STRUCTURAL HEALTH MONITORING OF COMPOSITE … · M.E.T.I-System Suite of Damage Detection Devices...

ProprietaryM.E.T.I-System Suite of Damage Detection Devices

STRUCTURAL HEALTH MONITORING OF COMPOSITE MATERIALS

Monitoring & Evaluation Technology Integration System: M.E.T.I-System Suite of Damage Detection Devices

Seth S. Kessler, [email protected]

46 Second Street • Cambridge, MA 02141 • 617.661.5616 • http://www.metisdesign.com

2ProprietaryM.E.T.I-System Suite of Damage Detection Devices

Outline

• Motivations for Structural Health Monitoring (SHM)• Components for necessary for a SHM system• Comparison of SHM damage detection methods• Lamb wave methods• Current research

test setup optimizationalgorithms and softwareresults

• Overall goals for SHM• Future Research ideas


SHM Motivations

• Structural Health Monitoring (SHM) denotes a system with the ability to detect and interpret adverse “changes” in a structure in order to reduce life-cycle costs and improve reliability

• Greatest challenge in designing a SHM system is knowing what “changes” to look for, and how to identify them

• Applicable to any field – highest payoff in air/spacecraft


SHM Minimizes Life-Cycle Costs

• Inspectiondramatically reduce or eliminate scheduled inspectionscurrently about 25% of aircraft life cycle cost is spent in inspectionseven greater percentage for spacecraft

• Maintenanceimprove efficiency and accuracy of maintenancecommercial airlines spend a combined $10 billion/year on maintenancecondition based maintenance could reduces these costs by 33%

• Designuse CBM over damage tolerant design – reduces weight up to 25%increased range and fuel economy of aircraftreduce operational down-time, thereby capturing more revenue


SHM Improves Structural Reliability

• Enhancement of detection methods and methodologiescan catch damage that may have occurred between scheduled intervalsmost inspection is currently visible, forms of damage can be overlooked

• Wide variety of aerospace applicationsmuch of the airline and military fleet are ageing aircraft, fatigue issuesintelligent structures are a key technology for quick turnaround of RLV’s

• Integration issuesretrofit SHM systems into existing vehicles to monitor damage growthintegrate SHM networks into new vehicle designs to guide inspections and dictate maintenance and repair based upon need


SHM in Composites

• Most new vehicles utilize advanced composite materials due to their high specific strength and stiffness

• Different areas of concern for NDEmetals: corrosion and fatigue vs composites: delamination and impactdamage below the visible surface is most important for composites

• Composite generally allows a more flexible SHM systemability to embed to protect sensors or actuators can tailor structure with SMA or E&M conductive materialshigher likelihood of sensors initiating damage

• May help relax public’s fear of commercially using composites ifthey are continuously monitored


SHM System Components• Architecture:

integration of system components for efficiency, redundancy and reliabilityreal-time VS discontinuous monitoring

• Damage characterization:identification of damage types for target application quantification of damage signature and effect on structural integrity

• Sensors: strain, vibration, acoustic emission, impedance, magnetic field, etc.active VS passive sampling methods

• Communication:both between neighboring sensor cells and global networkwired VS wireless

• Computation:locally control sensing systems and acquire data process and combine local and global data

• Algorithms: interpretation of damage location, severity, likelihood of failure• Power: store and supply electricity to each component• Intervention: actively mitigate damage, repair damage

Honeywell MEMS sensor

Rockwell RF receiver


Summary of Detection MethodsMethod Strengths Limitations SHM Potential

Strain gauge

embeddable simple procedurelow data rates

expensivelimited info

low powerlocalized results

Optical fibers

embeddablesimple resultsvery conformable

expensivehigh data ratesaccuracy?

