Transcript Hybrid Testing: Simulating Dynamic Structures in the Laboratory

UNIVERSITY OF OXFORD
DEPARTMENT OF ENGINEERING SCIENCE
Hybrid Testing
Simulating Dynamic Structures in the Laboratory
Tony Blakeborough and Martin Williams
SECED Evening Meeting
28 January 2009
Outline
 Introduction

Dynamic test methods – why do we need new ones?
 The real-time hybrid method

Displacement-controlled tests

Testing strategy and equipment

Numerical integration schemes

Compensation for transfer system dynamics
 Recent developments and applications

Tests under force control

Crowd-structure interaction

Distributed hybrid testing in the UK-NEES project
 Conclusions
Acknowledgements
 Numerous colleagues contributed to the work described here,
particularly:

Current researchers: Mobin Ojaghi, Ignacio Lamata

Past researchers: Antony Darby, Paul Bonnet, Kashif Saleem, Javier
Parra

Collaborators at Bristol, Cambridge, Berkeley, JRC Ispra
 We have received financial support from:

EPSRC

The Leverhulme Trust

The European Commission

Royal Academy of Engineering

Instron
Testing methods in earthquake
engineering
 Shaking tables – apply prescribed base motion to models

Can accurately reproduce earthquake input

Normally limited to small-scale models – expensive at large scale

Scaling problems (physical and time)

Control problems
SUNY Buffalo
Bristol University
Testing methods (cont.)
 Pseudo-dynamic test facilities:

Slow test, with inertia and damping components modelled
numerically, stiffness forces fed back from test specimen

Can be conducted at large scale

Best suited to flexible structures with concentrated masses

Expanded timescale can’t capture rate effects

Feedback loop can cause errors to accumulate
Lehigh University
JRC Ispra
Future trends
 Major upgrading initiatives, e.g. NEES (USA),
E-Defense (Japan)
 Very large shaking tables
 Enhancements to pseudo-dynamic methods:

Effective force testing

Real-time hybrid testing

Distributed hybrid testing
Minnesota EFT
facility
San Diego outdoor
shaking table
E-Defense, Japan




1200 tonne payload
amax = 1.5 g, vmax = 2 m/s, umax = 1 m
24 x 450 tonne actuators
15,000 l/min oil flow rates
Real-time hybrid testing
Emulated
system:
Numerical
substructure:
Physical
substructure:
Displacements
Dissipator
Forces fed
back from
physical
substructure
Computed
displacements
FD(t)
FL(t)
Displacement
applied by
actuator
d1(t) – d0(t)
d1(t)
FR(t)
d0(t)
Forces
Ground displacement
dg(t)
dg(t)
Real-time hybrid testing
 Advantages:

Avoids physical scaling problems

Avoids time scaling problems

Ideal for testing rate-dependent systems

Economical – only the key parts need to be modelled physically

Now being strongly pursued by NSF NEES programme
 Needs:

High-performance hardware and communications

Fast solution of numerical substructure

Compensation of transfer system dynamics
Typical test set-up
Actuator 2
Actuator 1
Command
Command
Feedback
Feedback
Monitoring
PC
Real-time
PC
Feedback
dSpace
board
GPIB interface
Command
Proprietary
controller
Structural Dynamics Lab
Structural Dynamics Lab @ Oxford
Hydraulic installation
The Flight Deck
Typical real-time control loop
Transfer system
Input
load
Numerical ddes
substructure
Outer-loop
compensation
dcom
Inner-loop
(proprietary)
controller
Servohydraulic
actuator
Displacement feedback
Force feedback
 Dual time-stepping implementation:

Numerical model runs at main steps ~ 10 ms

Controller runs at sub-steps ~ 0.2 ms
 Imperfect transfer system dynamics cause:

Errors in timing and amplitude of applied loads

Inaccuracy and/or instability of test
dact
Physical
substructure
Typical test strategy
1. Solve numerical substructure to give desired actuator
n 1
displacement at the next main step, d des
2. Curve fit to the current and the past few displacement points.
3. Use curve fit to extrapolate forward by a time equal to the
estimated actuator delay, to give the command displacement, d n 1
com
4. Use same curve fit to interpolate dcom values at sub-steps.
Send to the inner loop controller, together with the current
actuator position dact
5. Repeat step 4 at sub-steps, until the next main step.
Numerical integration schemes
 We require:

Very fast solution of numerical substructure (~10 ms)

Accuracy, stability, ability to model non-linear response
 Explicit integration (e.g. Newmark’s method)

All required data known at start of timestep

Quick, sufficiently accurate

Need short timestep for stability
 Implicit integration (e.g. constant average acceleration method)

Requires knowledge of states at end of timestep, therefore iteration
(or sub-step feedback)

