MPLM, COLUMBUS and NODE2 Modal Survey Tests and

advertisement
MPLM, COLUMBUS and NODE2 Modal Survey Tests and
Mathematical Models Validation
Marina Bellini
Alenia Spazio S.p.A., Torino, Italy
Abstract
The validation of the mathematical models of the International Space Station
pressurised modules, such as Mini Pressurised Logistic Module (MPLM), the
European scientific laboratory Columbus and the interconnecting element Node2
was performed by Modal Survey Tests (MST). The first module to be tested was
MPLM. The paper wants to underline how MPLM experience was very useful, for
the similar projects, which followed. In fact during Columbus and Node2 MST, the
previous experience of MPLM has allowed obtaining more quickly a very high
quality of empirical data. This was essential for the successful correlation between
the experimental and the analytical data.
Introduction
Three MPLM, already in Space Station operative cycle since years, are used to
transport
supplies
and
materials
between Earth and
NODE2
the
International
Space Station (ISS)
using the Space
COLUMBUS
Shuttle. Columbus
and Node2 are next
to be launched as
MPLM
final steps towards
completion of the
International Space
Station
assembly
(Figure 1).
Figure 1: MPLM, COLUMBUS and NODE2
Node 2 represents
accommodation in ISS
the interconnection
between the US part of the ISS and the International Partners section. It will be
attached to the forward port of the USLab Destiny and will provide docking and
utilities accommodation of the Japanese modules and Columbus. The Columbus
Laboratory is the centrepiece of the European contribution to the ISS. It provides a
pressurised and habitable environment for crew activities and experiment execution,
as well as an external facility for accommodation of unpressurised payloads [5].
During the Space Shuttle lift-off and landing events and during the connection with
the ISS, loads are produced by the interaction between Orbiter and Payloads in the
Cargo Bay and by the interaction between Payloads and ISS. The results of the
Orbiter/Payloads and ISS/Payloads coupled loads analyses, used for the verification
of the pressurized module structure can be affected by the uncertainty of the module
dynamic model. The scope of the modal survey test, imposing a flight-like and onorbit like constraint configurations, is to provide information about the significant
natural frequencies and mode shapes up to 50 Hz, for the correlation and the
validation of the module dynamic mathematical models.
Test fixture
The Test Fixture used for MPLM, Columbus and Node2 Modal Survey Tests was
the same and it was designed and house-made by Alenia [7]. The test stand has
been designed to support payloads with a weight up to 8000 kg. This test fixture
(Figure 2) consists of four
independent pillars fixed to
the ground of the test hall. On
top of each pillar a housing
was mounted with specific
suspension
elements
designed to carry the
trunnions of the test article
with a high fixation stiffness
in the restraint degrees of
freedom and a low (residual)
stiffness in the free degrees
Figure 2: Test fixture for MPLM, Columbus
of freedom. In order to
and Node2 MST Finite Element model
achieve a good and reliable
test prediction, before MPLM
MST an extensive experimental verification of the analytical model of the test
fixture was performed. The suspension elements (flexures) were instrumented with
strain gauges, calibrated to measure axial loads (and then effective masses), and
their axial, lateral and torsional stiffness values were determined statically. In
addition to validate the axial and lateral stiffness of the suspension elements under
operating conditions, a modal survey test was performed on the test-fixture loaded
by a dummy mass. During MPLM, Columbus and Node2 MST, these boundary
conditions were checked and verified again by a set of equivalent measurement
channels.
