impacts and mitigation of uncertainty shmii-2 · 2005-11-28 · impacts and mitigation of...

56
Impacts and Mitigation of Impacts and Mitigation of Uncertainty for Improving the Uncertainty for Improving the Reliability of Field Measurements Reliability of Field Measurements Kirk A. Grimmelsman & A. Emin Aktan Kirk A. Grimmelsman & A. Emin Aktan Drexel University Drexel University SHMII SHMII - - 2’ 2005 2’ 2005 The Second International Conference on Structural Health Monitor The Second International Conference on Structural Health Monitor ing of ing of Intelligent Infrastructure Intelligent Infrastructure November 16, 2005 November 16, 2005

Upload: others

Post on 19-Mar-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Impacts and Mitigation of Impacts and Mitigation of Uncertainty for Improving the Uncertainty for Improving the

Reliability of Field MeasurementsReliability of Field MeasurementsKirk A. Grimmelsman & A. Emin AktanKirk A. Grimmelsman & A. Emin Aktan

Drexel UniversityDrexel University

SHMIISHMII--2’ 20052’ 2005The Second International Conference on Structural Health MonitorThe Second International Conference on Structural Health Monitoring of ing of

Intelligent InfrastructureIntelligent InfrastructureNovember 16, 2005November 16, 2005

Page 2: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

ContentsContents

Introduction and definitionsIntroduction and definitionsContinuous monitoring of a pin and Continuous monitoring of a pin and hanger retrofithanger retrofitShort term testing and continuous Short term testing and continuous monitoring to evaluate fatigue lifemonitoring to evaluate fatigue lifeAmbient vibration testing of the Brooklyn Ambient vibration testing of the Brooklyn Bridge TowersBridge TowersSummarySummary

Page 3: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Health Monitoring ParadigmHealth Monitoring Paradigm

Health Monitoring: track health by data and analytical Health Monitoring: track health by data and analytical simulation so current and expected performance can simulation so current and expected performance can be described in a proactive mannerbe described in a proactive mannerHealth Monitoring paradigm offers great advantages:Health Monitoring paradigm offers great advantages:

Objective characterization of healthObjective characterization of healthProactive management of maintenanceProactive management of maintenanceEnabler of performanceEnabler of performance--based engineeringbased engineering

Requires integration of experimental, analytical, and Requires integration of experimental, analytical, and information technologiesinformation technologiesSystem Identification approach offers rational System Identification approach offers rational framework for optimum integration of these framework for optimum integration of these technologiestechnologies

Page 4: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Current State of Practice: Field Testing Current State of Practice: Field Testing and Monitoringand Monitoring

Practical objectives are primary motivation:Practical objectives are primary motivation:Reduce or quantify the uncertainty related to some Reduce or quantify the uncertainty related to some observedobservedbehavior, performance or mechanism (Reactive)behavior, performance or mechanism (Reactive)Improve the reliability of traditional design or analysis approaImprove the reliability of traditional design or analysis approachesches

Proactive structural health monitoring rarely consideredProactive structural health monitoring rarely considered

Lack of knowledge regarding capabilities/limitations of Lack of knowledge regarding capabilities/limitations of experiments experiments –– education emphasis is design instead of education emphasis is design instead of maintenance or renewalmaintenance or renewalLack of standards or guidelines for field experiments (need is Lack of standards or guidelines for field experiments (need is recognized, question is how to get there)recognized, question is how to get there)

Limited or no scientific characterization of how the Limited or no scientific characterization of how the parametrsparametrs/uncertainty in stages of experiment affect the quality and reli/uncertainty in stages of experiment affect the quality and reliability ability of the resultof the resultDesign, execution, and interpretation rely significantly on heurDesign, execution, and interpretation rely significantly on heuristicsistics

Page 5: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Types of Uncertainty Affecting Types of Uncertainty Affecting Reliability of Field Measurements Reliability of Field Measurements

Human Errors (HE)•Inattention/Thoughtlessness•Inexperience•Omission (Forgetfulness)•Commission (Bad Design)