requires laserlocalized results

Eddy current

surface mountablemost sensitive

expensive complex resultssafety hazard

high powerlocalized resultsdamage differentiation

Acoustic emission

inexpensivesurface mountablegood coverage

complex resultshigh data ratesevent driven

no powertriangulation capableimpact detection

Modal analysis


complex resultshigh data ratesglobal results

low powercomplex structuresmultiple sensor types

Lamb waves


complex resultshigh data rateslinear scans

high powertriangulation capabledamage differentiation


Size of Detectable Damage vs Sensor Size

Methods with best damage/sensor size ratio typically have low coverage, only Lamb wave and FR methods cover entire area, AE covers most


Size of Detectable Damagevs Sensor Power

Methods with lowest power requirement typically have lowest coverage; for Lamb wave and FR methods sensitivity scales with power level


Sensor Conclusions

• Piezoelectric materials are ideal for SHM applicationscan be used as actuators and sensors for a variety of NDE methodslight, low cost, low power, flexible, can be deposited

• Frequency response methodsuseful detection sensitivity to global damagecan be used for first or last line of defense

• Lamb wave methodssensitive to local presence of many types of damagepotential for triangulation of damage location and shape

• Recommendations for SHM system architecturedesign based upon analytical results and experimental tuninguse of multiple detection methods to gain maximum information


Lamb Wave Methods

• Form of elastic perturbation that propagates in a solid mediumactuation parameters determined from governing equationsexcite Ao wave for long travel distances and to minimize clutter

• Damage can be identified in several waysgroup velocity approximately ∝ (E/ρ)1/2, damage slows down wavesreflected wave from damage can be used to determine locations

• Present research utilizes piezoelectric actuators and sensors todetect energy present in transmitted and reflected waves


Thin Laminate FEA Results

• Figures on left show FEA results for coupon without damage• Figures on right show FEA results for coupon with 25 mm disbond• Movie files show z-displacement at 100 microsecond intervals• Can use to measure time-of-flight and observe reflections


Stiffened Plate FEA Results

• Figures on left show FEA results for stiffened plate without damage• Figures on right show FEA results for rib with 25 mm disbond• Movie files show z-displacement at 100 microsecond intervals• Disbond yields fringe pattern in both reflected and transmitted wave


Recent Lamb Wave Research

• Optimization of Lamb wave testing procedureselection of best actuating/sensing materials, adhesives, electrodesmore efficient actuating/sensing scheme for transmission and reflectionincreased reliability, robustness and signal strength by 4x

• Algorithm developmentdamage evaluation algorithms in MATLAB using filters and waveletstuned and validated by a large set of simple test resultscontinuing to improve and integrate algorithms into softwarecurrently yields report on presence, location, size and type of damage

• Test specimen databasetesting sensors/algorithms on composite plates with variety of damage testing sensors/algorithms complex sandwich structure specimensobserve effects of various core densities and thicknessobserve effects of disbonds, delams, hardpoints and gaps


Actuator/Sensor Schematic

Sensor

Actuator

Electrically conductive tape

Brass shim stock

Low temp thermoplastic tape

Complete actuator/sensor


Lamb Wave Test Setup

• Tests executed via PC laptop and NI data acquisition board• Completely portable, simple to use and automated results• HP oscilloscope and function generator have also been used


Data Reduction Procedure

• Automated software using adaptive algorithms to identify damage for a particular application eliminates human bias

• Goals for data interpretationpresence and severity of damagetype of damage differentiationlocation of damagesize and shape of damage

• Various forms of algorithms utilizedfilters and frequency separation used to decompose signalsseveral physical phenomena used to differentiate results series of reports generated comparing energy, frequency and time


METIS_v1 Software Executable

• Control version 1.0 of damage detection software

• Extracts data directly from data acquisition files in Excel

• Processes code in MATLAB, divided into several subroutines

• Uses several different physical phenomena to produce results

• Text output with damage information• Several series of plots to support

conclusions

%Damage detection software executable version 1.0

function metis_v1(y,x);% (control, test)