Test configuration
The first step, during the phase of predictions is to select, on the basis of the
existent mathematical model, a set of modes, defined target modes. The target
modes are identified as the modes with the maximum modal contribution to the
payload structural “design driver” parameters. For MPLM, Columbus and Node2
the “design driver” parameters were identified, because on the main load path of the
module primary structure, on module net acceleration, on module / Shuttle interface
forces and on module/rack interface forces. Anyway the set of target modes has to
be able to provide all information necessary to describe the dynamic behaviour of
the structure. On the basis of the set of target modes the test hardware configuration
is defined. In fact, it has to be avoided to perform the test with the fully
configuration in launch configuration. The main reason is to reduce the modal
density due to the presence of all the secondary structures. A high modal density
would not allow a correct identification of the modes during the test and would
complicate the following activity of correlation and updating of the mathematical
model. A further reason to define a test configuration is due to the wide
reconfigurability of the internal payloads, that produces different possible flight
manifests. This is the case of MPLM, designed for several missions, with different
payload amount and distribution. The module MPLM has the capability to
accommodate 16 racks around the perimeter of the module and attached to eight
longerons. The 25 missions planned by MPLM program consider different racks in
terms of mass and stiffness in miscellaneous arrangements. Then the internal
arrangement of the racks was
defined, driven by the necessity
to have a test configuration,
that, also if different from any
flight manifest, had to be able
to provide the necessary
information for the qualification
of the MPLM primary structure
mathematical model [2]. The
test configuration had to
provide the maximum possible
number of the 23 selected target
modes and the modes so
correlated had to present a
similar
strain
energy
Figure 3: Columbus MST configuration
distribution. Between the fully
loaded configuration and the
test configuration a good Modal Assurance Criterion (MAC) had to be obtained as
well. Different test configurations, suggested by a detailed analysis of the target
mode shapes, were studied. Only by the configuration chosen a good correlation
was obtained for 10 target modes and the modal strain energy distribution,
calculated for the 11 groups related to the different structural items, was similar to
the distribution presented by the fully loaded configuration. This good result had to
be attributed to the capability of this configuration to reproduce the dynamic
behaviour of the fully loaded configuration, as a result of the presence of four racks
constrained to the central ring and of their contemporaneous presence on
ceiling/port side, and on floor/starboard side locations. The frequency differences
were not considered critical because to be attributed only to the reduced mass; the
final weight of test hardware was about 6250 Kg while the fully loaded
configuration weighted 14100 Kg.
Using the same approach, the test configurations for Columbus and Node2 were
defined.
For these projects no
different manifests of launch are
foreseen, but internally several
structures of different typology are
included. In this case the purpose of
reducing the high modal density was
reached, taking into account the
necessity to have, as minimum, in the
test configuration one of each
different typology of the secondary
structures. This allowed acquiring,
during the test, data related to their
interface with the primary structure.
Before to perform Columbus and
Figure 4: Node2 MST configuration
Node2 MST, each different typology
of secondary structures was separately tested. More precisely for Columbus 5 MST
and for Node2 6 MST were carried out. The acquisition of the empirical data for
these structures resulted sometimes difficult, due to high non-linearity of some
modes. On the basis of these tests, the models of the secondary structures were
validated. During their validation, it was chosen to generate models characterized
by the frequencies more critical for the module. The use of the validated models of
the secondary structures was helpful in the prediction and correlation activity of the
module mathematical model. Figure 3 reports a sketch of Columbus Test
Configuration (2 External Payload Facility, Port Panel –Z, Starboard Panel +Y,
Starboard Panel –Y, Sub-System rack D1, 2 overhead racks located in the middle
locations) and Figure 4 a sketch of Node2 Test Configuration (2 Midday, 2
Alcoves, 2 DDCU Racks, 2 ISPR racks).
On-orbit mathematical model validation
The modules are attached to ISS by the Common Berthing Mechanism (CBM),
which is connected to their forward end cone. During the development of the
MPLM there was no requirement for verification of the on-orbit interfaces with the
ISS [4]. When this oversight was discovered, all the dynamic test stands had
already been disassembled. A method was needed that would not require an
extensive testing stand and could be completed in a short amount of time. The
residual flexibility modal testing method was chosen [1][8]. Anyway due to the
global cost-effectiveness of this method, in particular as concerns the elaboration of
data, for Columbus the Mass Loading method was preferred [3]. This method is in
fact the natural extension of the flight configuration modal survey. After a proper
analytical assessment, a test was performed on the flight configuration set-up
(Figure 5) after mounting on the Passive CBM a properly designed dummy mass
(Figure 6), in order to emphasise the modes involving the Port Cone and enabling
correlation with those exhibited in the on-orbit restraints configuration. The
presence of the dummy mass resulted necessary, because the Flight Configuration
set-up emphasised some difficulties in detecting the modes of the Passive
CBM/Port Cone Area. The totally different restraint configuration implied the shift
of the modes of interest towards higher frequency ranges. These modes proved
moreover to be of straightforward identification only when the test article was
emptied of the secondary structures; the
flight test set-up required meanwhile the
presence of these, as a minimum, to
reach the goal of the test. On the other
hand, it was impossible to consider the
possibility of performing the test of the
module without the secondary structures,
due to schedule reasons. The dummy
mass helped to increase the amplitude of
the Port Cone area displacements with
respect to the remaining secondary
structures and lower the frequencies of
these modes in a frequency range
Figure 5: Columbus on orbit MST
configuration
suitable for the test.
