free vibration analysis of micro-satellite...
TRANSCRIPT
86
CHAPTER 5
FREE VIBRATION ANALYSIS OF MICRO-SATELLITE
STRUCTURE
5.1 NORMAL MODE ANALYSIS
Normal modes analysis computes the natural frequencies and mode
shapes of a structure. The natural frequencies are the frequencies at which a
structure tends to vibrate if subjected to a disturbance. The deformed shape of
the structure at a specific natural frequency of vibration is termed as normal
mode of vibration. Some other terms used to describe the normal mode are
mode shape, characteristic shape, and fundamental shape. Each mode shape is
associated with a specific natural frequency. Natural frequencies and mode
shapes are functions of the structural properties and boundary conditions. If
the structural properties change, the natural frequencies change, but the mode
shapes may not necessarily change. If the boundary conditions change, then
both the natural frequencies and mode shapes change.
Normal mode analysis is also called the real Eigen value analysis.
Normal mode analysis forms the foundation for a thorough understanding of
the dynamic characteristics of the structure. Normal mode analysis is
performed for the following reasons:
1. Assessing the dynamic interaction between a component and
its supporting structure, if the natural frequency of the
supporting structure is close to an operating frequency of the
87
component then there can be a significant dynamic
amplification of the loads.
2. Assessing the effects of design changes on the dynamic
characteristics.
3. Using the modes in the subsequent forced response analysis.
4. Using the natural frequencies as a guide to select the proper
time or frequency step for a transient and frequency response
analysis.
5. Assessing the degree of correlation between modal test data
and analytical results.
6. In normal modes analysis the Eigen values and Eigen vectors
of the model are determined. For each Eigen value, which is
proportional to a natural frequency, there is a corresponding
Eigen vector, or mode shape.
Normal modes analysis for the undamped free vibrations is as
follows:
[K – liM] { fi }=0
Where ‘K’ is the system stiffness matrix, ‘M’ is the system mass matrix and
‘li’ and ‘fi’ are to be computed. ‘li’are the Eigen values and ‘fi’ are the Eigen
vectors or mode shapes. The Eigen values are related to the natural
frequencies as follows:
fi = li/2
Eventhough number of methods is available for Eigen value
extraction, Lanczos method is used for extracting Eigen values for the
dynamic analysis of the micro-satellite
88
5.1.1 Lanczos Method
The Eigen value extraction for a particular model depends on four
factors (i) the size of the model (the total number of degrees-of-freedom as
well as the number of dynamic degrees-of-freedom), (ii) the number of Eigen
values desired, (iii) the available real memory of computer, and (iv) the
conditioning of the mass matrix (whether there are mass less degrees-of-
freedom).
The Lanczos method overcomes the limitations and combines the
best features of the other methods. It requires that the mass matrix be positive
semi definite and the stiffness matrix be symmetric. Like the transformation
methods, it does not miss roots, but has the efficiency of the tracking
methods, because it only makes the calculations necessary to find the roots
requested by the user. This method computes accurate Eigen values and Eigen
vectors. The Lanczos method is the preferred method for the most medium to
large-sized problems, since it has a performance advantage over other
methods.
In general, the Lanczos method is the most reliable and efficient,
and is the recommended choice of the FEA software MSC NASTRAN.
5.2 FREE VIBRATION ANALYSIS
The natural frequencies of the satellite are of special interest as the
sensitive electronic instrumentation and onboard computers should not be
affected due to the vibrations developed on the satellite structure. The
frequencies of vibration of launch vehicle at the time of launching need to be
considered to avoid conditions of resonance occurring in the satellite during
the launch. In the current problem the frequencies of vibration of launch
89
vehicle during launch are specified. The aim is to design the satellite
structure or modify the given structure such that resonance would not occur.
In general there are four methods for analysis and dynamic control
of satellite which are:
1) Using passive or active damping: This method use materials
with high damping coefficient or in active case, the actuators
are used.
2) Controlling and decreasing the dynamic response of structure
in sensitive sections: In this method, actuators are used in
specified sections, which are very sensitive to vibration and
displacement. For instance, in the satellite, the camera location
is very important; the method is applied for reducing the
vibration at that specified region.
3) Controlling the applied load function: This method try to
control the applied load not to happen in the same frequency
as that of the system’s first natural frequency.
4) Frequency separation: This method is one of the most
common approaches used for vibration control. In this method
the designer try to shift the natural frequency of the structure
to a safe range far from the excitation loads frequency.
