60 Scientific Technical Review,2010,Vol.60,No.1,pp.

60-69

UDK: 533.6:517.962
COSATI: 01-03, 12-01

Computational Modal Analysis of the LASTA Aircraft

Jelena Dimitrijević1)
Predrag Kovačević1)

This paper presents the results of the Computational Modal Analysis of a complete finite element model of the
LASTA aircraft. The results of calculation, normal frequencies and normal mode shapes, shown here, will be used for
the further aeroelastic analysis to confirm that this aircraft is free from flutter in the range of the designed speed. A
short theoretical review of the Computational Modal Analysis and Lanczos Method is presented in this paper.

Key words: aircraft, propeller aircraft, trainer aircraft, eigen oscillations, modal analysis, numerical analysis, Finite
Element Method.

Introduction capture the significant physical behavior of the system. The

Finite Element Method is one of the most powerful popular
D URING designing, manufacturing and examination of
an aircraft prototype it is required to confirm that the
aircraft is free from flutter in the range of the designed
mathematics modeling techniques. In brief, the basis of the
FEM is the representation of a body or a structure by an
speed. The flutter analysis and the flight test are used to assemblage of subdivisions called finite elements. These
confirm it. elements are interconnected at joints, which are called
This paper presents the results of the Computational nodes or nodal points. Simple functions are chosen to
Modal Analysis, i.e. the computation of the natural approximate the distribution or variation of the actual
frequencies and the natural mode shapes of the LASTA displacements over each finite element. Such assumed
aircraft by using the Finite Element Method. First, a finite functions are called displacement functions. The final
element model of the LASTA aircraft is done. Then, the solutions yield to displacements at nodes. The attraction of
natural frequencies and the natural mode shapes are the FEM is that it is commercially available in the form of a
estimated using the Computational Modal Analysis. All this user-oriented computer program.
is done using the I-DEAS software. The results of the In practice, the finite element model of any structure
Computational Modal Analysis will be used for the Flutter approximates a real structure using finite elements. If the
Analysis, i.e. the calculation of unsteady generalized behavior of a structure is too complex, the recipe is to
aerodynamic forces and critical flatter speeds of the divide it into substructures that are more manageable [3]. If
LASTA aircraft. these substructures are still too complex, the subdivision
process is continued. The first step in the Finite Element
Method is to subdivide a structure, or an assembly, into
Finite Element Method and Dynamic Analysis many smaller elements such as plates, beams, shafts, etc.,
Finite element modeling is used in the design cycle of or substructures. The accuracy of the model depends on a
new mechanical parts or systems because it offers a number degree of the users’ skill in choosing the discretization. The
of advantages [1], [2]. Using the Finite Element Method most important is to choose an appropriate element type
(FEM), a mathematical model of the structure can be built and size for the parts or substructures, correctly connect all
much more before the first prototype structure is even built. parts and choose appropriate boundary conditions. Then the
In addition, it can be determined how the mass or stiffness overall equations of motion of the structure are made from
changes of the structure affect its static and dynamic the equations describing the motions of each of individual
characteristics and this is its great advantage. parts, or substructures, plus all the boundary conditions at
All structures, regardless of their simplicity, have an the connection points between parts or substructures.
infinitive number of degrees of freedom (DOF). With The Finite Element Method is typically used for the
generalization of the matrix methods and with the following analyses:
advantages of modern computers, engineers started using - Static analysis (static strength to observe the high stress
the Finite Element Method in the static and dynamic and strain areas under static loads),
analysis of large structures. One of the main objectives of - Dynamic analysis (modal analysis, dynamic responses to
selecting an FE model is to reduce the infinitive degree of external dynamic loads, dynamic strength and fatigue).
freedom of systems with complex geometry and material One of the most important fields of using the Finite
properties, to a model with a limited DOF, but which will Element Method is in the dynamic analysis [4]. By using

