1 real-time hybrid simulations p. benson shing university of california, san diego
TRANSCRIPT
1
Real-Time Hybrid Simulations
P. Benson ShingUniversity of California, San Diego
2
Better Known as the Pseudodynamic Test Method
Early Work:Hakuno et al. (1969)Takanashi et al. (1974)
Institute of Industrial Science, University of Tokyo
Hybrid: real-time testing; analytical substructuring; distributed testing and simulation; ……….
Pseudodynamic: slow rate of loading; dynamic properties simulated numerically
3
Pseudodynamic Test Method
Simple concept but requires care to execute. Precision of displacement control. Accumulation of experimental errors in numerical computation.
Advance to nexttime step: i = i + 1
Update and
Numerical solutionof eqs. of motion
-i i giirMa Cv Ma
ir
gia
id
ir
Test FrameDisplacement
Experimental Error Accumulation
4
Main source of systematic experimental errors:time-delay in servo-hydraulic loading apparatus
Shing and Mahin (1982)
5
Dermitzakis and Mahin (1985)
Substructure Test Methods
Advance to nexttime step: i = i + 1
Update and
Numerical solutionof eqs. of motion
-i i i gi Ma Cv r Mba
gia ir
Computer Model
Test Frame
Range of Configurations
6
Slow Fast Real Time
Discontinuous/Continuous
Local Laboratories
Geographically Distributed
7
Needs for Real-Time TestsComputer Model Test
Base IsolationDevices
TestActive/PassiveDampers
Computer Model
General Framework for Hybrid Simulation
8
S Ma Cv r f
( ) ( )A A E EAS
ES M C v r rM a C f
Structural Partitioning
,
,
E ES S B
AA
B
EE
EB
S
0 f
f f f
f
r r
r
A AAA AB
A A A ABA BB BE
A AEB EE
M M 0
M M M M
0 M M
0 0
0E E E EBB BE
A
IEEB
B
E
EEE
M M M r
M
0
0 aM
a
a a
Total Formulation
9
EA A AS M a C v r r f
E E E ES r M a C v r
Whole Frame Analytical Model Physical Substructure
4AM
3AM
2AM
1AM 1
EM
2EM 2d 2
Er
1d 1r
2d
3d
4d
,2A
Sr
,3Sr
,4Sr
+ =
Coupled Subdomain Approach
10
12 2 3 3 2 ,2 2 2
3 3 3 3 4 4 3 ,3 3
4 4 4 4 4 ,4 4
0 0 0
0 0 0
0 0 0 0
A A A A A IS
A A A A AS
A A AS
M a C C v r f f
M a C C C C v r f
M a C C v r f
1 1 11 1 2 22
2 2 2 22 2
1
22
00
0 E
A A A A
E IA A A
a v fM C C C
a v f fM C C
r
r
Physical Substructure
1EM
2EM 2d 2
Er
1d 1r
22AM
1AM
2If
Magonette et al. (1998)
Analytical Model
4AM
3AM
12AM 2d
3d
4d
,2A
Sr
,3Sr
,4Sr
2If
Implicit Scheme
Explicit Scheme
Dynamic Substructuring I
11
2 3 2 ,2 2
3 3 3 4 4 3 ,3 3
4
2
2
4 4 4 24 ,4 4
0 0 0 0 0
ˆ ˆ0 0 0 0
ˆ ˆ0 0 0 0
A A tt
t
A IS
A A A A AS
A A AS
tA
M C v r f
M a C C C v r M
M a C C v r M
a
a
a
Analytical Model
4AM
3AM
2AM 2d
3d
4d
,2A
Sr
,3Sr
,4Sr
2If 2
ta
Physical Substructure
1EM
2EM
2If 2
ta
ga
Dynamic Substructuring I
12
Actuator
Actuator
Specimen
Shake Table
Computational Model
Sivaselvan and Reinhorn (2004)
Dynamic Substructuring II
13
,11 1 11 1 2 2
,22 2 222 2 2
00
0
AA A A AS gA AA A A
S gI
ra v M aM C C C
ra v M aC C fM
2AM
2d
1d
,2A
Sr
,1Sr
2If
Physical Substructure
3EM
2EM
1AM
ga
4EM
2td 2
If
Dynamic Substructuring II
14
Actuator
Shake Table
Computational Model
SD
Smart-UPS
6 2 0
www.apcc.com
Actual Equipment Tested
ga
Horiuchi et al. (2000)Bayer et al. (2005)Bursi et al. (2008)
15
Nakashima et al. (1992, 1999) Horiuchi et al. (1996) Tsai et al. Darby et al. (1999) Magonette et al. (1998) Bayer et al. (2000) Shing et al. (2002) Wu et al. (2005, 2006)
Real-Time Hybrid Test Methods
Explicit IntegrationSchemes
Implicit IntegrationSchemes
Implicit-Explicit CoupledField Analysis
16
Newmark Implicit Method for Time Integration
1 1 , 1 1 1A A A E
i i S i i i M a C v r r f
21 1 1i i it d d a
1 1 1i i it v v a
17
* ( ) ( )1 1
k ki i K d R
1
( 1) ( ) ( )1 1i
k k ki i
d d d
Modified Newton Method
*0 02
A E A E
A Et
t
M M C CK K K
( ) ( ) ( ) ( )1 1 1 1 , 1 1 1
k k A A k E ki i i i S i i i R M d d C v r r f
18
Modified Newton Method
Number of iterations varies from time step to time step. Increment size decreases as solution converges.