By considering the characteristics of the micro-satellite and the
above four suggested methods, it could be realized that the best-fitted method
is third method. The first method needs viscous-elastic or vibration damping
system that increases the cost and weight of the satellite. The second approach
is not applicable due to lack of sensitive section in the micro-satellite based
on its mission (like camera). In the third approach, no control over the loads
caused by launcher, it was decided to take into account the launcher
90
specification as design constraints. Therefore, as a unique solution the
geometrical parameters has to be changed. Actually this can decrease the
system response to excitations loads and prevents from resonance
phenomenon. To avoid dynamic coupling between the low frequency vehicle
and satellite modes, based on the Launcher loads and frequency requirements
discussed earlier, the satellite should be designed with a structural stiffness
which ensures that:
Longitudinal frequency > 90Hz
Lateral frequencies > 45 Hz
5.2.1 The Finite Element Model
The finite element model of the satellite was constructed using the
pre and post-processing package, MSC PATRAN. The interface ring made of
aluminium alloy was modeled by shell elements made of an isotropic
material. The sandwich panels were modeled by laminate shell elements made
of composite material which use the properties of Aluminum alloy for the
faces and the properties of the aluminum honeycomb for the core. The
subsystems were modeled as lumped masses positioned at its center of gravity
and possessing both, translational and rotational inertia properties. The
subsystems were connected to the sandwich panels by beam elements which
are shown in Figure 5.2. Interconnections of panels are modeled by rigid
elements (NASTRAN’s RBE2 element) which are shown in Figure 5.3.
Boundary conditions were simulated by single point constraints located at the
attachment points of the cylindrical adapter to the rocket which is shown in
Figure 5.4.
91
5.2.2 Convergence study
The objective of convergence study is to determine the optimum
number of nodes required for carrying out a finite element analysis and to
minimize the memory requirements for the analysis without any compromise
on the accuracy of the obtained results. Convergence study usually involves
plotting of a particular result parameter against mesh size. The value of mesh
size at which the result parameter almost becomes a constant is the ideal one.
A mesh size finer than this indicates the wastage of memory without any
additional increase in accuracy, while a coarser mesh size than this indicates
inaccurate results. The finite element model of the micro-satellite structure without
solar panels and subsystems was generated for various mesh sizes (element size in
m), namely 0.2, 0.15, 0.1, 0.07, 0.04 and 0.01. The natural frequency for the first
three modes of vibration in each case was determined.
Figure 5.1 Convergence curves
From the Figure 5.1, it can be seen that natural frequencies for the firstthree modes converge to a steady value for a mesh size of about .01m and after
12000 nodes. Hence for all analyses to be carried out henceforth a mesh size same as
or finer than this will be used. The analysis has to be carried out after modeling the
satellite structure with solar panel substrates and the subsystems attached. The
summary of the finite element model of the micro-satellite is shown in Figure
5.5 is given below.
92
Number of grid points = 36715
Number of CBAR elements = 1215
Number of CONM2 elements = 46
Number of CQUAD4 elements = 34152
Number of CTRIA3 elements = 8
Number of RBE2 elements = 107
Figure 5.2 Equipment deck showing subsystems modeled as lumped
masses
93
Figure 5.3 RBE2 elements for connecting panels
Figure 5.4 SPC applied at the attachment point of
the adapter to the launcher
94
Figure 5.5 Finite element model of the micro-satellite
5.2.3 Subsystem modeling
The micro-satellite considered have several subsystems that are
generally mounted to the bottom deck, middle deck, top deck, cross webs and
vertical webs. The micro-satellite structural panels are designed to withstand
the dynamic loads. Frequencies of these excitations range up to 100 Hz and
therefore the frequencies of interest for structural design are often limited to
the spacecraft modes ranging up to 120 Hz. Theoretically the natural
frequency of the micro-satellite is determined using the finite element analysis
software MSC PATRAN/MSC NASTRAN. The determination of the natural
modes of the micro-satellite structural panels involves modeling of the large
number of subsystems also.
95
The Finite element model of the subsystem for determining thenatural modes of the structural panels can be created by the followingmethods.
1. Modeling of subsystems in detail. The subsystems aregenerally designed to have their fundamental naturalfrequencies to be well above 120 Hz to reduce the dynamicloads. The frequencies of the fundamental modes of thesubsystem panels are expected to be around 30 – 40 Hz.Hence for the prediction of the first few natural modes there isno need to have the detailed FE model of the subsystemswhich make the FE model to be very huge, but simplemathematical models should be sufficient.
2. Distributing of mass uniformly over the panels or deck plates.The distribution of mass on the deck plates is not sensitive indetermining the more number of out of plane bending mode.
3. The mass of the subsystem lumped at its C.G and connected tothe deck by using RBE2 elements or RBE3 elements. RBE2and RBE3 elements are built in multi-point constraintequations in the software and it acts as rigid link between thedependent and independent nodes. Dependent nodes are thenodes taken at the locations where the subsystem is connectedto the deck plates and the independent nodes are taken at thenode where the mass is lumped. The rigid elements connectedto the deck add unnecessary stiffness to the deck.