% Setup values from user values

% Initialize variablesinit

% Filter and wavelet decompositiondata

% Piecewise max wavelet coefficients and plotswavecoefwaveplots

% Computingsignalplotenergycomputearrivalcomputefrequencycomputereflectioncompute

% Reportingenergyreportarrivalreportfrequencyreportreflectionreportfinalreport


METIS_v1 Software Initializationfunction metis_v1(y,x);% (control, test)











• Code is executed as a function with control and test variable matrices as arguments

• Each matrix contains columns for time, near and far sensor voltage and actuated signal

• Values subroutine contains constants for analysis including frequency, actuator/sensor locations and thresholds

• Init subroutine gives variables initial constant values or zero


METIS_v1 Software Decomposition









• Data subroutine uses custom Butterworth bandpass filters and truncates data into desired format

• Wavecoef subroutine performs custom wavelet decompositions using modified Morlet mother wavelet

• Waveplots initializes each of the subsequent plot formats

• At this point in the code, all of the useful data has been extracted and “messaged” to be readily analyzed




METIS_v1 Software Computation









• Computation section is the “gut” of software• Four separate but related physical

phenomena used to interpret and compare energy integrationwave speeds arrival timesfrequency spectrareflected wave patterns

• Each subroutine is independent for upgrade • Share common function to locate wavepeak



METIS_v1 Software Algorithms

• Energy Reportintegrates magnitude of sensed voltage for near and far sensordamage suspected if energy-loss threshold exceeded

• Wave Arrival Reportcalculates actuation peak and arrival times at near and far sensorsdamage suspected if time-lag threshold exceeded

• Frequency Band Reportfrequency bandwidth of received signal calculateddamage suspected if band-shift threshold exceeded

• Wave Reflection Reportcalculates wave velocities with wavelet methods for triangulationthresholds are set for known feature and boundary reflections

• Cumulative Damage Summary Reportdamage presence and type from Energy, Arrival and Frequency reportsdamage location and size from Wave reflection report


Wavelet Analysis Waterfall Plots

• Waterfall plot displays all frequencies over signal time in 3-D• Allows accurate extraction of signal peaks and waveform trends


Waterfall Plot Cross SectionsWaveform Peak

controldamage

• Plotted for actuating frequency, points represent highest scales• Troughs represent peak of waveform, threshold used for noise


Reflection from Damage

Waveform PeakRegion in Question

Subtracted Resultant

• Undamaged wavelet coefficients subtracted from test signal• Reflected wave location is exposed in resulting array


METIS_v1 Software Reporting









• Reporting section extracts results from computation section to draw conclusions

• Text output produced for each method to support plots generated by code

• Finalreport subroutine amalgamates the results of all the methods to interpret overall state of damage

• Final output of likelihood of damage presence, type and location



METIS_v1 Software Output• Text output produced from

Reporting section of code• Reports for each method

integrated energy levelswave arrival timesfrequency bandwidth peakswave reflection locations

• Damage Summary combines reports to determine damage state

• Self-calibrating, uses wave between actuator and adjacent sensor to calculate wavespeed

• Each method checks on actuating signal to confirm actuator/sensors are functioning properly

------------------------------------------------------------------------------------------------------------------------ METIS_v1 Damage Detection Software Report -------------------------------------------------------------------------------------------------------------------------------------------------------- Energy Analysis ----------------------------------------------

Control Damage Difference RatioE Near Sensor 30.09 35.53 -5.44 0.85E Far Sensor 10.95 9.88 1.06 1.11

There appears to be a 18% gain of energy at the near sensorThe structure may be damagedFar energy losses are nominal. No damage has been detected with energy method

---------------------------------- Wave Arrivals -----------------------------------------------Control Damage Difference

Near Sensor 196.00 197.00 1.00Far Sensor 529.00 560.00 31.00

Sensor report exceeds damage tolerance. There may be a delamination

--------------------------------- Frequency Peaks ---------------------------------------------Control Damage Difference

Peaks Act 14.00 14.00 0.00Peaks Near 14.32 13.70 -0.62Peaks Far 14.65 14.00 -0.65

Tested frequency shift of 0.00 compared to nominal frequency shift of 0.65Band shift within tolerances.