Convergence is guaranteed as long as
is positive definite (Shing and Vannan 1991).
Problems for Real-Time Tests:
2
A E A E
A ES S
t
t
M M C CK K K
19
Fixed Number of Iterations with InterpolationShing et al. (2002)
Response Correction and Update
20
11 1 , 1
, 1
1 1 ( )
( ) 1
A A A Ai i i S i
AS i i i
Ei
Ei
M a C v C v
r
r
f
r
r f
( ) ( )1 1
k d ki i d d
( ) * ( ) ( )1 1 1 1
E m k E d k m ki i i i r r K d d
-Method
Compatibility
Equilibrium
Nonlinear Structure
21
0 5 10 15 20 25 30-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
Time (sec)
Dis
pla
cem
en
t (in
ch)
CDM t = 0.001
RTI t = 0.01, N =10
RTI t = 0.01, N =50
-1 -0.5 0 0.5 1 1.5 2 2.5 3-30
-20
-10
0
10
20
30
Disp. (inch)
Fo
rce
(ki
ps)
5.42 5.43 5.44-900
-800
-700
-600
-500
-400
-300
-200
-100
0
Time (sec)
Re
sid
ua
l Fo
rce
Err
or
(kip
s)
CDM t = 0.001
RTI t = 0.01, N =10
RTI t = 0.01, N =50
RTI t = 0.01, N =10
RTI t = 0.01, N =50
22
System Configuration
NEES@Colorado
23
Real-Time Substructure Test Platform
dEr
PID Controller
Real-Time Processor
SCRAMNet Card 2
Analytical Substructure Model
ExperimentalElement/Substructure
Target PC –Real-Time Kernel
SCRAMNet Card 1
Special Element
Data-AcquisitionProgram
Actuators
Specimen
OpenSEES
24
Issues in a Real-Time Test
Actuator time-lag caused by dynamics of servo-hydraulic system and test structure.
Communication delays among processors.
Accounting for real inertia and damping forces.
Convergence errors in numerical scheme.
Interaction of numerical computation with system dynamics.
25
Phase-Lag Compensation Methods
( )( ) ( ) ( ) ( ) ( )c
PID ff p i d ff p
de t di t i t k e t k e t dt k k d t
dt dt
( ) ( ) ( 1) ( 1)1 1
c k k c n m ni i DFC i id d k d d
PID with Feedforward
Discrete Feedfordward Correction
Phase-Lead Compensator
d
d d
1( )
1
T sPLC s
T s
System Model for Test Simulation
26
force
Physical test system
Responses update
Iteration with interpolation
Displacement and restoring force,
2se
Delay(2 ms)
1ˆ
id
f
,
,i i
i i
d v
a r
1ˆ
id
m m1 1,i id r
ECd1id
c1id
cp, 1id
m m1 1,i id r
1ˆ
id
1 1
1 1
,
,i i
i i
d v
a r
Error compensation
Explicit part
Converged responses,
m m1 1,i id r
1 1 1 1, , ,i i i id v a r ( Outer loop)
( Inner loop )
Physical Test System
27
Controller
e PID
Servo-valve and hydraulic actuator
md
-+
displacement response
cpd
ppk A
ffk
Δ pressure feedback
Feed-forward loop
+
+i
d
dt
Test specimen
mr
-
se
Delay
mr
Physical testing system
28
System Transfer Function (Linear System)
Consider dynamics of servo-hydraulic actuators and test structure.