Random Phenomena (RP)

Epistemic Uncertainty (EU)•Less Understood Phenomena (LUP)•Unknown Phenomena (UP)

EU - LUP

HE

RP

EU - UP

Page 6: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

INPUT• Non-stationary• Echoes/Reflections• Bandwidth• Directionality• Select Harmonics• Interference/Noise

STRUCTURAL SYSTEM• Non-stationarity due to changes in environment • Nonlinearity • Incomplete free body/Appendage tests• Lack of observabilitydue to insufficient sensor density • Scale-induced complexity

OUTPUT (DATA)• Asynchronous• Filters• Sensor calibration• Noise & bias• Spurious pulses• Bandwidth• Window length• Freq. resolution

DATA PROCESSING• Data quality measures• Error ID/Cleaning• Filtering, averaging, and windowing• Post-processing

MECH PROPERTIES• Frequency band• Modal order• Spatial adequacy• 3D vs. idealized• Separation• Amplitude & phase• Damping

ANALYTICAL MODEL• Completeness • Material variability• Geometry• BC & CC• Temporal/spatial Nonlinearity & non-stationarity

PARAMETER ID• Parameter grouping• Sensitivity• Bandwidth• Modality• Objective Function• Optimization

VERIFICATION• Modality • Independence

TEST DESIGN• Access• Excitation• Sensor density and modality• Diagnose/Mitigate malfunctions

Some Sources of Uncertainty in ExperimentsSome Sources of Uncertainty in Experiments

Page 7: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Suspended SpanCantilever Arm

Cantilever Arm

Anchor Span Anchor Span

822 ft 411 ft 822 ft 411 ft 822 ft

3,288 ftP.P. 27 P.P. 45

Fix.Exp. Fix.Fix.Exp. Exp.

822 ft 822 ft411 ft411 ft 822 ft

3,288 ft

Commodore Barry Bridge

Continuous Monitoring to Characterize a Pin & Continuous Monitoring to Characterize a Pin & Hanger RetrofitHanger Retrofit

Page 8: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

1

2

1

3

4

Retrofitted Truss Hanger Locations

Each hanger is retrofitted with four 7.5-inch diameter stainless steel rodsThe rods are pre-tensioned to remove DL from the truss hangers.

Suspended Span Hanger RetrofitSuspended Span Hanger Retrofit

Page 9: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Typical Pin and Hanger MemberTypical Pin and Hanger Member

Top Joint Suspended SpanCantilever Arm

Han

ger

Pin

Pin

Pin

Hanger

Top and bottom pins prevent bending moments due to longitudinal expansion/contraction of the truss or live loads

from developing on the hanger cross section30

m

Page 10: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Auxiliary Support SystemAuxiliary Support System

The 4 pin and hanger members are The 4 pin and hanger members are retrofitted with an auxiliary support system retrofitted with an auxiliary support system to provide added redundancyto provide added redundancySupport system consists of:Support system consists of:

4 large diameter stainless steel rods4 large diameter stainless steel rodsfour spreader beams located at the upper and four spreader beams located at the upper and lower truss chordslower truss chordssplice couplers, nuts, and washerssplice couplers, nuts, and washers

Rods extend vertically from the upper chord Rods extend vertically from the upper chord of the cantilever arm to the lower chord of of the cantilever arm to the lower chord of the suspended span the suspended span

Page 11: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Auxiliary Support SystemAuxiliary Support System

Spreader Beam @ U.C.

Spreader Beam @ L.C

Splice Coupler

7.5” Diameter SS Rod

Hanger Sensors

Page 12: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Scope and Objectives of the MonitoringScope and Objectives of the MonitoringScope: uncontrolled testing before, during, and Scope: uncontrolled testing before, during, and after installation of the auxiliary support systemafter installation of the auxiliary support systemObjectives: ImmediateObjectives: Immediate

Axial force change for each hangerAxial force change for each hangerAxial forces in rodsAxial forces in rodsIdentify if bending of hangers occursIdentify if bending of hangers occursEvaluate consistency of tensioning at 4 Evaluate consistency of tensioning at 4 hanger locationshanger locationsAre any significant stresses induced in other Are any significant stresses induced in other membersmembers

Objectives: LongObjectives: Long--TermTermMonitor performance of systemMonitor performance of system

Page 13: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Typical Instrumentation SchemeTypical Instrumentation Scheme

Geokon VSM 4050 Vibrating Wire Strain Gage Hanger Elevation View

A

Hanger

Hanger Rod (Typ.)