1)
Military Technical Institute (VTI), Ratka Resanovića 1, 11132 Belgrade, SERBIA
 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT 61

the finite element model, it is possible to simulate the and modal frequency analysis can be based on the results of
response of the real world to an external stimulus. External the natural frequency and natural mode shapes. Another
forces might be of short duration and high amplitude, i.e. important field of use is in the experiment when the natural
impulse forces, which cause immediate structure crash. frequencies and natural mode shapes can be used to guide
Furthermore, they may be of long duration and cyclic in the experiment, for example to help the researchers to
nature and could cause fatigue damage of the structure indicate at which frequencies to vibrate the structure and to
during a long period. In that way, it is possible to predict indicate the best location for the accelerometers. In
structure damage, its crash or fatigue life even before the addition, design changes can also be evaluated by using
first prototype is constructed. natural frequencies and natural mode shapes. Finally, the
In the static analysis, the element size is smaller and a knowledge of the natural frequencies and natural mode
total number of elements is larger. In the dynamic analysis, shapes for a particular structure is important for all types of
the element size is bigger because the calculation method is dynamic analyses.
more complex and then the total number of elements is The dynamic response verification process [7] of the
smaller. Time-varying characteristics make the dynamic project is shown in Fig.1. The prototype can be virtual
analysis more complicated and more realistic than the static (finite element model) or real (physical prototype).
analysis [5]. Two basic aspects of the dynamic analysis, Analytical, computational results (natural frequencies and
which differentiate it from the static analysis, are: natural mode shapes) are obtained based on the virtual
- Dynamic loads are applied as a function of time. prototype using the Computational Modal Analysis and the
- Time-varying external dynamic load induces a time-varying experimental results based on the real prototype using
response (displacements, velocities, accelerations, forces, Experimental Modal Analysis, Vibration Tests, or Flight
and stresses). Tests. The verification of the finite element model and the
The primary task for the dynamic analyst is to determine results of the Computational Modal Analysis can be done
the type of analysis to be performed. The choice of the after the experiments when it is possible to compare the
appropriate mathematical approach depends on the nature modal parameters (natural frequencies and natural mode
of the dynamic analysis in many cases. The extent of the shapes) obtained computationally and experimentally. The
information required from a dynamic analysis also dictates correction of the finite element model and then natural
the necessary solution approach and steps. frequencies and natural mode shapes can be done according
Two basic classifications of the dynamic analysis are: to the experimental results, Fig.1. A new, calibrated virtual
- free vibrations and prototype may be further used for some other kinds of
- forced vibrations. dynamic phenomena or for the dynamic analysis and design
The free vibration analysis is used to determine the basic modification.
dynamic characteristics of the system when the external
forces are zero, i.e., there is no applied load. Free vibrations Design Vibration
can be damped and undamped. If damping is neglected, Requirements Tests
than it is an undamped free vibration.
Flight
Tests
Natural Frequency and Natural Mode Shape
Computational Modal Analysis Physical Modal Experimental
Prototype Test Modal Model
The determination of natural (normal) frequencies and
natural (normal) mode shapes of the structure with damping Prototype
neglected is the first step in the dynamic analysis. Pre-Test
According to natural frequencies and natural mode shapes Correlation
that characterize the basic dynamic behavior of the Virtual Prototype Model Up-
Analytical
structure, we have the indication of how the structure will Modal Mo- dating
respond to dynamic loading. The natural frequencies of a
structure are the frequencies at which the structure naturally
tends to vibrate if subjected to a disturbance. An Calibrated Virtual
Prototype
appropriate deformed shape of the structure at a specific
natural frequency of vibration is its normal mode shape, or
the normal mode of vibration. Natural frequencies and Response Analysis Design
natural mode shapes of any structure are estimated owing to Modifications
the Computational Modal Analysis using finite element
model of that structure [6]. They depend on the structural Figure 1. Dynamic response verification process
properties and boundary conditions.
Computation of natural frequencies and natural mode
shapes of a structure could be of great significance. They Theoretical review equation of motion for
tell us at what frequencies the structure can be excited into
resonant motion. In many cases, this information is
undamped free vibration
sufficient for modifying the structural design in order to Natural frequencies and natural mode shapes are
reduce noise and vibration. In addition, a dynamic obtained as the results of the eigenvalue problem [8], i.e.
interaction between a component and its supporting equation of motion for undamped free vibrations.
structure is very important because the component could The general equation of motion in the matrix form is:
cause structural damage or failure if its operating frequency
is close to one of the natural frequencies of the structure. A
[M ]{u} + [C ]{u} + [K ]{u} = {P} (1)
transient response, frequency response, response spectrum where:
62 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT

[M ] - mass matrix 2. If det ([K ] − ω 2 [M ]) = 0 , the solution is non-trivial

[C ] - damping matrix {φ} ≠ 0 .
[K ] - stiffness matrix The eigenvalue problem is then reduced to the following
{P} - force vector equation:
{u} - displacement vector det ([K ] − ω 2 [M ]) = 0 (7)
Disregarding the dissipative forces [C ]{u} and when the
external (excitation) forces are zero, {P} =0, the equation of and for ω 2 = λ is:
motion (1) reduces to equation of motion for undamped free
vibrations and its matrix form is: det ([K ] − λ [M ]) = 0 (8)

[M ]{u} + [K ]{u} = 0 (2) The determinant is zero only at the set of discrete
eigenvalues λi , or ωi2 . The corresponding i-th eigenvector
Dissipative forces are neglected because there is no
possibility to estimate the damping matrix in this phase of is {φ}i . So the eq. (3) can be rewritten as:
the project. Damping effects can only be included
approximately and that is way damping effects are ([K ] − ω [M ]){φ} = 0
2
i i i=1,2,3… (9)
neglected altogether in many analysis. Dumping is obtained
The number of eigenvalues and eigenvectors is equal to
experimentally after the Experimental Modal Analysis
the number of degrees of freedom. In structural dynamics,
when natural frequencies and natural mode shapes are
each eigenvalue corresponds to the natural frequency and
obtained as well.
each eigenvector to the natural mode shape. Together they
Using the Modal Decomposition Method and the
define one free vibration mode of the structure. The i-th
condition that it is a harmonic vibration {u} = −ω 2 {u} , eigenvalue λi is related to the i-th natural frequency as
eq.(2) reduces to the eigenvalue problem: follows:
([ K ] − ω 2 [ M ]){φ} = 0 (3)
fi =
ωi
(10)
2π
where:
{φ} - eigenvector, or natural mode shape where:
ω - circular natural frequency. fi - i-th natural frequency
Eq. (3) is called eigenproblem and can be interpreted as
a set of homogeneous algebraic equations. It represents ω i = λi
base for the eigenvalue problem calculation. The first
necessary step in the dynamic analysis of the structure is to The mode shape of a linear elastic structure that is in free
solve the characteristic equation. or forced vibration, at any given time, is a linear
By pre-multiplying eigenequation (3) by the matrix combination of all of its natural mode shapes, or
mathematically:
[ M ]−1 ,or by the matrix [ K ] , it can be written as a
−1

standard eigenvalue problem: {u} = ∑ {φ}i ξi (11)

i
[A]{φ} = λ{φ} (4)
where:
where: {u} - vector of physical displacements
[ A] = [ M ] [ K ] - square dynamic matrix
−1
{φ}i - i-th natural mode shape
λ = ω2 - eigenvalue ξi - i-th modal displacement
The matrix [ M ] must be non-singular. A consistent
−1 There is a property of orthogonallity between
eigenvectors. It is very easy to present the evidence for
mass matrix always satisfies this condition. A stiffness these properties. From eq. (9) with respect to the equation
matrix will satisfy this condition if the rigid body motion
degrees of freedom are suspended. If the rigid body motion ωi2 = λ i , for one vector {φ}i is obtained:
degrees of freedom exist, the Shift Frequency Method [9] is
applied and it transforms eq. (3) to:
[K ]{φ}i = λ i [M ]{φ}i (12)

