Upload
aiyubi2
View
227
Download
0
Embed Size (px)
Citation preview
8/12/2019 Horta Torus Structure
1/25
Langley Research Center
Static and Dynamic Model Update of anInflatable/Rigidizable Torus Structure
(Work In Progress)
Lucas G. Horta
and Mercedes C. Reaves
Structural Dynamics Branch
NASA Langley Research Center
FEMCI Workshop 2005
May 5, 2005
8/12/2019 Horta Torus Structure
2/25
Langley Research Center
Background The TSU* Hexapod is a generic test bed
that incorporates design features of
structures that can be packed for launchand deployed in space
Inflatable/rigidizable structures can be
inflated to more than 12 times their
packaged size enabling new space mission
scenarios
However, new fabrication techniques also
bring new modeling challenges
*TSU= Tennessee State University
8/12/2019 Horta Torus Structure
3/25
Langley Research Center
Objective and MotivationObjective-To develop a mathematical
and computational procedure to
reconcile differences between finite
element models and data from staticand dynamic tests
Motivation- All missions envisioned
for inflatable and rigidizable systemswill require state of the art controls
and accurate analytical models. In
the Hexapod case, the initial model
missed many important modes
10 20 30 40 50 60 70 80 90 100 11010
-5
10-4
10-3
10-2
10-1
Freq. (Hz)
Mag
Scale Factor = 0.11241 Sensor No. 12 Shaker 2
Test
NASTRAN
10 20 30 40 50 60 70 80 90 100 110-150
-100
-50
0
50
100
Freq. (Hz)
Phase(deg)
Initial ModelShaker No. 2
8/12/2019 Horta Torus Structure
4/25
Langley Research Center
ApproachFEM developed using commercial tools
Static test results used to update stiffness values
Dynamic test results used to update mass distribution
Probabilist ic assessment used to compute most likely set
of parameters
Update parameters computed via nonlinear optimization(genetic and gradient based)
Computational tools based on MATLAB
System ID using the Eigensystem Realization AlgorithmIncremental update of components starting with torus
8/12/2019 Horta Torus Structure
5/25
Langley Research Center
Model Update Procedure
Develop
baseline
fini te element
model Select
parametersfor update
Define parameter
bounds & probability
distribution
Compute output
probability &sensitivity valuesReduce parameter set
using sensitivity and
probability resultsSolve
optimization for
updated
parameters &probability
8/12/2019 Horta Torus Structure
6/25
Langley Research Center
Computational Framework
NASTRANStatic
Dynamic
Bulk DataFile
Output DataFile
MATLABOptimization
Probability
Sensitivity
System ID
External Inputs
1- Parameter selection
and prob. distribution
2- Experimental data
8/12/2019 Horta Torus Structure
7/25
Langley Research Center
Torus Description and Nomenclature
Urethane
Joint (typ.)
Tube (typ.)
O.D. 0.0181 m
1.858 m Joint flanget = 0.00508 m
Inner joint
t = 0.0063 m
Outer joint
t = 0.0032 m
Tube (typ.)
t = 0.00042 m
8/12/2019 Horta Torus Structure
8/25
Langley Research Center
Main Sources of Modeling
Uncertainties
Fabrication irregularities due to materials and
fabrication methods (irregular cross-section,irregular geometry, etc)
Unknown stiffness of adhesive bonded joints
Composite material property uncertainties,
effective single ply and rule of mixture
approximations
8/12/2019 Horta Torus Structure
9/25
Langley Research Center
STATIC TEST RESULTS
8/12/2019 Horta Torus Structure
10/25
8/12/2019 Horta Torus Structure
11/25
Langley Research Center
Input Load and Sensor Location for Static
Tests
Support 1
Support 2Support 3
Loading
Point
1
23
4
5
6
7
8
9
10
11121314
1516
17
18
19
20
21
2223
24
8/12/2019 Horta Torus Structure
12/25
Langley Research Center
Static Test Parameters Selected
for Update1. Torus tube thickness
2. Urethane joint:
- Inner joint thickness
- Outer joint thickness- Joint flange thickness
3. Torus tube orthotropic material properties:
E1, E24. Urethane joint modulus of elasticity E
8/12/2019 Horta Torus Structure
13/25
Langley Research Center
Output Probability Statement
How probable is it to predict the measured
output?