A

C.L. Pin

B B

Section A-A Section B-B

6’-0”

Hanger Rod (Typ.)

22’-0”

Page 14: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Sensor Installation on RodsSensor Installation on Rods

Covers to minimize temperature effects on sensor readings

Strain gages bonded to S.S. rod with epoxy

VW Strain Gage

Page 15: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Results Results –– Strain Response of Hanger Strain Response of Hanger

Page 16: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Results Results –– Strain Response of RodsStrain Response of Rods

Page 17: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Observations from Raw DataObservations from Raw Data

Consistent behavior observed at all 4 Consistent behavior observed at all 4 hanger locations as rods were tensionedhanger locations as rods were tensionedMagnitudes of residual strains in rods and Magnitudes of residual strains in rods and hangers after tensioning slightly different hangers after tensioning slightly different at each hanger locationat each hanger locationAverage strains on rod and hanger cross Average strains on rod and hanger cross sections used for subsequent analysessections used for subsequent analyses

Page 18: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Uncertainty Uncertainty -- Data AnalysisData Analysis

Axial forces on rods and hanger calculatedAxial forces on rods and hanger calculatedHooke’sHooke’s Law (F = Law (F = ε ε xx E x A) E x A) Isolated free body assumptionIsolated free body assumption

Nominal material properties used to Nominal material properties used to calculate forces from strain measurementscalculate forces from strain measurementsSignificant difference in total force on rods Significant difference in total force on rods and change in force in hangerand change in force in hanger

Difference was 1760 Difference was 1760 kNkN in one casein one case

Further analyses required to interpret the Further analyses required to interpret the measurement results measurement results

Page 19: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Force Discrepancy AnalysisForce Discrepancy Analysis

Verify all measurement data validVerify all measurement data validDefine normalized parameters for use in Define normalized parameters for use in further analysis (Structural Identification)further analysis (Structural Identification)

Strain Ratio (SR): Absolute value of the ratio Strain Ratio (SR): Absolute value of the ratio of the average rod strains to the average of the average rod strains to the average hanger strain. hanger strain. Force Ratio (FR): Absolute value of the ratio Force Ratio (FR): Absolute value of the ratio of the change in hanger axial force to the sum of the change in hanger axial force to the sum of the rod forces of the rod forces

Page 20: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

3D FEM Analyses3D FEM AnalysesComponents added 3D Components added 3D FEM of bridgeFEM of bridge

Rod postRod post--tensioning tensioning simulated with simulated with temperature loads temperature loads

Temperature loads Temperature loads modified until model rod modified until model rod and hanger strains and hanger strains converged to measured converged to measured strainsstrains

3D FEM of Hanger Region

Page 21: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

ResultsResultsSR and FR from analysis were correlated with SR and FR from analysis were correlated with experimental counterpartsexperimental counterparts

Analytical SR & measured SR correlate closely when Analytical SR & measured SR correlate closely when modular ratio of the hanger steel to the rod steel is 1.30 modular ratio of the hanger steel to the rod steel is 1.30 (actual modular ratio not available (actual modular ratio not available –– nominal ratio ~ nominal ratio ~ 1.01)1.01)

Nominal and calibrated FEM give FR of 0.94 (6% of sum Nominal and calibrated FEM give FR of 0.94 (6% of sum of rod forces lost to 3D effects) if modular ratio is of rod forces lost to 3D effects) if modular ratio is assumed 1.30assumed 1.30

Uncertainty due to incomplete information for calculating Uncertainty due to incomplete information for calculating forces (material test results)forces (material test results)