4. Mass of the subsystem lumped at its C.G and connected to thedeck at the interface points through beam or rod elements.Both these elements can simulate the inertial characteristics ofthe subsystems and can also determine large number of out ofplane bending modes of the structural panels. The beam androd element do not add unnecessary stiffness to the deck plateand can best predict the theoretical modes of the deck platesand finally the modes of the micro-satellite.
96
For the estimation of the modes of the deck plates or the micro-satellite it is not necessary to model the subsystems in detail and hence it isrequired to lump the mass of the subsystem at its C.G and attach the lumpedmass element to the deck plate at its interface points through beam elements.
5.3 SIZE OPTIMIZATION OF MICRO-SATELLITE
STRUCTURE
The optimization of the micro-satellite structures like bottom deck,top deck, and middle deck, cross webs, vertical webs and solar panels whichwere all made by sandwich panel structure can be done by
Altering the face sheet thickness
Altering the honey comb core thickness
Increasing the face sheet thickness is not desirable since theincrease in mass due to slight change in thickness will be high for the facesheet owing to its high density (2700 kgm-3) when compared to that ofhoneycomb (32 kgm-3) and decreasing the face sheet thickness is not possibleas the minimum available face thickness is 0.20 mm.
Figure 5.6 Effect of increasing panel core thickness
97
From the Figure 5.6, tc2 > tc1 (where ‘tc1’ and ‘tc2’ are the core
thickness of the deck plates) and hence I2 > I1 (where ‘I2’ and ‘I1’ are the
Moment of Inertia values). Thus increasing the thickness increases stiffness
(since K I) and natural frequency.
Based on the available aluminium core thickness of 15mm, 20mm,
30mm and 40mm size, optimization study was done to identify the micro-
satellite structure with less mass and more stiffness values to satisfy the
frequency constraints. The optimization was started with the core thickness
and mass values as shown in Table 5.1.
Table 5.1 Core thickness values before optimization
Core thickness of Micro-satellite structure (facesheet thickness keptconstant) in mm
Mass
(kg)
Natural Frequency
LateralX (Hz)
LateralY (Hz)
LongitudinalZ (Hz)
Bottom Deck = 30
Middle Deck = 30
Top Deck = 30
Vertical Web= 30
Cross Web = 30
Solar Panel = 15
38.56 46.86 47.86 128.94
Altair Optistruct the optimization software was used and the
constraints given were to identify the structure with less mass having lateral
frequencies greater than 50 Hz and the longitudinal frequency greater than
100Hz.The results obtained were shown in Table 5.2 and percentage decrease
in mass and percentage increase in natural frequency is shown in Table 5.3.
98
Table 5.2 Core thickness values after optimization
Core thicknessvalues arrived afteroptimization (mm)
Mass
(kg)
Natural Frequency
LateralX (Hz)
LateralY (Hz)
Longitudinal
Z (Hz)
Bottom Deck =29.8
Middle Deck = 15
Top Deck = 15
Vertical Web= 15
Cross Web =15
Solar Panel = 15
33.98 50.7 52.0 132.92
Table 5.3 Percentage variation in mass and frequencies
Percentagedecrease inmass (%)
Percentage increase in NaturalFrequency
LateralX (%)
LateralY (%)
LongitudinalZ (%)
12 7.5 8 3
Since the huge interface ring was attached to the bottom deck and
the subsystems are placed on both sides of the bottom deck and middle deck
the structure needs extra thickness for the inserts and fasteners. Hence it was
decided to keep the micro-satellite structure with the following specifications
as shown below.
Bottom deck - 40mm thick core, Face sheets 0.25mm thickTop deck - 15mm thick core, Face sheets 0.20mm thick
99
Middle deck - 20mm thick core, Face sheets 0.25mm thickCross web - 15mm thick core, Face sheets 0.20mm thickVertical web - 15mm thick core, Face sheets 0.20mm thickSolar panel - 15mm thick core, Face sheets 0.20mm thickTotal mass = 35 kg
The results of the analysis for free vibration analysis of the micro-
satellite with above dimensions of the primary structure with all the
subsystems, solar panel substrates, antennas and interface ring are given in
Table 5.4.
Table 5.4 Free vibration analysis results
Mode Number Frequency(Hz) Mode shape
1 26.99 Antenna mode
2 27.12 Antenna mode
3 27.20 Antenna mode
4 27.26 Antenna mode
5 27.27 Antenna mode
6 27.41 Antenna mode
7 27.43 Antenna mode
8 27.46 Antenna mode
9 56.96 Lateral XX mode
10 58.11 Lateral YY mode
11 111.19 Torsion mode
12 135.84 Longitudinal mode
From the Table 5.4 it is evident that the first eight modes of the
analysis show the antenna mode and one of the antenna modes is shown in
Figure 5.7. The ninth and tenth modes show the lateral modes which are
100
shown in Figure 5.8 and Figure 5.9. The eleventh mode shows the torsion
mode which is shown in Figure 5.10 and the twelfth mode shows the
longitudinal mode which is shown in Figure 5.11. The above obtained natural
frequency values satisfy the launcher requirements.