Communication delays.
Error compensation schemes.
Interaction of numerical computation with physical system.
Jung and Shing (2006)
29
Implicit Integration Scheme
1d F EP ICid d d d
( )( ) ( )F EF sd s f s
( )( ) ( )EP mEP sd s d s
( )( ) ( )IC mIC sd s d s
External Force
Explicit Prediction
Implicit Correction
30
2
( ) ( ) ( ) ( )( )
( ) ( ) ( ) ( ) ( ) ( ) ( )
m
s
d s EF s EC s P sRTPD s
f s e IC s EC s P s EP s EC s P s
System Block Diagram and Transfer Function
31
Physical Test System
32
Validation with Simulink Model
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 21
2
3
4
5
6
7
8
9
Frequency (Hz)
Am
p. F
act
or
D
kp = 1
kp = 5
Transfer Function, K = Kt
Simulink Model, K = Kt
Transfer Function, K = Kt * 1.1
Simulink Model, K = Kt * 1.1
( 1) ( 1) ( 1)1 1 1 1
m n d n m ni i ini i i
r r K d d
Error Correction:
33
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 21
2
3
4
5
6
7
8
9
Frequency (Hz)
Am
p. F
act
or
D
70 75 80 85 90 95 100 105 110 115 1200
0.01
0.02
0.03
0.04
0.05
Frequency (Hz)
Am
p. F
act
or
D
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-50
0
50
100
150
200
250
Frequency (Hz)
Ph
ase
An
gle
(d
eg
ree
s)
Analytical Sol.
kp = 1
kp = 3
kp = 5
kp = 7
kp = 7, kp
= -0.0002
Analytical Sol.
kp = 1
kp = 3
kp = 5
kp = 7
kp = 7, kp
= -0.0002
Analytical Sol.
kp = 1
kp = 3
kp = 5
kp = 7
kp = 7, kp
= -0.0002
System Performance (PID Only)
34
PID with Feedforward
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 21
2
3
4
5
6
7
8
9
Frequency (Hz)
Am
p. F
act
or
D
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50
50
100
150
200
250
Frequency (Hz)
Ph
ase
An
gle
(d
eg
ree
s)
Analytical Sol.
kff = 0
kff = 0.04
kff = 0, k
p = 5
Analytical Sol.
kff = 0
kff = 0.04
kff = 0, k
p = 5
35
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 21
2
3
4
5
6
7
8
9
10
Frequency (Hz)
Am
p. F
acto
r D
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50
50
100
150
200
250
Frequency (Hz)
Pha
se A
ngle
(deg
rees
)Analytical Sol.k
DFC = 0
kDFC
= 0.5
kDFC
= 1
kDFC
= 0, kP = 5
Analytical Sol.k
DFC = 0
kDFC
= 0.5
kDFC
= 1
kDFC
= 0, kP = 5
Discrete Feedforward Correction (DFC)
36
Inertia Effect in Real-Time Tests
Advance to nexttime step: i = i + 1
Update and
Numerical solutionof eqs. of motion
-A Ai i giirM a C v Ma
,S ir
gia
id
ir
Test Frame
tM a+
37
0 0.5 1 1.5 2 2.5 30
1
2
3
4
5
6
7
8
9
Frequency (Hz)
Am
p.
Fa
cto
r D
0 0.5 1 1.5 2 2.5 30
20
40
60
80
100
120
140
160
180
200
Frequency (Hz)
Ph
as
e A
ng
le (
de
gre
es
)
Analytical Sol.
Mt = M * 0.1
Mt = M * 0.4
Mt = M * 0.7
Mt = M * 1.0
Analytical Sol.
Mt = M * 0.1
Mt = M * 0.4
Mt = M * 0.7
Mt = M * 1.0
Influence of Inertia Force Feedback
38
Actual Test with Inertia Force Removal
0 5 10 15 20 25 30-1
-0.5
0
0.5
1
1.5
Time (sec)
Dis
pla
ce
me
nt (in
ch
)
CDM t = 0.001 Without Inertia Force Correction With Inertia Force Correction
1.5 2 2.5 3 3.5 4-1
-0.5
0
0.5
1
1.5
Time (sec)
Dis
plac
emen
t (in
ch)
CDM t = 0.001 Without Inertia Force Correction With Inertia Force Correction
Mt/M = 4.7%
Influence of Support Flexibility
39
0.5 1 1.5 2 2.50
1
2
3
4
5
6
7
8
9
Frequency (Hz)
Amp.