( [K ′]− (ω 2
) )
± S 2 [M ] {φ} = 0 (5)
and after pre-multiplying by another transposed vector,
{φ}Tj the previous equation becomes:
where:

[K ′] = [K ] ± S 2 [M ] (6) {φ}Tj [K ]{φ}i = λ i {φ}Tj [M ]{φ}i (13)

Eq. (3), eigenequation, has two possible solutions: In the same way, starting from the vector {φ} j and pre-
1. If det ([K ] − ω [M ]) ≠ 0 , the only possible solution is
2
multiplying by vector {φ}i equation (12) becomes:
T

{φ} =0. This is a trivial solution, which does not give

any useful information from a physical point of view, {φ}Ti [K ]{φ} j = λ j {φ}Ti [M ]{φ} j (14)
except that there is no motion.
 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT 63

If [K ] and [M ] are real symmetric matrixes, it can be The methods of eigenvalue extraction belong to one of
written: the following groups [10]:
- Methods with direct egenvalue extraction finding zero
{φ}Ti [M ]{φ} j = {φ}Tj [M ]{φ}i {φ}Ti [K ]{φ} j = {φ}Tj [K ]{φ}i determinants.
- Iterative methods (Inverse Iteration Method, Vector
Therefore, after subtracting eq. (14) from eq. (13) it is Iteration Method, Subspace Iteration Method) - In an
obtained: iterative method, a certain number of roots are extracted
at a time by iterative procedures applied to the original
( )
0 = λ i − λ j {φ}i [M ]{φ} j
T
dynamic matrix.
- Transformation methods (Jacobi Method, Ritz Method,
( )
And then if λ i − λ j ≠ 0 , i.e. if i ≠ j : Lanczos Method, Givens Method, QL Hausholder
Method, QZ and QR Method) - In the transformation
{φ}Ti [M ]{φ} j = 0 and {φ}i [K ]{φ} j = 0
T
(15)
method, the matrix of coefficients is first transformed into
a special form (diagonal or tridiagonal) from which the
eigenvalues may easily be extracted.
Eq. (15) define the orthogonal properties of the vectors
- Reduction method (Guyan Method) - In the reduction
with respect to the system mass or stiffness matrixes.
method, the structure is divided into substructures and
If i = j , then these two vectors are not orthogonal and
after the eigenvalue extraction, they are synthesis again.
then eq. (14) is equal to some non zero scalar constant: The characteristics of the Sturm series, the procedure for
reortogonalization and deflection, can be used to improve
{φ}Tj [M ]{φ}j = m j , m j - ј-th generalized mass (16) methods. Some of these methods, as presented in Fig.2, are
better for small degree of freedom systems (QL Hausholder
Method, QZ and QR Method, Guyan Method and Jacobi
and Method), while the others (Subspace Iteration Method,
Givens and Lanczos Method) are superior for large
{φ}Tj [K ]{φ}j = k j , structures.
Number of
2 Eigenvalues desired
k j = ω m j - ј-th generalized stiffness (17)

In addition, from eq. (16) and eq. (17), Rayleigh’s

equation is obtained:

{φ}Tj [K ]{φ} j JACOBI, RITZ

GIVENS
INV, SVI, GIVENS,
LANCZOS, +

ω2j = (18) QL, QR, QZ REDUCTION

{φ}Tj [M ]{φ} j
INV

Order of problem
number of nodal DOF
Eq. (15) are known as the orthogonal property of normal
modes, which means that each normal mode shape is
Figure 2. Selection of the method for eigensolution
unique and different from any other. In addition, one
normal mode shape cannot be obtained through a linear When the number of degrees of freedom increases, then the
combination of any other normal mode shapes. necessary computer memory and time for extraction also
For practical examination natural mode shapes can be increase. The Iterative Methods and the Lanczos Method are
scaled. If one of the elements of the eigenvectors is the most efficient for egenvalue extraction for the structures
assigned a certain value, the rest of the (n-1) elements are with a few hundred thousand degrees of freedom.
also fixed because the ratio between any two elements is
constant. Thus, process of adjusting the elements of the
natural mode shapes to make their amplitude unique is Lanczos Method
called normalization. The resulting scaled natural modes The Lanczos Method overcomes the limitations and
are called orthonormal modes. The mass (inertial) combines the best features of the transformation and
normalization is the default method of eigenvector iterative methods [11]. It does not miss roots
normalization and scales each eigenvector, natural mode (transformation methods), and it only find the roots
shape, to a unit value of the generalized mass: requested by the user (iterative methods). The Lanczos
Method is the preferred method for most medium and large
{φ}Ti [M ]{φ} j = 1 (19) sized problems, since it has a performance advantage over
other methods. In addition, the advantage is its capability of
When the structure is free, without support, then, besides parallel processing.
elastic modes, it has rigid-body modes. The natural In 1950, Lanczos published his procedure for efficient
frequency of the rigid-body mode is equal to zero. determination of m first or last in series egenvalues and
appropriate eagenvectors of nxn symmetric matrix.
Methods for natural frequency and natural mode However, not until 1975, Lanczos Method was included in
shape estimation software NASTRAN and in many other known FEM
software [12]. Lanczos Method is included in the static and
There are large numbers of numerical methods of the real dynamic analysis software that is developed in theVTI.
eigenvalue extraction for its different type and size. While The symmetric matrix [A] is obtained (eq. (4)) using
most of the methods can be applied to all problems, the choice certain transformation of the matrixes [M] and [K] so this
is often based on the efficiency of the solution process. procedure can be applied to natural frequency extraction.
64 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT

Pre-multiplying by the matrix [V ] and using the LASTA Aircraft – Basic Information
T

replacement {φ } = [V ]{y} , eq. (4) can be written as:

The LASTA aircraft is intended for the basic figural,
[V ] [A][V ]{y} = λ [V ] [V ]{y}
T T
(20) navigational, instrumental and night traning. It is shown in
Fig.3.
where,
[V ] - matrix of Lanczos vector v, size nxm
{y} - new eigenvector dimension m
If it is assumed that Lanczos vectors [13] are
orthonormal with respect to the mass matrix, {v}i {v} j = δ ij
T

( δ ij - Kronecker delta), than [V ] [V ] = [I ] is valid, where

T

[I ] is the unit matrix, eq. (20) can be transformed into the

standard eigenvalue problem:

[T ]{y} = λ{y} (21)

[T ] = [V ] [A][V ] , size mxm, and can be written as:

T

⎡α1 β1 ⎤
⎢β α β2 ⎥
⎢ 1 2 ⎥
⎢ β2 α3 β3 ⎥
[T ] = ⎢ ⎥ (22)
⎢ • • • ⎥
⎢ Figure 3. Aircraft LASTA
β m−2 α m−1 β m−1 ⎥
⎢ ⎥ The LASTA aircraft is a low wing metal construction
⎣⎢ β m−1 α m ⎦⎥
aircraft, with two, one behind another, classical non-
The matrix [T ] is three diagonal and symmetric and the ejection seats and a tricycle landing gear. The power group
consists of one piston engine with six cylinders ''AVCO-
vectors v are defined in such a way to be orthogonal. When LYCOMING AEIO-540-L1.B5D" and two-blade propellers
the matrix [T ] is defined as eq. (22), especially when it is "HARTZELL HC-C2YR-4CF/FC 84756". The aircraft
mxm (m<<n) size, then eq. (21) can easily be solved using cabin area has heating and aeration. Windscreen defogging
the direct determinant egenvalues calculation. and de-icing is enabled. The LASTA aircraft uses concrete
The Lanczos procedure for every new vector calculation and prepared grass runways and posses possibility of
uses the information that all previous calculated vectors independent launching.
contain. For the standard eigenvalue problem, a new vector
is presented as the linear combination of ј orto-vectors v,
previously determined, and the residue {r} j that is Finite Element Model of the LASTA Aircraft
orthogonal to all of them: The first step in finite element modeling is to separate
{r } j = [A]{v} j , the structure, here aircraft, and view the rest as masses and
forces. Second, the aircraft structure is divided to
or substructures - wings, fuselage, stabilizers, engines, landing
gears, etc. Then each substructure is reduced into its
{r } j = {r} j + α j {v} j + β j −1 {v} j −1 + components - rings, ribs, spars, cover plates, actuators, etc.
(23) It is continued through as many levels as it is necessary.
+ β j − 2 {v} j − 2 + ... + β1 {v}1
Then, the element type and the element size are chosen
Therefore, it is necessary to determine the coefficients before the beginning of the finite element modeling of the
α j , β j −1 ,..., β1 and the vector {r} j . After the Lanczos vector LASTA aircraft [14]. Elements have to be small enough to
give more precise results, but, on the other hand, it is
determination, eq. (23) can be rewritten as: important that a total number of the degree of freedom
would not be too much large because of the computer
[A]{v} j = β j {v} j +1 + α j {v} j + β j −1{v} j −1 (24)
possibility. The size and type of elements are adapted to the
structures. The aircraft structure is modeled by shell, beam
or in a matrix form:
and three-dimensional finite elements. All parts of the
[A][V ] = [V ][T ] + β m vm+1 {e}
T
(25) aircraft structure are modeled separately. The masses of the
non supporting parts are modeled as the nonstructural or
where, the vector {e} = {0 0 0 ... 1} is size m. In the concentric masses. The structural parts are joined into
T

subassemblies and then the stiffness calculation and the

end, the matrix [T] is obtained from the equation modal analysis are done. Finite element models of some
[V ]T [A][V ] = [T ] . substructures are shown from Fig.4 to Fig.8.
 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT 65

Figure 4. Wing rib and the stringer FE model

Figure 8. Fitting FE model

The main assembly of the aircraft finite element model is

made by join subassemblies [15]. The subassemblies are
connected in points where the tie-down fittings are
positioned. These connections are modeled as beams and
strings. In the end, the masses of pilots, fuel in reservoir,
oil, equipment and flight controls are added to the aircraft
FE model. The complete FE model of the LASTA aircraft
has 42.129 finite elements, 37.367 nodes and 220.536
degrees of freedom and it is shown in Fig.9. The boundary
conditions are free-free and the Shift Frequency Method is
applied.

Figure 5. Fuselage frame and the stringer FE model

Figure 9. LASTA aircraft FE model

Figure 6. Horizontal stabilizer and the elevator FE model

Computational Modal Analysis
The Computational Modal Analysis or the calculation of
the natural frequencies and the natural mode shapes are
done for the symmetric and anti-symmetric case according
to the recommendation from [16]. The difference between
these two cases is in the aileron string stiffness. The aileron
string stiffness is k=200kN/m for the symmetric case and
k=20kN/m for the anti-symmetric case.
The frequencies of the elevator and rudder rotation are
fitted according to the command control circuits stiffness
obtained experimentally.
After the modal analysis, symmetrical natural mode
shapes are chosen from the symmetric case calculation and
anti-symmetric natural mode shapes from the anti-
symmetric case calculation. All local mode shapes are
neglected. All natural mode shapes are inertial normalized.
Figure 7. Fin and the rudder FE model
66 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT

Natural Mode Shapes

The frequencies and mode shapes description of the finite
element model of the LASTA aircraft are presented in Table 1.
All natural mode shapes are shown from Fig.10 to Fig.26.