From the optimization study the following information wereobtained
1. The natural frequencies increase with increase in bottom deckcore thickness which was verified by strain energy analysis.
2. The natural frequencies decrease with increase in top deckcore thickness.
3. The natural frequencies decrease with increase in middle deckcore thickness.
Figure 5.7 Antenna mode of vibration of the satellite
101
Figure 5.8 Lateral vibration of the satellite in the XX direction
Figure 5.9 Lateral vibration of the satellite in the YY direction
102
Figure 5.10 Torsion mode of vibration of the satellite
Figure 5.11 Longitudinal vibration of the satellite in the ZZ direction
103
5.4 MODAL STRAIN ENERGY
Modal strain energy is a useful parameter for identifying the
elements of the finite element model for design changes to meet the frequency
requirements. Elements with large values of strain energy in a mode indicate
the location of large elastic deformation (energy). These elements are those
which directly affect the deformation in a mode mostly. Therefore, changing
the properties of these elements with large strain energy should have more
effect on the natural frequencies and mode shapes than if elements with low
strain energy were changed. The Figure 5.12 shows the element strain energy
values of the micro-satellite structure and Figure 5.13 shows the element
strain energy values of the bottom deck. The Table 5.5 shows results of the
modal strain energy analysis of the micro-satellite.
Figure 5.12 Element strain energy values of micro-satellite structures
104
Figure 5.13 Element strain energy values of bottom deck
Table 5.5 Modal strain energy results
Micro-satellite structure Strain energy values (Nm)
Interface of bottom deck and crossweb
9.50*10-3
Bottom deck 3.28*10-3
Interface of ring and bottom deck 3.28*10-3
Middle deck 1.78*10-3
Top deck 1.29*10-3
Solar panel 0.185*10-3
Vertical web 0.381*10-3
From the Table 5.5 the element strain energy values are more in theinterface of bottom deck and the cross web and in the bottom deck. Increasingthe thickness in and around that element will increase the natural frequencyvalues. Hence increasing the thickness of the bottom deck will have moreeffect in increasing the frequency values compared to increase in the thicknessof any other structural members of the micro-satellite.
105
5.5 FREE VIBRATION ANALYSIS OF SUBSYSTEMS
The subsystems are designed to have their fundamental naturalfrequencies to be well above 120 Hz to reduce the dynamic loads. A greatmany creative, innovative and highly useful electronic products have failed tobe profitable because of low reliability, premature failure or fragilepackaging. Failure of satellites can be of mechanical rather than electricalmalfunction. The launch environment usually provides the most severe shockand vibration loads on the satellite and subsystem assembly that will ever beseen during normal operation. Each subsystem design varies according to itsfunctional requirements and it is necessary to do its intended function at anysevere load environment. All the subsystems that are to be accommodatedinside the micro-satellite considered were made of machined aluminiumcomponents. The proper stiffening members are provided in the sidewalls andin top and bottom of the subsystems. Cutouts are provided for the connectorsand sufficient numbers of insert locations were identified to make thesubsystem stiffer. The printed circuit boards are kept inside the subsystem insuch a way that any kind of load does not affect its functioning.
Figures 5.14, 5.17, 5.20 and 5.23 show the CAD model of thesubsystems payload baseband, power electronics, bus electronics and powerdistribution respectively. Figures 5.15, 5.18, 5.21 and 5.24 show the finiteelement model of payload baseband, power electronics, bus electronics andpower distribution respectively created using MSC/PATRAN. Figures 5.16,5.19, 5.22 and 5.25 show the free vibration analysis results of payloadbaseband, power electronics, bus electronics and power distributionrespectively. The Table 5.6 shows the fundamental frequency values of largesubsystems whose values are well above the launcher constraint.
Table 5.6 Fundamental frequency values of subsystems
Subsystem Fundamental frequency (Hz)
Payload baseband 481Power electronics 942.5Bus electronics 424Power distribution 703
106
Figure 5.14 Payload baseband
Figure 5.15 Finite element model of Payload baseband
Figure 5.16 Free vibration analysis result of Payload baseband
107
Figure 5.17 Power Electronics
Figure 5.18 Finite element model of Power Electronics
Figure 5.19 Free vibration analysis result of Power Electronics
108
Figure 5.20 Bus Electronics
Figure 5.21 Finite element model of Bus Electronics
Figure 5.22 Free vibration analysis results of Bus Electronics