Unconditionally stable
 Two-step methods (e.g. operator-splitting)

Explicit predictor step, implicit corrector
Test system
 Simple mass-spring system
 All springs in numerical model
have bi-linear properties
 Increase DOFs in numerical
model to test algorithms
Physical substructure
Base
motion
Numerical substructure – n-DOF
 10-DOF numerical
Explicit
Results
 Sine sweep input
through several
resonances
 5 ms main-step
 0.2 ms sub-step
Two-step methods
substructure
 Red = numerical
 Blue = hybrid test
Implicit
simulation
 In frequency domain
Explicit
Results
substructure
 Sine sweep input
through several
resonances
 5 ms main-step
Two-step methods
 10-DOF numerical
 0.2 ms sub-step
simulation
 Blue = hybrid test
Implicit
 Red = numerical
 50-DOF numerical
substructure
 Sine sweep input
Explicit
Results
through several
resonances
 25 ms main-step (15
 0.2 ms sub-step
 Implicit schemes
unable to compute
in real time
 Red = numerical
simulation
 Blue = hybrid test
Two-step methods
ms Newmark)
 50-DOF numerical
substructure
 Sine sweep input
Explicit
Results
through several
resonances
 25 ms main-step (15
 0.2 ms sub-step
 Implicit schemes
unable to compute
in real time
 Red = numerical
simulation
 Blue = hybrid test
Two-step methods
ms Newmark)
Actuator dynamics
 Both timing and amplitude errors exist, and may vary during test
 Delay of the order of 5 ms is unavoidable
 Delay has an effect similar to negative damping  instability
Compensation schemes
Two components:
 Forward prediction scheme

Aims to compensate for known or estimated errors through scaling
and extrapolation

Exact polynomial extrapolation

Least squares polynomial extrapolation

Linearly extrapolated acceleration

Laguerre extrapolator
 Delay estimation

Delay and amplitude error estimates are updated as test proceeds
Validation experiments – Test A
 Linear, 2DOF system, single actuator
Emulated system
Base
motion
m2
Numerical:
m2
m1
Physical:
Feedback
force
m1
Actuator
Test B
 Non-linear, 2DOF system, single actuator
Emulated system
Base
motion
m2
m1
Gap
non-linearity
Numerical:
m2
Physical:
F
m1
Test C
 Linear, 3DOF system, two actuators
 Asynchronous input motions, stiff coupling
Emulated system
g2
m3
m2
g1
m1
Numerical:
Numerical:
g2
m3
F1
F2
Physical:
m2
m1
g1
Effect of forward prediction
Test A, with fixed
delay estimate, exact
polynomial
extrapolation
Hybrid test
Analytical response
Synchronization
plots:
Comparison of forward prediction schemes
RMS errors (%) over a test with constant delay and amplitude error
estimates
Test A
No compensation
Test B
Test C
Test C
Act#1
Act#2
unstable
unstable
unstable
unstable
Exact extrapolation
1.8
1.5
2.9
2.5
Least squares extrapolator
1.9
2.0
-
-
Linear acceleration
1.9
1.6
3.4
2.7
Laguerre extrapolator
1.8
1.7
unstable
unstable
Delay updating results
Delay estimates produced by updating scheme in Test C:
Effect of delay updating
 RMS errors (%) over a test with with third order exact
extrapolation
 Tests A and B used 0.5 ms sub-steps
 Test C used 0.2 ms sub-steps
Test A
Test B
Test C
Test C
Act#1
Act#2
No update
1.8
1.5
2.9
2.5
With updating scheme
0.9
1.1
2.3
1.8
Developments and applications
 Tests under force control

Dorka and Jarret Damper
 Crowd-structure interaction

Grandstand simulation rig
 Distributed hybrid testing

Oxford-Bristol-Cambridge
EU NEFOREE project
comparison of testing methods
Single storey test building designed by
Prof Bursi at Trento
3m
Parallel tests on shaking table, reaction
wall and real time hybrid substructuring
8630kg mass
Two dissipative devices to be tested
- Dorka shear device and Jarret dampers
Natural frequency
Unbraced 2.6Hz
Braced 8.6Hz
2% damping
5% damping (Dorka)
3m
Seismic testing of dampers
NEFOREE – EU study
Shaking table set-up
(elevation)
Hybrid test of device
800
400
200
80
40
900
660
125
2000
170
200
40
400
40
200
710
2000
Dorka and Jarret devices
Dorka shear panel:
shear diaphragm in SHS hysteretic damping
Jarret dampers:
Non-linear visco-elastic
devices
Control problems





