Simple, Robust Hybrid Test
Systems for Non-linear Structural
Dynamic Research and
Development
K.J. Mulligan, J.G. Chase, R.B.
Elliott, B. Horn, and G. Danton
The Scene
• You are designing a new type of structural system,
connection or device
• You’ve modeled it to death – it looks good!
• However, it’s really expensive
– You’d hate to get it wrong
– And your colleagues are hogging the shake table
• How do you maximise results and minimise risk?
You Could.....
• Rely on even more modelling
Or
• Do a few full-size or scaled tests ($$$)
OR........
Use Hybrid Testing
Basic Elements:
• Virtual-real real-time interface control
• Main structure model that captures all
fundamental dynamics
• A test rig or actuators to supply the motion
required/dictated
• Sensors to measure response of sub-structure
• A test device and/or sub-structure
Why Hybrid?
• An intermediate step between models and
expensive, time-consuming large-scale tests
– Where you often forget to measure something that turns
out to be important anyway.
• Can potentially run many hybrid tests versus fewer
large-scale, especially examining devices
• Increases confidence before the next design and
test step and allows a chance to iterate designs or
models before taking a “bigger leap”
• Thus, it improves overall design cycle times
Where?
How? Processing Loop
External inputs
Structural model
calculations
Returned Signals
Conversion factors
Virtual System
Response of substructure measured
Test rig
implements
commands
Commands
Physical System
Real-time interface and data
exchange
Specifications
dSpace™ real-time operating system
• 1 to 10 kHz operating rate – much faster than
what you are testing
• Calculations done in corresponding time step guaranteed
• Incorporates virtual-real interface, as well as
data gathering and storage
Advantages
• Easy set up
• Readily available and widely accepted systems
and equipment
• Real-time means no additional calculations
required
• Highly flexible
• More systematic testing = greater certainty in
final application
Limitations and Solutions
• Signal processing lag
Returned
(measured)
signal
-3
2.5
displacement (m)
use returned signals
3
x 10
2
1.5
Command
signal
1
• Optimising sensor resolution and
bandwidth
0.5
0
16.6
16.8
17
17.2
time (sec)
17.4
17.6
equipment designed for type of application,
transparent control
• Efficient model computation
trade off between computational
complexity and efficiency
Rocking Wall case study
device being examined is incorporated into a ‘smart’ tendon to control rocking dynamics
equation of motion
Detailed how it works
Physical system
test rig
Virtual System
main structure
Displacement command
Valve Control
sub-structure
Measured Force and
Displacement
Results
ground motion
motion
2
(m/s )
2
(m/s )
2
0
4
theta
(rad)
ground
-2 2
10
0.02
20
25
30
35
40
45
50
0
0
-2
25
-0.02
10
30
35
40
45
15
20
25
30
35
40
45
50
15
20
25
30
35
40
45
50
0.01
0.02
0
(rad)
command
displacement
theta
(m)
15
0
-0.01
10
-0.02
returned
force
(N)
2000
1000
-0.04
25
0
-1000
10
30
15
20
35
time (sec)
25
30
tiem (sec)
time
(sec)
40
35
40
45
45
50
Conclusions
• Simple robust system that is easy to set up
• Runs in real time very fast to avoid iterating
• Large number of tests can be done quickly to
improve confidence
• Utilises widely accepted tools and equipment
Acknowledgments
• EQC Research Foundation Grant #03/497
• All co-authors and contributors