Table 1. Natural mode shapes

f
No. Mode shape description Figure
(Hz)
1 4.913 Aileron anti-symmetric rotation Fig.10

2 7.316 Rudder rotation Fig.11

3 8.140 Elevator rotation Fig.12

4 11.397 First wing bending Fig.13

Figure 12. Mode 3 - Elevator rotation
5 11.745 Fuselage torsion Fig.14

6 14.596 Aileron symmetric rotation Fig.15

7 15.428 Horizontal stabilizer anti-symmetric bending Fig.16

8 17.529 Fuselage symmetric bending in the vertical plane Fig.17

Anti-symmetric coplanar wing and

9 20.688 Fig.18
fuselage bending

10 22.711 Second wing bending Fig.19

11 23.094 Symmetric coplanar wing Fig.20

12 23.831 Horizontal stabilizer symmetric bending Fig.21

13 24.367 Flap anti-symmetric rotation Fig.22

14 24.759 Flap symmetric rotation Fig.23

15 26.335 Fuselage anti-symmetric torsion Fig.24 Figure 13. Mode 4 - First wing bending

16 29.380 Fin bending Fig.25

17 32.384 Wing symmetric torsion Fig.26

Figure 14. Mode 5 - Fuselage torsion

Figure 10. Mode 1 - Aileron aniy-symmetric rotation

Figure 11. Mode 2 - Rudder rotation Figure 15 Mode 6 - Aileron symmetric rotation
 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT 67

Figure 16. Mode 7 - Horizontal stabilizer aniy-symmetric bending Figure 20. Mode 11 - Symmetric coplanar wing

Figure 17. Mode 8 - Fuselage symmetric bending in the vertical plane Figure 21. Mode 12 - Horizontal stabilizer symmetric bending

Figure 18. Mode 9 - Anty-symmetric coplanar wing and fuselage bending Figure 22. Mode 13 - Flap anti-symmetric rotation

Figure 19. Mode 10 - Second wing bending Figure 23. Mode 14 - Flap symmetric rotation
68 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT

LASTA aircraft will be used for the calculation of unsteady

generalized aerodynamic forces and critical flatter speeds. The
natural frequencies and the natural mode shapes demonstrate
where to expect modes on the Ground Vibration Test (GVT) –
Experimental Modal Analysis (EMA).
A real value of the natural frequencies, the natural mode
shapes and the damping will be obtained after the Ground
Vibration Test. These results will be then used for the final
calculation of the unsteady generalized aerodynamic forces
and critical flatter speeds. In addition, according to these
results of frequencies and damping, the aircraft finite
element model will be verified and fit for further use in the
analysis of the aeroelastic phenomena (aeroelastic twisting,
reverse, critical flutter speed calculation, aeroelastic
Figure 24. Mode 15 - Fuselage aniy-symmetric torsion response to gust, etc.) and aircraft dynamic response (under
armament action, landing, ground roll and taxiing ).