Two actuators – equal but opposite forces
Dorka cell - very stiff specimen
Significant rig/specimen interaction
LVDT noise 30mm rms produced significant forces
Not possible to run under displacement control
Solution
 Run test in force-control
 Two MCS controllers – one for magnitude and other
for force imbalance
 Displacement feedback into numerical model
Force control loop
Apply measured
displacements to
numerical substructure
External earthquake
loads
Physical substructure
Numerical substructure
Measure deformation of test
specimen
Calculate forces
at interface between physical
and numerical substructures
Command actuators to
apply forces to
physical substructure
Numerical substructure
8630kg mass
x
xg
Columns - kc
3m
Mass
m
Damper - 
Braces - kb

3m
x  2cb
 
 x   1

F
 1   
 
0  xg 
 b2   x  b2  u2
   
0   x  0
 F  0  m b2  u2
 
1
 x  0
  x   m
  
  x 
 u2
0
2
b

0   
 
0 xg 
Earthquake records
El Centro
Normalised acceleration
1
0.5
0
-0.5
-1
2
4
6
8
10
Time(s)
12
14
16
18
20
4
5
Time(s)
6
7
8
9
10
1
Normalised acceleration
Synthesised
EC8 record
0
0.5
0
-0.5
-1
0
1
2
3
Response of Dorka device (El Centro 0.2g)
8
6
4
Force (kN)
2
0
-2
-4
-6
-8
-10
-12
0
5
10
15
Time (s)
20
25
Detail - EC8 synthesised earthquake tests
10
Force demand
Measured force
Error
0.2g pga
Force (kN)
5
0
-5
-10
-15
3.1
3.2
3.3
3.4
3.5
Time (s)
3.6
3.7
60
Force demand
Measured force
Error
Force (kN)
40
1.2g pga
3.8
20
0
-20
-40
-60
3.2
3.3
3.4
3.5
Time (s)
3.6
3.7
3.8
3.9
Specimen hysteresis curves
15
50
40
10
5
Specimen force (kN)
Specimen force (kN)
30
0
-5
20
10
0
-10
-20
-10
-30
-15
-0.15
-0.1
0.1
0.05
0
-0.05
Specimen displacement (mm)
EC8 0.2g
0.15
0.2
-40
-0.6
-0.4
-0.2
0
0.2
0.4
Specimen displacement (mm)
EC8 0.6g
0.6
0.8
Large hysteresis loops
80
60
60
40
20
Specimen force (kN)
Specimen force (kN)
40
0
-20
20
0
-20
-40
-40
-60
-60
-1.5
-1
-0.5
0
0.5
1
Specimen displacement (mm)
EC8 0.9g
1.5
2
-80
-3
-2
-1
0
1
2
Specimen displacement (mm)
EC8 1.2g
3
4
Conclusions – Dorka device




Real time hybrid tests successful
Simulated behaviour in 8Hz frame with 5% damping
Stiff specimen required force feedback loop
Device robust enough for use
Jarret devices
Response to square wave input
Brace demand
Measured force
Error
8
0.15g
alternating sign
(square wave)
ground
acceleration of
period 2s
6
Force (kN)
4
2
0
-2
-4
1.8
Specimen force (kN)
Specimen force (kN)
2.2
Time (s)
10
10
5
0
-5
-10
-2
2
-1
0
Specimen displacement (mm)
1
5
0
-5
-10
-100
-50
0
50
Specimen velocity (mm/s)
100
2.4
2.6
2.8
Response of Jarret devices
5
Force (kN)
El Centro
record with a
pga of 0.2g
around the
peak at 3.3s
Force demand
Measured force
Error
0
-5
3
2
3.4
3.6
3.8
4
Time (s)
4.2
4.4
Force demand
Measured force
Error
1
Force (kN)
3.2
.... and at end
of record
0
-1
-2
21
21.5
22
Time (s)
22.5
23
4.6
4.8
Response of Jarret devices
Force demand
Measured force
Error
10
0
-10
-20
0
5
10
15
10
15
Time (s)
6
4
Displacement (mm)
Force &
displacement
response of to
the EC8 record
with a pga of
0.6g
Force (kN)
20
2
0
-2
-4
-6
0
5
Time (s)
Response of Jarret devices
20
15
10
Force (kN)
Force against
displacement
and velocity for
the EC8 record
with a pga of
0.6g
5
0
-5
-10
-10
0
-15
-200
-150
-100
-50
0
50
100
Velocity (mm/s)
150
200
10
Displacement (mm)
Response of Jarret devices
20
15
10
20
Force (kN)
5
0
15
-5
10
Force (kN)
-10
-15
-10
-5
0
-5
0
placement (mm)
5
-200
-10
5
10
-200
-150
-100
-50
0
Velocity (mm/s)
50
100
Velocity projection
150
200
0
-15
-6
-4
-2
0
2
Displacement (mm)
4
200
6 Velocity (mm/s)
Displacement projection
EC8 record with a pga of 0.6g
Conclusions – Jarret device
 Tests successfully completed
 Realistic tests at low velocities
 Problems at higher velocities due to extreme non-linear
response in velocity
 Student just starting work on this – possibly use
velocity feedback with improved displacement
measurements
Human-structure interaction in grandstands
EPSRC funded study
RA – Anthony Comer
Grandstand rig
 15-seater grandstand rig
 Standard design – typical rake
& seat distances
 Test crowd coordination
 Effect of grandstand movement
on coordination
 Simulate various natural
frequencies and mass ratios
Grandstand rig design
 Aluminium alloy fabricated
rakers and stretchers
 Light & stiff – lowest
internal natural frequency
>30Hz
 Air spring at each corner to
take out mean load
 Electro-mechanical
actuator at each corner to
control rig
 Load cell under each
spectator
Control problems
 Force feedback from load cells at actuators suffered large
levels of interference from e/m fields emitted by motors
 Filtering would introduce too much lag for stability
 Digital displacement feedback available from linear encoders
(resolution 3μm) immune from e-m interference
 Use force control with displacement feedback
Control strategy
 Three significant degrees of freedom