Uncertainty only identified in trying to reconcile forces Uncertainty only identified in trying to reconcile forces from the rods and hangersfrom the rods and hangers

Page 22: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

ShortShort--Term Structural Load Testing: Term Structural Load Testing: Remaining Fatigue Life EvaluationRemaining Fatigue Life Evaluation

453 ft

1736.5 ft

453 ft 830.5 ftAnchor Span Anchor SpanMain Span

Brent Spence Bridge

Page 23: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Test StatisticsTest Statistics48 strain gages installed on 12 truss members48 strain gages installed on 12 truss members

Sampling rate of 30 Hz for continuous monitoringSampling rate of 30 Hz for continuous monitoring

Wireless network established for data acquisitionWireless network established for data acquisition

Controlled load testing conducted to support analysis and Controlled load testing conducted to support analysis and interpretation of datainterpretation of data

Monitor all installed strain gages for 311 hoursMonitor all installed strain gages for 311 hours

Stress range histograms developed from critical gages on Stress range histograms developed from critical gages on each member using each member using rainflowrainflow counting methodcounting method

Calculate effective stress range from stress range Calculate effective stress range from stress range histograms for each member to evaluate fatigue lifehistograms for each member to evaluate fatigue life

Page 24: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Installed Strain GagesInstalled Strain Gages

Strain Gages in Chord Strain Gages in Diagonal

w

dStrain Gages: 350 Ohm

Weldable Resistance Strain Gage

Weldable Strain Gage

Typ. Gage Layout

Page 25: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Data Acquisition Cabinet @ Lower Chord

Wireless Local Area Network for Wireless Local Area Network for Fatigue MonitoringFatigue Monitoring

Wireless Bridge

WB02

Server

YagiAntenna

Wireless Bridge

YagiAntenna

~ 1.5 MilesWireless Transmission

Path Burgess & Niple Office Cincinnati, OH

Workstation

Drexel UniversityPhiladelphia, PA

Internet

WB01

Control DA SystemCollect, Process, & Archive Data

Page 26: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

0 1 2 3 4 5 60

2000

4000

6000

8000

10000

12000

Cyc

le C

ount

Sress Range (ksi)

Stress Range Histogram - Member UT-L2L4

Stress Range Histogram Stress Range Histogram -- ExperimentExperiment

Data Range: May 04, 2004 to May 19, 2004Total Cycles = 27,504RMC Effective Stress = 0.697 ksi

Member L2L4Upstream Truss

Page 27: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Average Stress Influence Line for Member L4L6Truck 01 = Lane L1, Truck 02 = Lane L2

-0.40

-0.30

-0.20

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0 2 4 6 8 10 12 14 16 18 20 22 22p 20p 18p 16p 14p 12p

Trucks @ Panel Point No.

Stre

ss, k

si

UT_L4L6_FLANGE

UT_L4L6_WEB

DT_L4L6_FLANGE

DT_L4L6_WEB

Ohio Tower

Direction of Truck Travel(KY to OH)

Kentucky Tower

Gag

e Lo

catio

n

Processed Data: Two Truck Load CaseProcessed Data: Two Truck Load Case

Page 28: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Conclusions Conclusions -- Uncertainty in Fatigue Uncertainty in Fatigue Evaluation ExperimentEvaluation Experiment

Spurious spikes in data Spurious spikes in data –– mitigate through digital mitigate through digital signal processing during analysissignal processing during analysis

Establishment of threshold stress level for fatigue Establishment of threshold stress level for fatigue damage (mitigated through controlled load testing)damage (mitigated through controlled load testing)

Stress cycle counting method used Stress cycle counting method used –– influences data influences data postpost--processing requirementsprocessing requirements

Duration of monitoring Duration of monitoring –– statistical analyses to statistical analyses to establish stationary resultsestablish stationary results

Even with uncertainties Even with uncertainties –– experiment more reliable experiment more reliable than purely analytical approachthan purely analytical approach

Page 29: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Ambient Vibration Testing of the Ambient Vibration Testing of the Brooklyn BridgeBrooklyn Bridge