-------------------------------Wave Reflection Report----------------------------------------

Reflected waves at 510.00 corresponding to 4.75 inchesWaves traveling at a speed of 23810 inches/second

------------------------------------------------------------------------------------------------------------------------------------ DAMAGE SUMMARY ----------------------------------------------------------------------------------------------------------------------------------------There may be a delamination at 4.75 inchesActuator/sensors are functioning properly.----------------------------------------------------------------------------------------------------


• Several pairs of sensors tested on undamaged plates• Reproducible response across different specimens

Comparison of Control Signals

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0-0 . 1

0

0 . 1

sign

als

(V)

F a r S e n s o r R e c e p t io n

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

5

1 0

1 5

Far t

otal

ene

rgy

(ms)

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Far w

ave

arriv

al

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 50

0 . 5

1

Far p

eak

freq

(kH

z)

N o D a m a g eTe s t A rt ic le


Delaminated Results: Far Sensor

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0-0 . 1

0

0 . 1

sign

als

(V)

F a r S e n s o r R e c e p t io n

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

5

1 0

1 5

Far t

otal

ene

rgy

(ms)

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Far w

ave

arriv

al

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 50

0 . 5

1

Far p

eak

freq

(kH

z)


• Delaminated signal is time-lagged, slightly lower energy content• Frequency bandwidth remains similar


• Energy plot shows reflection from delamination • Control plot flattens out as no reflections arrive at near sensor

Delaminated Results: Near Sensor

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 0-0 . 5

0

0 . 5

sign

als

(V)

N e a r S e n s o r R e c e p t io n

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

Nea

r tot

al e

nerg

y (m

s)

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Nea

r wav

e ar

rival

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 50

0 . 5

1

Nea

r pea

k fre

q (k

Hz)



0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Act

uate

d S

igna

l

Un d a m a g e d s p e c im e nTe s te d s p e c im e n

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Far S

enso

r

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Nea

r Sen

sor

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Sub

tract

ed N

ear S

enso

r

Delaminated Results: Subtracted

• Reflection revealed at around 510 us (also reflection from end)• Corresponds to delamination at 4.75” (actual at 4.8”)


Matrix Cracks Results: Subtracted

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Act

uate

d S

igna

l

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Far S

enso

r

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Nea

r Sen

sor

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 1 0 0 00

2 0

4 0

6 0

8 0

Sub

tract

ed N

ear S

enso

r


• Reflection revealed at around 493 us (also reflection from end)• Corresponds to cracking at 4.72” (actual at 4.8”)


0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Act

uate

d S

igna

l

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Far S

enso

r

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Nea

r Sen

sor

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Sub

tract

ed N

ear S

enso

r


• Reflection revealed at around 238 us• Corresponds to delamination at 4.76” (actual at 4.75”)

HD Al Core Results: Subtracted


0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Act

uate

d S

igna

l

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Far S

enso

r

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Nea

r Sen

sor

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 00

2 0

4 0

6 0

8 0

Sub

tract

ed N

ear S

enso

r


HD/S Al Debond Results: Subtracted

• Reflection revealed at around 202.5 us• Corresponds to debond at 4.73” (actual at 4.75”)


Lamb Wave Testing Conclusions

• Thin laminatescontrol laminates are very reproducible, used to calibrate thresholdsdelaminated specimens yield reflections with little change to bandwidthmatrix cracks yield frequency changes with some reflections

• Sandwich structureshigh density Al core specimens easily show disbond and impact damagelow density Al core specimens yield similar results, slightly dampedNomex & Rohacell core specimens are further dampedsolid thick and thin Al core have faster waves speeds, cleanest signals

• Mixed sandwich structureshigh density Al core with center gap easily differentiated from disbondhalf high density half solid Al produces wave with reflection at interfacehalf high density half solid Al and disbond can be measured over joint


