size of Damage

3.0019

1.91421.47581.04540.46340.0142Size of Damage (Output of the Neural Network)

4.3759

2.26591.46790.84110.28390RMS of Residuals (Input of The Neural Network)

Results for Determination of Damage Size via Neural Network for Uni-Directional Composite Damage

Page 8: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

The line with maximum Residual

Two red lines that are parallel with the line of maximum residualAnd the distance is decided by the output of the neural network (size of damage).The damage should be in the area between two red lines.

The circle is the truth damage.Other blue lines denote the largest residuals in direction of 0, 30, 45, 60, 90, 120, 135, and 150 degrees.The right line shows the estimated damage geometry by the operator using the information from the left figure.

Determination of the Damage Geometry

Half of the output of the NN

Page 9: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Size of truthdamage

LocationOf Damage

GeometryOf Damage

0.5 inch

3 inch

2 inch

1.5 inch

1 inch

circle

circle

circle

circle

circle

Size of DamageFrom Neural Network

0.4634 inch

1.0454 inch

1.4758 inch

1.9142 inch

3.0019 inch

Location and Geometry(red area) UsingTriangulation +Knowledge about sizeOf damage

Truth Data-

Diagnosis Results

Uni-directional Composite Results

Page 10: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Accelerometer-Based Feature Extraction and Analysis

Flexural Wave Propagation in a Laminate Composite Plate

Page 11: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Feature Extraction Using Wavelet Packets and Higher-Order-Statistics

Raw Signal of Sensor 1

Raw Signal ofSensor 4Raw Signal of Sensor 3

Raw Signal ofSensor 2

Adaptive Energy Detector Adaptive Energy Detector

Adaptive Energy DetectorAdaptive Energy Detector

Time Feature:Time at which wave arriving time from impact location to

each sensor

Energy Feature: The maximum energy

amplitude

Page 12: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Energy & Arrival Time Feature Analysis

Area 1

Area 4Area 3

Area 2

Time FeatureDistance from

Impact Location to Sensor 1

Sensor 1

NeuralNetwork 4

Time FeatureDistance from

Impact Location to Sensor4

Sensor 4

NeuralNetwork 3

Time FeatureDistance from

Impact Location to Sensor 3

Sensor 3

NeuralNetwork 2

Time FeatureDistance from

Impact Location to Sensor 2

Sensor 2

Energy Analysis Arrival Time Analysis

NeuralNetwork 1

Page 13: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

NN Outputs and Triangulation for Seven Training Cases

Page 14: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Estimated impact location

Truth impact location

Test Case 1

Sensor 1

Sensor 3 Sensor 4

Sensor 2

Page 15: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Test Case 2

Estimated impact location

Truth impact location

Sensor 1

Sensor4Sensor 3

Sensor 2

Page 16: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Integrated Approach for Determining Damage Location and Severity for Composite StructuresStructural Features – Resistivity, AccelerationModels - Finite Element Approach for Determining the Impact LocationEvaluation of Independent Approaches to Structural Health Monitoring Continuing Work - GEAE Engine Half-Case

Composite Effort Summary

Page 17: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Carl Byington, Ryan Brewer, Sanket Amin, Vijit Nair, Adam Mott

Impact Technologies, LLC

Page 18: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Corrosion PHM Framework and Elements

Page 19: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Corrosion-based Fault-to-Failure Prognostic Horizon

TRACE WATER

WATER CRITIC

AL

> SATURATIO

N LIMIT

STRESS CORROSIO

N

CRACKING

LOCALIZED

MICRO PITTIN

GMACRO PITTIN

GSPALL A

ND SIGNIFIC

ANT

MATERIAL LOSS

GEAR/BEARIN

G

FAILURE

GEARBOX FUNCTIONAL

FAILURE

PROGNOSTIC HORIZON PROVIDED

Smart Oil Sensor™

GearMod™ & ImpactEnergy™

FX,Y,Z FA,B,C

FADEC CONTROL

OBSERVABLES

ROOT CAUSES & EFFECTS

Proposed Corrosion Monitoring Enhancements

Corrosion Life Modeling

104

106

108

1010

-50

0

50

100

150

200

250

3008620 Stress-Cycle Diagram

CyclesSt

ress

(kps

i)