Page 30: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Scope and ObjectivesScope and Objectives

Ambient vibration testing a component of Ambient vibration testing a component of seismic evaluation and retrofit studyseismic evaluation and retrofit studyResults used for system identification to Results used for system identification to improve reliability of FE modelsimprove reliability of FE modelsFocus on towers, but span responses also Focus on towers, but span responses also measuredmeasuredIdentify frequencies, mode shapes and Identify frequencies, mode shapes and dampingdamping

Page 31: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Description of ExperimentDescription of Experiment

Fixed array of 43 accelerometers located Fixed array of 43 accelerometers located on towers and spanson towers and spans

Measure longitudinal, transverse and Measure longitudinal, transverse and torsionaltorsional vibrations of towers vibrations of towers

Measure vertical, transverse and Measure vertical, transverse and torsionaltorsionalvibrations of main span and side span vibrations of main span and side span adjacent to Brooklyn Toweradjacent to Brooklyn Tower

Page 32: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Description of ExperimentDescription of Experiment

Wind speed and direction measuredWind speed and direction measured

Ambient temperature measurementsAmbient temperature measurements

Dynamic range adjusted after observing actual Dynamic range adjusted after observing actual responses responses –– minimum range utilizedminimum range utilized

Vibration data sampled at multiple rates (100 Vibration data sampled at multiple rates (100 Hz, 200 Hz, but primarily 20 Hz and 40 Hz)Hz, 200 Hz, but primarily 20 Hz and 40 Hz)

Long data records Long data records –– hours to dayshours to days

Measurements recorded over a monthMeasurements recorded over a month

Page 33: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Tower Instrumentation SchemeTower Instrumentation Scheme

Brooklyn Tower Brooklyn Tower ElevationElevation

Level HLevel H

Level GLevel G

Level FLevel F

Level ELevel E

Level CLevel C

Level BLevel B

Level ALevel AManhattan Tower Manhattan Tower

ElevationElevation

Level DLevel D

272’

272’

-- 6”6”

153’

153’

-- 0”0”

119’

119’

-- 6”6”

Page 34: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Accelerometer LayoutAccelerometer LayoutIntermediate Tower Levels between Top and Deck LevelIntermediate Tower Levels between Top and Deck Level

Transverse AccelerometerTransverse Accelerometer

Longitudinal AccelerometerLongitudinal Accelerometer

T

L

TL TL TL

Page 35: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Accelerometer LayoutAccelerometer LayoutTower Levels Below DeckTower Levels Below Deck

Transverse AccelerometerTransverse Accelerometer

Longitudinal AccelerometerLongitudinal Accelerometer

T

L

TL L

Page 36: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Span Instrumentation SchemeSpan Instrumentation Scheme

V T V T V T

V T

V

V V V

Main SpanMain Span

240’240’

Bro

okly

n To

wer

Brooklyn Brooklyn Bound Bound

Traffic LanesTraffic Lanes

Manhattan Manhattan Bound Traffic Bound Traffic

LanesLanes

Partial PlanPartial Plan

Side SpanSide Span

185’185’393’393’454’454’

Page 37: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Accelerometer Layout for SpansAccelerometer Layout for Spans

RoadwayRoadway RoadwayRoadway

Pedestrian Pedestrian WalkwayWalkway

T

STIFFENING TRUSSESSTIFFENING TRUSSESCross SectionCross Section

VV

Out

er T

russ

Out

er T

russ

Inne

r Tru

ssIn

ner T

russ

Inne

r Tru

ssIn

ner T

russ

Out

er T

russ

Out

er T

russ

33’33’--0”0” 33’33’--0”0”16’16’--6”6”

Transverse AccelerometerTransverse Accelerometer

Longitudinal AccelerometerLongitudinal Accelerometer

T

L

Page 38: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Accelerometer InstallationAccelerometer Installation

Page 39: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Accelerometer InstallationAccelerometer Installation

Page 40: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Uncertainty in ExperimentUncertainty in Experiment