Heave (vertical displacement)

Roll

Pitch
 Feedforward

Measure loads applied by ‘spectators’

Resolve into resultant vertical load and roll & pitch moments

Apply equivalent forces at actuators to balance force resultants and
keep rig stationary
 Numerical model

Simulate vertical and rotational damped springs numerically to
control dynamics of grandstand

Apply a proportion of vertical resultant load to excite the rig
Response to 130kg male jumping
4500
Vertical force (N)
Pitch moment (Nm)
Roll moment (Nm)
4000
3500
Force resultant
3000
2500
2000
1500
Vertical response only
1000
500
0
-500
40
45
50
55
Time (s)
1.4
Vertical
Pitch
Roll
Equivalent displacement (mm)
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
40
45
50
Time (s)
55
Rotations successfully
tared off
Conclusions – grandstand simulation
 Controlled tests possible on grandstand with spectators
jumping and bobbing
 Can also be used to wobble seated and standing
spectators to assess the acceptability of motion (main
dynamic use in project)
 Can be used to simulate human-structure interaction
Split-site testing –
hybrid testing over the internet
 Numerical and physical
substructures at separate
locations
 Possibility of testing very
large components
 Possible only over the
internet
Network architecture
JANET internet route
Communication interruptions
achieved bristol (measured ox)
command oxford
sine 5mm command to bristol rig from oxford
6
4
displacement (mm)
 JANET delays
 ~10ms - OK
 Inconsistency
0
-2
-4
-6
causes problems
 Solution
0
1
2
3
4
5
time (s)
6
7
8
9
10
8
9
10
network "delay" sine 5mm command to bristol rig from oxford
0.35
Use UK-light –
a dedicated link
0.3
0.25
Delay (s)

2
0.2
0.15
0.1
0.05
0
0
1
2
3
4
5
time (s)
6
7
Oxford-Bristol test
Results of test on Monday
Achieved displacements
achieved bris(ox)floor 1
achieved oxford floor 2
achieved oxford floor 3
15
10
displacement mm
5
0
-5
-10
-15
0
5
10
15
time s
20
25
Limitations
 Physical substructure

Limits set by equipment

Response times of actuators

Control problems at limits of actuator capacity

Stiffness of frames
 Reduce uncertainty

Proof testing (strength/performance guarantee)

Check individual items

Assess design under realistic loading

Validate computer models used in design
Architecture of 3 site test – radial model
State of work in split site testing
 Ethernet not a problem provided use a dedicated link
 Tests possible and seem to work
 Future work

Increase natural frequencies of systems – currently “3Hz but up to 10
should be possible

Investigate different interconnection links

At moment there is a central numerical model with physical sites
as servers at end of radial spokes – other arrangements are
possible
 Investigate force control
 Extend to the rest of the world – planning links with EU in FP7
research
Conclusions
 Simulation of real time behaviour

It works for ‘stiffness’ and ‘rate dependent’ components

Reproduces rate/time dependent effects

Useful for more realistic component testing

Allows devices to be checked in much more arduous circumstances

Copes with non-linear behaviour in both physical and numerical
substructures
General conclusions
 What test at all?

Reduce uncertainty


Proof testing (strength/performance guarantee)
 Check individual items
 Assess design under more realistic loading
 Validate computer models used in design
Challenging activity

Push current control techniques and test equipment to limits
 Trickle down effect – improved techniques help standard testing