Various combinations of the inertial
parameters of the dummy mass were
tested in order to find the optimum
solution fitting for the purposes of the
analysis. After the feasibility study, a
cylindrical mass of about 1000 kg with
centre of gravity at about 0.5 m from the
Passive CBM interface was produced. Its
first
hard-mounted
fundamental
frequency was estimated at 120.06 Hz,
and thus sufficiently higher than the
frequencies of the target modes selected
for the test.
Figure 7: Node2 MST- on
orbit radial configuration
Figure 6: Dummy mass finite
element model
Figure 8: Node 2 MST - on
orbit axial configuration
For Node2, the same approach was followed. After the acquisition of the modes in
the flight configuration, two dummy masses, exactly the same studied for
Columbus were applied to the radial CBM (Figure 7), and after to the axial CBM
(Figure 8), to acquire data to validate Node2 mathematical model, also as concerns
interface with the ISS. Figure 9 and Figure 10 show, for Node2, the strain energy
distribution when dummy mass is connected to the radial and the axial CBM. This
distribution is similar to that obtained when the real on–orbit constraints are
applied.
Figure 9: Node2 - MST on orbit
radial configuration -strain energy
distribution – tangential mode
Figure 10: Node2 - MST on orbit
axial configuration -strain energy
distribution – lateral mode
Test set-up and test performance
The method applied, during all these MST, was the classical modal tuning (phase
resonance). Based on the results of the test predictions, for MPLM, 20 excitation
points were defined. The basic set of structural measurements consisted of 109
measurement points on the primary structure and 26 measurement points on each of
the four installed internal racks,
each of them tri-axial. Together
with some additional measurement
chains, i.e. the accelerometers and
strain gauges to check the boundary
conditions at the trunnions and the
exciter forces and response
accelerations, a total number of 690
measurement channels had to be
acquired [6]. To satisfy this
requirement, IABG’s large modal
test facility was used for this
project. Some high non-linearities
were evidenced at the beginning of
Figure 11: Special tool for keel
the test due to the rack braces
“loose boundary” effect and to the gap between the keel trunnion and the fitting
body, present by design, in the magnitude of 0.04 mm. To avoid the first problem
the braces were preloaded, while the second problem, impacting only a reliable
acquisition of the first global Y mode, imposed the installation of a special tool for
the tuning of this mode (Figure 11). To check that the frequency measured using the
special tool was representative of the flight behaviour of MPLM, it was necessary
to excite the structure, without the special tool, at the expected flight acceleration.
With a new exciters set-up allowing higher input forces it was demonstrated that the
frequency of this mode increased asymptotically when the MPLM acceleration
responses were increased, keeping similar shapes (Figure 12). The same problems
arose during the tests of Columbus and Node2, performed with the Alenia mobile
Modal Test System. But thanks to MPLM experience the problems were quickly
solved.
During Columbus MST 20 exciters have been used, and 741 channels
acquired. Further 72 channels, distributed to have an effective representation of the
modal displacement of the area Port Cone/CBM were acquired during the test for
the on orbit mathematical model validation [3]. During Node2 MST 23 exciters
have been used, differently distributed, according to test configuration. For Node2
the number of channels acquired was 822 channels (flight condition), 759 (radial
configuration) and 819 (axial configuration).
Figure 12: Node2 MST-Y mode frequency versus force
Updating and correlation results
The correlation criteria, established before starting the prediction activity, were of
type quantitative and qualitative .The deviation in terms of natural frequency had
not to be greater than 5%. The correlation in terms of the mode shapes was
considered good for cross-orthogonality values equal or greater than 0.9 for the
diagonal values and less than 0.1 for the off-diagonal values for the target modes.