100%

0%

IncreasingContamination

Decreases Component Fatigue Life

SURFACE

CORROSION

Page 20: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Corrosion Reduces Fatigue Life

104 106 108 1010-50

0

50

100

150

200

250

3008620 Stress-Cycle Diagram

Cycles

Stre

ss (k

psi)

100%

0%

IncreasingContamination

Decreases Component Fatigue Life

Potential Fatigue Life Reduction Can Be Orders of Magnitude

Formulas Updated To Incorporate SOS Measurements

Phase I Focus Gear Pitting Initiation

AGMA Surface DurabilityFormulas

RTH

HLacwc CKS

CCss =

xcsMvoP

pc CCKKKIFd

TCs 2

2=

2

000,126 ⎟⎟⎠

⎞⎜⎜⎝

⎛=

RTpH

HLac

xcsmvo

Pac CKCS

CdCsCCKKK

IFnP

Page 21: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Phase I Corrosion Failure Progression

Corrosion was induced by incrementally adding water contamination to the oilThe test rig was operated under accelerated loading conditions to produce initiation and some progression

Corrosion life demonstrated was consistent with model predictions but accelerated testing extrapolation needs to be addressed

Tooth Flank Corrosion

Shaft Corrosion

Page 22: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Seeded Corrosion Fault

A more severe level of damage was created by subjecting gear teeth on a separate gearbox to FeCl3 acid

Damage clearly detectable with vibration and verified with visible determination

Corroded Tooth

Page 23: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Corrosion Effort Summary

Identified gearbox contamination sources, root causes of critical corrosion failure modes, and prognostics frameworkDemonstrated SOS ability to detect water in Aeroshellgearbox oilImplemented vibro-acoustic algorithms for gearbox failure testingEvaluate and selected corrosion model for Phase I feasibility demonstrationImplemented PHM addresses on-demand prognosticsfor gearbox mechanical components

Page 24: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Gear Prognostics Module Example

Solid Model Development

FEA and Fracture Mechanics

Crack

1 cm Correlation

Avinash Sarlashkar, Joel Berg, Eric Schenk - Impact

Page 25: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Adaptation Iteratively Improves Prognosis

Dam

age

a1

L1 L2

a2

Corrected Models

a3

L3

Current time

Time

Current time Current time

MeasurementUpdate

Page 26: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Test 1 & 2 - Failure Progression Results

Page 27: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

95%

42%

MTTF = 3.1 hrsMTTF = 3.1 hrs

100%

5% MTTF = 4.45 hrsMTTF = 4.45 hrs

No Vibration Feature Indications

Features Present -Model Calibrated

Time

Time

Distributionon Crack Initiation

Distributionon Crack Propagation

Benefits of Predictive Model Updating

Page 28: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

DARPA Structural Prognosis Approach

Characterize component materialmicrostructure, residual stresses, manufacturing defects, etc.

Use physics of failure models with anticipated usage to predict component damage and damage accumulation

Employ state awareness tools (system operating sensors, specialized, and virtual sensors) in real or near real time to monitor operating environment and detect extent of damage

temperature, stress, vibratory modes, etc.

Integrate model predictions and sensor data through reasoners to make a probabilistic assessment of future capability given intended mission Parameters

David Muench, Liang Tang, Brian Walsh, Avinash Sarlashkar, Joel Berg, Mike Koelemay, Greg Kacprzynski - Impact Technologies

Steve Engel, et al. – Northrop Grumman

Page 29: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Applied to Aircraft Panel

Page 30: Impact Technologies PHM and Structural Health Managementcjl9:byington_impact.… · Impact Technologies PHM and Structural Health Management Carl Byington, P.E. Director, Impact Technologies,

© 2007 Impact Technologies, LLC. All Rights Reserved.

Summary

Impact is involved in a range of PHM and SHM programs dealing with understanding of failure modes, symptom/effects, sensing capability, failure physics modeling, and predictive tools PHM/SHM needs span across services and DARPA has recognized need for structural prognosis capabilityGood collaborative interests with Penn State Center

Projects and Research FacultyEngineering Co-op and Graduate Students

Thank you!