Hybrid Testing; Benchmarks and Validation

Post on 03-Jan-2016

26 views 0 download

description

Gary Haussmann Eric Stauffer. Hybrid Testing; Benchmarks and Validation. Outline. Validation: Test vs. Analytical Models Sources of error Results and comparison Capabilities, from PSD to Real-time When do you need real-time? The same test at multiple speeds Limitations Model size - PowerPoint PPT Presentation

transcript

CU-NEES 2008 FHT Workshop1

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Hybrid Testing;Benchmarks and Validation

Gary Haussmann

Eric Stauffer

CU-NEES 2008 FHT Workshop2

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Outline

• Validation: Test vs. Analytical Models– Sources of error– Results and comparison

• Capabilities, from PSD to Real-time– When do you need real-time?– The same test at multiple speeds

• Limitations– Model size– Velocity, displacement limits

CU-NEES 2008 FHT Workshop3

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Validation Process

• Well Understood Model

• Predict Analytical Response

• Compare With FHT Test Results

CU-NEES 2008 FHT Workshop4

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Validation: Error Sources• Actuator Lag

– For a stiff specimen, lag leads to negative “damping”

– For a damping specimen, lag leads to negative “mass”

• MR Damper Model– Nonlinear Response– Use a Linearized Model

• Measurement Error– Instrument Noise

CU-NEES 2008 FHT Workshop5

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Validation: MR Damper

• Perform Simple Tests to Characterize Damper– Variety of velocities and displacements

• Generate a Damper Model for Analytical Solution

CU-NEES 2008 FHT Workshop6

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Validation: Results

CU-NEES 2008 FHT Workshop7

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Validation: Results

• Large amplitude tests match analytical

• As amplitude decreases, lag increases in the test results

• Primary error source: MR Damper model– Linearized model doesn't match near-zero

response of the actual damper– Not noticeable with large signals

• Revise Damper Model

CU-NEES 2008 FHT Workshop8

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Capabilities: PSD

• Pseudodynamic testing

• Time is not a factor

• Local or Distributed Testing

• No Limits on Model Size

• Three Independent or Coordinated Actuators

• 12+ Measurement Channels

CU-NEES 2008 FHT Workshop9

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Capabilities: Real-Time• Soft or Hard Real-Time

• Limitations:– Local only– Max Model Size– Actuator Velocity Limits– Data Channel Limit

(~80) – Pausing “ill-advised”

CU-NEES 2008 FHT Workshop10

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time Testing?

• You don't always need real-time– Stiffness primary phenomenon– Low-frequency excitation

• Real-time Situations– Velocity-based components– Rate-dependent forces– Can we model it instead?

CU-NEES 2008 FHT Workshop11

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

When Real-Time Matters• A Simple Hybrid Test With MR Damper

• Try Various Real-Time Methods and Compare

CU-NEES 2008 FHT WorkshopNEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Various Modeling Schemes

• Hard Real-Time

• Soft Real-Time– “Fake” Soft Real-Time

• OpenFresco OS Soft Real-Time– Interpolate/Extrapolate– Ramp-and-Hold

• Scaled time: *10, *100

CU-NEES 2008 FHT Workshop13

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time vs. Stretched Time

• Velocity-based components– Impact is reduced– Slower time leads to less response

• Stiffness-based components still valid

• Virtual forces can emulate a velocity-based response if no real-time available– Compute velocity-based response in slow test– Requires reasonable damping model

CU-NEES 2008 FHT Workshop14

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time Test Results

• Slower Tests Reduce Damper Effect

• Hard To See The Difference

CU-NEES 2008 FHT Workshop15

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time Test Results

• Slower Tests Reduce Damper Effect

• Hard To See The Difference

CU-NEES 2008 FHT Workshop16

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time Results

• Exaggerate The Differences

• Assign Hard Real-Time = 100%

• Assign No Damper = 0%

• Scale The Rest Appropriately

CU-NEES 2008 FHT Workshop17

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time Results

CU-NEES 2008 FHT Workshop18

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time TestingAnalysis Limitations

• Standard update rate: 0.97mS period– 60Hz Monitor: 16mS

• Model Size– Degrees-of-freedom: 130– Ever-increasing

• Material Modeling– Complex materials take more time– Elastic Linear easiest

CU-NEES 2008 FHT Workshop19

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time TestingSpecimen Limitations

• Actuator Limitations– Displacement: +-6 inches– Velocity: 10/20 inches/sec– Force: 110 or 220 Kips– Use actuator model to predict these results

• Coupled and Multi-DOF test Configuration– Crosstalk and parasitic coupling– Exacerbated in real-time

CU-NEES 2008 FHT Workshop20

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Real-Time TestingBonus Capabilities

• MATLAB xPC/Real-time Workshop– Also hooked into SCRAMnet– Additional modeling/analysis during test

• PXI Instrument/DAQ Chassis– Analog and Digital I/O– Acquire quantities in real-time over SCRAMnet

• Streaming analysis data– Visualize OpenSEES model during test– Debugging, Entertainment

CU-NEES 2008 FHT Workshop21

NEES at CU Boulder

The George E Brown, Jr. Network for Earthquake Engineering Simulation

01000110 01001000 01010100

Feb. 19, 2008

Conclusion

• Hybrid Test Validation– Compare test and analytical solution– Damper model main source of discrepancy

• Real-time vs. slower time– Damping effects are reduced– Slower tests approach no-damping results

• Hybrid testing limitations– Model size– Actuator limits