The qualitative criteria are based on the comparison of the mode shapes plots, the
modal assurance criterion (MAC), the effective and generalized masses differences.
For all the projects the correlation goals have been achieved, especially thanks to
the good quality of the empirical data and to the detailed activity of prediction. Due
to the different typology of the secondary structures, the activity of updating and
correlation was different for the various projects. The updating of the mathematical
models is resulted always an exacting and expensive activity. Also the updating
suggested by software dedicated to the optimizations is resulted scarcely helpful for
these complex structures. As example of the quality reached by the correlation, the
data related to Node2 are reported. The first two tables are related to the flight
configuration. Table 1 (related only to the target modes) and Table 2 report the
cross-orthogonality, calculated before and after the updating activity. Practically all
the first target modes respect the correlation criteria. The requirements of frequency
shift < 5% is achieved for all the target modes.
Table 1: Node2 MST- Flight configuration- preliminary cross-orthogonality
TEST
ANALYSIS FR [Hz.]
1
2
3
4
5
6
7
8
9
10
11
13
15
16
17
18
19
20
28
35
12,40
14,92
16,11
18,01
18,55
19,92
20,66
21,34
21,95
23,15
23,79
25,46
26,67
27,68
29,46
30,81
32,94
33,89
36,95
41,83
1
64
12,42
1,00
2
51
14,50
3
3
15,67
4
4
17,94
5
5
18,50
6
6
19,90
7
7
20,71
8
49
21,38
9
9
21,91
0,99
-0,18
-0,27
-0,93
10
10
24,31
11
62
24,58
0,94
0,14
0,87
12
45
25,67
13
52
27,51
14
15
28,39
0,18
-0,15
-0,91
0,19
-0,14
0,19
0,16
0,29
0,93
0,23 -0,74
15
17
29,56
16
43
31,71
17
18
33,41
18
53
34,24
24
63
38,02
-0,99
-0,99
-0,98
-0,99
0,98
0,99
0,97
0,12
0,95
0,22
-0,88
0,12 -0,73
0,11
0,14
0,64
Table 2: Node2 MST – Flight configuration – final cross-orthogonality
Table 3 and Table 4 are summary tables, related to the MST dedicated to acquire
data for the validation of the model in on-orbit restraints configurations (shaded
values are related to the target modes). The requirement of frequency is satisfied.
The Cross and the MAC values are not always good, because affected by the
behaviour of the secondary structures that above 25 Hz does not result perfectly
correlated. In the on orbit restraints conditions, anyway the areas of interest are the
radial and the axial ports. Consequently to show the quality of the correlation and
the compliance with the requirements, the piece-wise technique (using a sub-set of
degrees of freedom) was used in the MAC calculation.
Analytical
Frequency
Mode
[Hz]
mode
Frequency [Hz]
Error %
4
22.16
10
21.57
-2.66
0.87
0.75
0.87
0.83
1
23.33
12
22.73
-2.57
0.93
0.88
0.89
0.93
2
27.02
17
25.74
-4.74
0.55
0.28
0.66
0.80
12
41.44
34
39.85
-3.84
0.73
0.33
0.73
0.87
10
44.92
42
44.32
-1.34
0.69
0.20
0.46
0.76
8
46.46
---
---
---
---
---
---
---
6
48.01
48
46.01
-4.17
0.55
0.00
0.68
0.77
7
49.40
51
47.05
-4.76
0.56
0.00
0.72
0.82
Test Test Frequency Analytical
Cross
MAC
MAC
MAC
Orthogonality (759 DOFs) (378 DOFs)
(72 DOFs)
Table 3: Node2 MST- On orbit radial configuration – final correlation
Test Test Frequency Analytical
Analytical
Frequency
Cross
MAC
MAC
MAC
Mode
[Hz]
mode
Frequency [Hz]
Error %
1
29.20
18
28.36
-2.88
0.87
0.73
0.75
0.82
2
31.88
20
31.06
-2.58
0.93
0.86
0.92
0.97
3
33.72
21
32.79
-2.75
0.96
0.86
0.96
0.96
12
38.74
30
37.39
-3.50
0.67
0.15
0.80
0.88
7
39.74
32
38.79
-2.40
0.87
0.42
0.80
0.92
5
40.29
33
39.38
-2.26
0.53
0.26
0.42
0.48
10
41.55
36
40.41
-2.73
0.79
0.51
0.79
0.71
8
42.32
41
42.56
0.57
0.82
0.37
0.72
0.89
6
43.72
42
42.81
-2.09
0.82
0.65
0.61
0.50
11
44.29
43
43.11
-2.66
0.52
0.26
0.35
0.30