Recent Research Conclusions

• Optimized Lamb wave actuator/sensors and test configurationincreased signal strength nearly a factor of 4 over prior researchmore reproducibility in signals helps interpretation; more work needed

• METIS_v1 damage detection softwareexhibits 83% accuracy, nearly 100% for delaminationeliminates most subjectivity with automation and thresholdsstill need to refine threshold levels and actuator/sensor locations

• Experimental resultsseveral combinations of small and controlled damage still yieldsmeasurable effect, the more damage, the more dramatic the resultsneed to refine wiring technique to yield more consistent voltage leveloverall highly successful project, able to detect damage in all specimens


Newly Funded Research

• Patterned electrode formations for actuator beam steeringwould enable a sonar-like imaging capability detect damage using either Lamb wave or acoustic emission

• Development of a thin-film polymer rechargeable batterywireless inductive-loop recharging systemmatching size and power specifications for an SHM system

• Study of packaging techniques needed for all of the SHM componentsisolate from harsh natural, mechanical,and electrical environments

• Testing of the developed sensor patchtesting on large built-up structuresnoisy operating environments (natural, mechanical and electrical)


Proposed Research

• Comparison of behavior and reliability of SHM NDE methods different materials (metals, CFRP, GFRP)manufacturing processes (uni-tape, woven fabric, filament wound)

• Attenuation studycompare sensor density and power requirements for compositevarying thickness, stiffness, density, core material and curvature

• Development of a wireless SHM capabilitiesdata acquisition and actuator function generating system matching size, power and speed specifications for an SHM system

• Economic study of SHM system implementation strategiesprovide justification for the use of SHM systemsoptimal strategy for their introduction into military and civil applications


M.E.T.I-System Suite of Damage Detection Devices

• M.E.T.I.-NDEportable quick and accurate non-destructive evaluation device solution for composite structures based on Lamb wave methods

• M.E.T.I.-Stripcapable of detecting damage present between any pair of narrow stripsideal for cylindrical structures

• M.E.T.I.-Patchpatches to be mounted in critical regions by a thermoplastic backingyields a detailed damage report for the area located below its surface

• M.E.T.I-Chipsmall chips will be designed to scan large area (~2m)detect presence and location of both surface and subsurface damage


Proposed SHM Device Architecture

• Piezoelectrics and other components on a thermoplastic patchstrain, vibration, acoustic emission, Lamb wavessome on chip processingwireless relay from patchto be placed in key locations, detect damage within patch area

• Neural network behavior (ant colony scenario)system to be calibrated pre-operation to understand orientationsseveral “dumb” sensors collectively making “smart” decisionssensors behave passively with AE and strain, occasional FRMwhen event occurs, will actively send Lamb waves to quarry damage,upon verification of damage convey to central processor

• Could gather information through ethernet port upon landing and run full vehicle pre-flight test as a preliminary insertion step


M.E.T.I-System Team

• Metis Design Corporation (Dr. Kessler)team lead, system architecture, system integration, algorithms, software

• MIT Department of Aeronautics/Astronautics (Dr. Spearing)sensor optimization, structure interface, manufacturing, packaging

• MIT Department of Materials Engineering (Dr. Sadoway)polymer battery, wireless recharging system

• University of Michigan Aerospace Department (Dr. Cesnik) actuator optimization, electrode patterning, environmental testing

• MicroStrain (Mr. Arms)wireless data acquisition, wireless actuation, data storage

• AFRL Advanced Composites Office (Lt. Kuenzi)end-user voice, interface requirements, weight/cost limits, testing

STRUCTURAL HEALTH MONITORING OF COMPOSITE … · M.E.T.I-System Suite of Damage Detection Devices...

Documents

Transcript of STRUCTURAL HEALTH MONITORING OF COMPOSITE … · M.E.T.I-System Suite of Damage Detection Devices...