Spurious spikes in data Spurious spikes in data –– remove during remove during preprocessingpreprocessingQuality of ambient excitation:Quality of ambient excitation:

Low level excitationLow level excitationNonNon--stationary excitationstationary excitationAmbient excitation primarily from traffic on Ambient excitation primarily from traffic on spans spans –– transfer to tower only occurs through transfer to tower only occurs through connections with deck, main cables, and staysconnections with deck, main cables, and stays

Identification of critical tower modesIdentification of critical tower modesDamping estimatesDamping estimates

Page 41: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Comparison of Tower and Span Comparison of Tower and Span Ambient ResponsesAmbient Responses

0 2 4 6 8 10 12 14 16 18 20-0.01

-0.005

0

0.005

0.01Filtered Time Domain Data for Several Span & Brooklyn Tow er Top Sensors

Time (s)

Acce

lera

tion

(g)

2.5 3 3.5 4 4.5 5-0.01

-0.005

0

0.005

0.01Zoomed View of Filtered Time Domain Data for Several Span & Brooklyn Tow er Top Sensors

Time (s)

Acce

lera

tion

(g)

HNLN51VN59VS51THST

HNLN51VN59VS51THST

Span Accelerations

Tower Accelerations (Top)

Page 42: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Effect of Spurious SpikesEffect of Spurious Spikes

500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500

-5

0

5

x 10-4 Filtered Time Domain Data with Noise for Channel #13

Time (sec)

Acc

eler

atio

n (g

)

0 1 2 3 4 5 610-12

10-10

10-8

10-6

PSD for Channel #13

Frequency (Hz)

PS

D (g

2 /Hz)

With NoiseAvg of SegementsComposite of Valid SegmentsComposite of All Segments

TypTyp. Spikes. Spikes

Segments

Page 43: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

NonNon--Stationary ExcitationStationary ExcitationFrequency Domain Frequency Domain –– Tower Transverse AccelerationTower Transverse Acceleration

(22(22--NOVNOV--04 16:01 04 16:01 –– 20:00)20:00)

Duration = 4 HoursDuration = 4 Hours

0 1 2 3 4 5 610-14

10-12

10-10

10-8

10-6Brooklyn Tower Transverse - 16:01 to 20:00 - Bandpass 0.3 to 10.0 Hz

Frequency (Hz)

g2 /Hz

0 1 2 3 4 5 610-8

10-6

10-4

10-2

100

Frequency (Hz)

g2 /Hz

HSTGMTFSTESTCSTBSTFNTFMTGST

ANPSD-T

HH

GG

FF

EE

CC

BB

N

MS

Average Normalized PSD

1.587 Hz2.695 Hz 4.468 Hz, 4.531 Hz

4.639 Hz, 4.668 Hz5.000 Hz

Page 44: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

NonNon--Stationary ExcitationStationary Excitation

HH

GG

FF

EE

CC

BB

Time Domain Amplitude Time Domain Amplitude –– Tower Transverse AccelerationTower Transverse Acceleration(22(22--NOVNOV--04 17:01 04 17:01 –– 17:16)17:16)

3600 3800 4000 4200 4400-5

0

5x 10-4

Time (s)

Acc

eler

atio

n (g

) HST

3600 3800 4000 4200 4400-5

0

5x 10-4

Time (s)

Acc

eler

atio

n (g

) GMT

3600 3800 4000 4200 4400-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) FST

3600 3800 4000 4200 4400-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) EST

3600 3800 4000 4200 4400-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) CST

3600 3800 4000 4200 4400-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) BST

Duration = 15 minutesDuration = 15 minutes

Page 45: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

NonNon--Stationary ExcitationStationary ExcitationFrequency Domain Frequency Domain –– Tower Transverse AccelerationTower Transverse Acceleration

(22(22--NOVNOV--04 17:01 04 17:01 –– 17:16)17:16)HH

GG

FF

EE

CC

BB

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

HST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

GMT

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

FST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

EST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

CST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

BST

Page 46: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

NonNon--Stationary ExcitationStationary ExcitationTime Domain Amplitude Time Domain Amplitude –– Tower Transverse AccelerationTower Transverse Acceleration