Fac
tor
D
Mr=M
t* 8, K
r=K
t* 8, f
r=f
t * 1
Mr=M
t* 8, K
r=K
t* 32, f
r=f
t * 2
Mr=M
t* 8, K
r=K
t* 128, f
r=f
t * 4
Mr=M
t* 32, K
r=K
t* 128, f
r=f
t * 2
Rigid Support
40
Nonlinear Structures (2-DOF, Method)
2(1 )
t
s
MK KConvergence: has to be positive definite
0 5 10 15 20 25 30-0.6
-0.3
0
0.3
0.6
Time (sec)
Dis
pla
cem
en
t (in
ch)
CDM t = 0.001PIDPID + DFCPID + PLCPID + FFRTI w/o Actuator
0 5 10 15 20 25 30-1.2
-0.9
-0.6
-0.3
0
0.3
0.6
0.9
1.2
Time (sec)
Dis
pla
cem
en
t (in
ch)
CDM t = 0.001PIDPID + DFCPID + PLCPID + FFRTI w/o Actuator
Strain Hardening
Strain Softening
Simulation Setup
41
Controller PC
Digital controller andData-acquisition System
Servo-hydraulic actuators and test structure
Rae-Young jung
Rae-Young jung
Rae
-Yo
ung
jung
Rae-Y
oung
jun
g
cd
,m md ri
SCRAMNet
Valve command, Sensors feedback
,m md r
Host-target pair 1
Real-timeTarget PC
Host PC
Host-target pair 2
Real-timeTarget PC
Host PC
SCRAMNet
Switch
Two-Story Frame
42
Two-DOF Real-Time Tests
43
Two-DOF Real-Time Tests
44
0 5 10 15 20 25 30-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Time (sec)
Dis
pla
cem
en
t (in
ch)
CDM t = 0.001 PID PID + DFC PID + PLC PID + FF
3.5 4 4.5 5 5.5-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Time (sec)
Dis
pla
ce
me
nt (in
ch
)
CDM t = 0.001 PID PID + DFC PID + PLC PID + FF
45
Real-Time Substructure Test with a Single Column
Actuator
Analytical Model in OPENSEES
Test Column
46
Real-Time Substructure Test
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
0 2 4 6 8 10 12 14
Time (sec)
Dis
pla
ce
me
nt
(in
)
FHT Test Pure Simulation
47
Test of a Zipper Frame
Georgia TechU. At BuffaloUC-BerkeleyUC-San Diego/U. of ColoradoFlorida A&M
48
Test Setup
49
Test Results
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0 1 2 3 4 5 6 7 8 9
Time [sec]
Dis
plac
emen
t [in
]Analysis
Hybrid Test
-2.5
-1.5
-0.5
0.5
1.5
2.5
3.5
0 1 2 3 4 5 6 7 8 9
Time [sec]
Dis
plac
emen
t [in
]
Analysis
Hybrid TestBrace
Rupture
80% LA 22
200% LA 22
50
Brace Response
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2-40
-30
-20
-10
0
10
20
30
40
50
60
Axial Displacement [in]
Axi
al F
orc
e [k
ips]
Hybrid Test
Numerical Model
Brace Rupture
(a) Brace 1
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5-50
-40
-30
-20
-10
0
10
20
30
40
50
60
Axial Displacement [in]
(b) Brace 2
Brace Rupture
51
Brace Damage
52
Future Challenge - Improve Computational Speed
* ( ) ( ) 1 1
1 1
ˆns ns
s k s kee e n e n
s sA A
K d R
*2ˆeeK1( )
1k
e nR 2( ) 1k
e nR*1ˆeeK *ˆ ns
eeK ( ) 1
ns ke nR
Parallel Computing
53
Future Challenge - Develop Mixed Control Strategy
Displ. Control
Computer ModelTest Specimen
Force Control
ShearWall
54
Dr. Rae-Young Jung, Former Grad. Student at CU
Dr. Zhong Wei, Former Grad. Student at CU
Dr. Eric Stauffer, Formerly at NEES@Colorado
Andreas Stavridis, Grad. Student at UCSD
Rob Wallen, NEES@Coloarda
Thomas Bowen, NEES@Colorado
Contributors
Development supported by NSF under NEES Program.
Acknowledgments
55
Thank You