17
46.55
47
45.18
-2.95
0.49
0.14
0.36
0.57
9
50.67
59
49.12
-3.06
0.79
0.47
0.85
0.91
Orthogonality (819 DOFs) (438 DOFs) (144 DOFs)
Table 4: Node2 MST – On orbit axial configuration – final correlation
Conclusions
The validation of Columbus and Node2 mathematical models had taken significant
advantages from the previous experience of MPLM. For all these projects, related
to the ISS and to be launched using Space Shuttle, a MST was performed, to
acquire the data necessary to update the mathematical model. The experience of
MPLM MST has allowed for Columbus and Node2 to obtain a very high quality of
empirical data, reducing the time and the costs of the test. Before MPLM MST a
test fixture, reproducing the boundary constraints at payloads trunnions, as in
Shuttle cargo bay, was designed and experimentally verified by Alenia. The same
text fixture was used for Columbus and Node2. Due to the various manifests of
launch, for MPLM a deep study was performed to define a test configuration, able
to satisfy the test requirements and representative of any flight manifest. To reduce
the high modal density, for Columbus and Node2, characterized by the presence of
different typology of secondary structures, a test configuration was defined using
the same approach used for MPLM. During the tests some problems of nonlinearity arose. But considering that the causes were the same that generated the
problems during MPLM MST a quick solution was applied. To validate the area of
connection between Columbus, Node2 and the ISS the method of the residual
flexibility, applied for MPLM, was abandoned, due to the difficulty in treating the
data acquired. An application of Mass Loading Method was preferred for Columbus
and Node2.
For each project the updating activity was a very hard task that the previous
experiences and the software dedicated were not able to simplify. But thanks to the
quality and reliability of the empirical data, the validation of the mathematical
models was always successfully completed.
References
[1] Admire J.R., Tinker M.L. and Ivey E.W., Residual Flexibility Test Method for
verification of Constrained Structural Models, AIAA Journal, Vol.32, No. 1,
170-175, January 1994.
[2] Bellini M. and Fleming P., Definition of the Mini Pressurized Logistic
Module(MPLM) Test Analytical Model (TAM) to better changes of
conducting an adequate Modal Survey Test, Proceedings of the Eleventh
Symposium on Structural Dynamics and Control, Virginia Tech, Blacksburg ,
Virginia, USA, May 1997.
[3] Bellini M, Quagliotti F and Spinello D., Columbus Modal Survey Test for the
On-Orbit Mathematical Model Validation, Proceedings of the Fourth
International Symposium on Environmental testing for Space Programmes,
Liège, Belgium, June 2001.
[4] Bookout P., Rodriguez P.I., Tinson I. and Fleming P., MPLM On-Orbit
Interface Dynamic Flexibility Modal Test, AIAA 2001-1382.
[5] Brondolo D. and Gargioli E., ISS Modules that have seen an optimised design,
validation and qualification approach, IAC Congress, Bremen, Germany, 2003
[6] Fleming P., Bellini M., Calvi A., Dillinger S. and Grillenbeck A., Modal
Survey Test of the Mini Pressurized Logistic Module (MPLM), Proceedings of
the Sixteenth International Modal Analysis Conference, Santa Barbara,
California, USA, February 1998.
[7] Notarnicola M., Quagliotti F. and Fleming P., Test Stand for Space Shuttle
payloads modal survey test design and verification, Proceedings of CEAS
International Forum on Aerolasticity and Structural Dynamics, Roma,
Italy,June 1997.
[8] Tinker M.L. and Cutchuns M.A., Model Correlation issues in Residual
Flexibility Testing, ASME Design Engineering Technical Conferences, 1997.
Download