(22(22--NOVNOV--04 18:31 04 18:31 –– 18:46)18:46)

9000 9200 9400 9600 9800-5

0

5x 10-4

Time (s)

Acc

eler

atio

n (g

) HST

9000 9200 9400 9600 9800-5

0

5x 10-4

Time (s)

Acc

eler

atio

n (g

) GMT

9000 9200 9400 9600 9800-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) FST

9000 9200 9400 9600 9800-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) EST

9000 9200 9400 9600 9800-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) CST

9000 9200 9400 9600 9800-5

0

5x 10

-4

Time (s)

Acc

eler

atio

n (g

) BST

HH

GG

FF

EE

CC

BB

Duration = 15 minutesDuration = 15 minutes

Page 47: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

NonNon--Stationary ExcitationStationary ExcitationFrequency Domain Frequency Domain –– Tower Transverse AccelerationTower Transverse Acceleration

(22(22--NOVNOV--04 18:31 04 18:31 –– 18:46)18:46)

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

HST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

GMT

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

FST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

EST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

CST

0 1 2 3 4 510-12

10-10

10-8

10-6

Frequency (Hz)

PS

D (g

2 /Hz)

BST

HH

GG

FF

EE

CC

BB

Page 48: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

BROOKLYN TOWER LONGITUDINAL MODE SHAPES

0

50

100

150

200

250

300

-1.5 -1 -0.5 0 0.5 1 1.5

Normalized Amplitude

Hei

ght A

bove

Bas

e (ft

)

0.8060.9721.1961.2941.3871.4501.5191.8071.8361.8701.8801.8992.0313.4963.5113.5213.5303.5353.6083.7063.7653.7943.8043.8133.8433.8624.6684.766

Preliminary Tower Longitudinal Mode Shapes from Preliminary Tower Longitudinal Mode Shapes from Peaks in Amplitude SpectraPeaks in Amplitude Spectra

Bold shapes have best coherence & largest peak in frequency spectra

Tower TopTower Top

Tower BaseTower Base

Page 49: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Brooklyn Tower Resonant Frequencies & Mode ShapesBrooklyn Tower Resonant Frequencies & Mode ShapesTower Mode 1Tower Mode 1

11stst Longitudinal ModeLongitudinal ModeTower Mode 4Tower Mode 4

22ndnd Longitudinal ModeLongitudinal Mode

Hei

ght

(ft)

0

50

100

150

200

250

300

0.0 0.2 0.4 0.6 0.8 1.0 1.2

PP (1.387Hz)

PTD&RD(1.386Hz)

PTD&CORR(1.381Hz)

Amplitude

Top

Base0

50

100

150

200

250

300

-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2

PP(3.765Hz)

PTD&RD(3.768Hz)

PTD&CORR(3.771Hz)

Amplitude

Top

Base

Page 50: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Brooklyn Tower Mode ShapesBrooklyn Tower Mode Shapes

0

50

100

150

200

250

300

0.0 0.4 0.8 1.20

50

100

150

200

250

300

0.0 0.4 0.8 1.2

PP(1.597Hz)

PTD&RD(1.588Hz)

PTD&CORR(1.588Hz)

0

50

100

150

200

250

300

0.0 0.4 0.8 1.2

Hei

ght

(ft)

Amplitude Amplitude Amplitude

North LegNorth Leg Middle LegMiddle Leg South LegSouth LegTop

Base

Tower Mode 2 Tower Mode 2 -- 11stst Lateral ModeLateral Mode

Page 51: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

0

50

100

150

200

250

300

-0.8 -0.4 0.0 0.4 0.8 1.2 -0.8 -0.4 0.0 0.4 0.8 1.2-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2

PP (4.668 Hz)

PTD&RD(4.669Hz)

PTD&CORR(4.667Hz)

North LegNorth Leg Middle LegMiddle Leg South LegSouth Leg

Hei

ght

(ft)