Linear Torus Solution
Probability
Bands
w range
w
v
v range
Output No. 1
Output No. 2
Updated parameters
Sample Space 300 points
8/12/2019 Horta Torus Structure
14/25
Displacement Results: Test, Baseline, and
Updated Model
Measurement locations
Target 1
Z = -4.54 mm
Load =42.1 N
Support 1
Support 2
Support 3
Torus structure displacement under load
-6
-5
-4
-3
-2
-1
0
1
2
3
1 2 3 4 6 7 8 9 10 11 12 14 15 16 17 18 19 20 22 23 24
Measurement location
Baseline
Update
Update Gen
Average Test
VerticalDisplacement(mm)
Torus static test results
=1.1 =0.01
Support 1
Support 2Support 3
Loading
Point
12 3
45
6
7
8
910
11121314
1516
17
18
19
20
21
2223
24
8/12/2019 Horta Torus Structure
15/25
Langley Research Center
Static Update Summary
14.1-13.90.008730.011570.011680.00860.01016
Flange
thickness
16.920.00.005270.005070.007610.005070.00634Inner jointthickness
-11.019.60.003520.003790.00380.002530.00317
Outer joint
thickness
9.4130.00039120.0003760.0004960.00030240.000432
Tube
thickness
-6.103.21E+093.03E+093.33E+092.73E+093.03E+09Joint E
5.405.83E+106.16E+106.77E+105.5E+106.16E+10Tube E1,E2
%
Change
Genetic
%
Change
FMIN
Optimum
Genetic
Optimum
FMIN
Upper
Bound
Lower
BoundBaselineParameters
8/12/2019 Horta Torus Structure
16/25
8/12/2019 Horta Torus Structure
17/25
Langley Research Center
Dynamic Test Configuration
Force gageSpring/cable
suspension
Suspension 1
Suspension 2
Laser
Vibrometer
Shaker
Suspension 3
Retro-
reflective
Target (typ.)
8/12/2019 Horta Torus Structure
18/25
Langley Research Center
Shaker and Sensor Locations for Dynamic
Tests
Suspension 1
Shaker 1Z
Suspension 2Shaker 2ZC
CC
C
C
C
C
C
C
C
C
C
14
16
17
1
5
4
2
7
8
11
10
1315
18
3
6
9
12
Shaker4 XY
Suspension 3
8/12/2019 Horta Torus Structure
19/25
Langley Research Center
Sample Set of Identification Results
(Drive Point 1 & 2)
0 20 40 60 80 100 12010
-5
100
105 Predicted (- -) and Real FRFs for input No. 1 and output No. 11
Frequency (Hz)
Amp
litud
e
0 20 40 60 80 100 120-200
-100
0
100
200
Frequency (Hz)
Phase
(deg
)
0 20 40 60 80 100 12010
-4
10-2
10
0
102 Predicted (- -) and Real FRFs for input No. 2 and output No. 17
Frequency (Hz)
Amp
litude
0 20 40 60 80 100 120-200
-100
0
100
200
Frequency (Hz)
Phase
(deg
)
Shaker No. 1
Shaker No. 2
8/12/2019 Horta Torus Structure
20/25
8/12/2019 Horta Torus Structure
21/25
1279.81433.81194.8955.8Joint 7 density
990.761433.81194.8955.8Joint 8 density
1063.51433.81194.8955.8Joint 10 density
1150.51433.81194.8955.8Joint 11 density
1239.21433.81194.8955.8Joint 13 density
1433.81433.81194.8955.8Joint 14 density
1080.21433.81194.8955.8Joint 16 density
1231.41433.81194.8955.8Joint 17 density
10931433.81194.8955.8Joint 1 density980.591433.81194.8955.8Joint 2 density
996.081433.81194.8955.5Joint 4 density
1204.61433.81194.8955.5Joint 5 density
1770.42200.81834.01467.2Torus tube density
2400.43502.22918.52334.8Spring 3
3169.83502.22918.52334.8Spring 2
2752.93502.22918.52334.8Spring 1
Optimum ValueUpper BoundNominal ValueLower BoundParameter^
Dynamic Update Summary*
* list of parameters not based on statistical analysis
8/12/2019 Horta Torus Structure
22/25
8/12/2019 Horta Torus Structure
23/25
8/12/2019 Horta Torus Structure
24/25
Langley Research Center
Work To Be Done Establish repeatability of test data
Complete probabilistic assessment of dynamic FRF to verifyparameter selection
Compute updated parameters for dynamic case using non-gradient based optimizer
Verify computational accuracy with refined FEM
Implement error localization algorithm to examine potential FEM
problem areas Compute Modal Assurance Criterion using reduced mass matrix
Re-test with increased sensor count to improve spatialresolution for test and analysis modes
Refine system ID results to improve areas near FRF zeros
8/12/2019 Horta Torus Structure
25/25
Langley Research Center
Concluding Remarks
Computational procedure using MATLAB in place
Initial updates completed using static and dynamic tests
results
Updated solution for static case is in good agreement with test
Selected parameters for static analysis showed low output
probability values when used to predict observed solution
Two step approach using output probabilit ies fol lowed by
optimization provides good physical insight into the problem
Dynamic test and analysis results need to be re-visited
Procedure computationally intensive but well suited for multi-
processor systems