Amplitude Amplitude Amplitude

Top

Base

Brooklyn Tower Mode ShapesBrooklyn Tower Mode ShapesTower Mode 5 Tower Mode 5 –– 22ndnd Lateral ModeLateral Mode

Page 52: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

0

50

100

150

200

250

300

-1.4 -0.7 0.0 0.7 1.4

PP (2.705Hz)PTD&RD(2.717Hz)PTD&CORR(2.726Hz)

0

50

100

150

200

250

300

-1.2 -0.6 0.0 0.6 1.2 1.8

PP(4.766Hz)PTD&RD(4.766Hz)PTD&CORR(4.768Hz)

Brooklyn Tower Mode ShapesBrooklyn Tower Mode ShapesTower Mode 3Tower Mode 3

11stst TorsionalTorsional ModeModeTower Mode 6Tower Mode 6

22ndnd TorsionalTorsional ModeMode

Hei

ght

(ft)

Amplitude Amplitude

Top

Base

Top

Base

Page 53: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

ConclusionsConclusionsUncertainty due to data quality and errors (bias error, Uncertainty due to data quality and errors (bias error, spikes) spikes) –– corrected through digital signal processing and corrected through digital signal processing and manual removal of spikesmanual removal of spikes

Uncertain excitation due to transfer of traffic excitation Uncertain excitation due to transfer of traffic excitation through structural connections to the towersthrough structural connections to the towers

NonNon--stationary excitation stationary excitation –– sample for longer timesample for longer time

Tower oscillations coupled with span oscillations Tower oscillations coupled with span oscillations –– reflection reflection of motions between the two componentsof motions between the two components

Uncertainty related to identifying the tower modes from the Uncertainty related to identifying the tower modes from the coupled motions of the spans and towerscoupled motions of the spans and towers

Seismic evaluation requires identifying those modes that Seismic evaluation requires identifying those modes that determine the tower demands during earthquakedetermine the tower demands during earthquake

Page 54: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

ConclusionsConclusions

Dynamic testing and modal analysis of real structures Dynamic testing and modal analysis of real structures are often driven by real engineering objectivesare often driven by real engineering objectivesA process oriented approach to taking data, processing, A process oriented approach to taking data, processing, and identifying the modal properties may fail to satisfy and identifying the modal properties may fail to satisfy the real engineering objectivesthe real engineering objectivesThe physics of the problem must be considered during The physics of the problem must be considered during each stage of experiment and analysis so that we may each stage of experiment and analysis so that we may reach meaningful interpretations of the results reach meaningful interpretations of the results –– we we cannot simply present a large quantity of identified cannot simply present a large quantity of identified frequencies/mode shapes for the purpose of seismic frequencies/mode shapes for the purpose of seismic retrofittingretrofittingidentifying the subset of peaks in frequency spectra identifying the subset of peaks in frequency spectra associated with resonant motion of the tower is essential associated with resonant motion of the tower is essential and this represents a significant challenge in ambient and this represents a significant challenge in ambient vibration testingvibration testing

Page 55: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

SummarySummaryThe mechanisms of uncertainty in testing of actual The mechanisms of uncertainty in testing of actual structures are abundantstructures are abundantExplicit consideration and mitigation of uncertainty Explicit consideration and mitigation of uncertainty in test design, execution, processing, and in test design, execution, processing, and interpretation is critical for providing real interpretation is critical for providing real engineering benefits from the experimentengineering benefits from the experimentAlthough some uncertainty can be mitigated through Although some uncertainty can be mitigated through proper design and execution of a field experiment, proper design and execution of a field experiment, we may have to accept that some level of we may have to accept that some level of uncertainty will always remain uncertainty will always remain –– heuristics related to heuristics related to structure and test objectives an important structure and test objectives an important component of interpretation and decision making component of interpretation and decision making based on experimentbased on experiment

Page 56: Impacts and Mitigation of Uncertainty SHMII-2 · 2005-11-28 · Impacts and Mitigation of Uncertainty for Improving the Reliability of Field Measurements Kirk A. Grimmelsman & A

Thank YouThank You