Literature
[1] BATHE,K.J., WILSON,E.L.: Numerical Methods in Finite Element
Analysis, Prentice-hall, inc., Englewood Cliffs, New Jersey, 1976.
[2] KIENTZY,D.W., RICHARDSON,M.H.: Combined Testing and
Analysis on a PC using MSC/pal and STAR, Structural Measurement
Systems, Inc., San Jose, California MacNeal-Schwendler World
Users’ Conference March 23, 1988.
[3] FELIPPA,C.A.: Introduction to Finite Element Methods, Department
of Aerospace Engineering Sciences, September 2006.
[4] WILSON,E.L.: Numerical Methods in Offshore Engineering, John
Wiley & Sons, 1987.
[5] NX Nastran, Basic Dynamics Analysis User’s Guide
Figure 25. Mode 16 - Fin bending [6] RICHARDSON,M.H.: Measurement and Analysis of the Dynamics
of Mechanical Structures, Hewlett-Packard Conference for
Automotive and Related Industries, Detroit, MI, October 1978.
[7] PEETERS,B., CLIMENT,H., DE DIEGO,R., … :, Modern Solutions
for Ground Vibration Testing of Large Aircraft, IMAC 26, Orlando,
USA, February 2008.
[8] WILKINSON,J.H.: The Algebric Eigenvalue Problem, Clarendon
press, Oxford, 1965.
[9] OJALVO,I.U.: Proper Use of Lanczos Vectors for Large Eigenvalue
Problems, Computers & Structures, 1985, Vol.20, No.1-3.
[10] KOVAČEVIĆ,P.: Analiza dinamike strukture letelica metodom
konačnih elemenata, Vojnotehnički institut VJ, 1995.
[11] WEINGARTEN,V.I., RAMANATHAN, R.K., CHEN,C.N.: Lanczos
Eigenvalue Algorithm for Large Structures on a Minicomputer,
Computers & Structures, 1983, Vol.16, No.1-4.
[12] KOVAČEVIĆ,P.: Razmatranje uticaja statičkih aeroelastičnih
pojava i sopstvenih oblika oscilovanja na minimizaciju mase
Figure 26. Mode 17 - Wing symmetric torsion strukture uzgonskih površi, Univerzitet u Beogradu, Mašinski
fakultet, Beograd, 1989.
[13] HUGHES,T.J.R.: The Finite Element Method, Static and Dynamic
Conclusion Finite Element Analysis, Prentice Hall, 1987.
[14] DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: Računska modalna analiza
The results of the Computational Modal Analysis of the aviona LASTA, Vojnotehnički institut VS, Beograd, 2009.
LASTA aircraft are presented in this paper. [15] CRAIG,R.R., BAMPTON,M.C.C.: Coupling of Substructures for
All aircraft parts are modeled by finite elements and they Dynamic Analysis, AIAA Journal, Vol. 6, No. 7, July 1968.
are joined into the aircraft finite element model that has [16] FEDERAL AVIATION ADMINISTRATION: Advisory Circular,
220.536 degrees of freedom. The natural frequencies and Means of Compliance With Section 23.629-1B, Flutter, 2004.
the natural mode shapes of seventeen modes are obtained
by the Computational Modal Analysis and they are
presented in Table 1. All natural mode shapes are shown
from Fig.9 to Fig.16. Received: 03.03.2010.
The results of the Computational Modal Analysis for the
 DIMITRIJEVIĆ,J., KOVAČEVIĆ,P.: COMPUTATIONAL MODAL ANALYSIS OF THE LASTA AIRCRAFT 69

Računska modalna analiza aviona LASTA

U ovom radu su prikazani rezultati računske modalne analize kompletnog modela konačnih elemenata aviona
LASTA. Rezultati proračuna, sopstvene frekvencije i sopstveni oblici oscilovanja, koji su ovde prikazani, biće
korišćeni za dalju analizu aeroelastičnih pojava radi provere njegove bezbednosti od pojave flatera u okviru anvelope
upotrebnih brzina leta.

Ključne reči: avion, elisni avion, školski avion, sopstvene oscilacije, modalna analiza, numerička analiza, metoda
konačnih elemenata.

Расчётный модальный анализ самолёта ‫״‬Ласточка‫״‬

В настоящей работе показаны результаты расчётного модального анализа целостной модели конечных
элементов самолёта ‫״‬Ласточка‫״‬. Результаты расчёта, собственные частоты и собственные формы колебаний,
здесь показаные, будут использованы для дальнейшего анализа аэроупругих явлений с целью проверки её
надёжности из-за возникновения флаттера в границах диапазона употребительных скоростей полёта.

Kly~evwe slova: самолёт, винтовой самолёт, учебный самолёт, собственные колебания, modальный анализ,
цифровой анализ, метод конечных элементов.

Analyse modale de computation de l’avion LASTA

L’analyse de computation modale pour le modèle complet des éléments finis de l’avion LASTA est exposée dans ce
travail. Les résultats du calcul, de la fréquence naturelle et des formes de l’oscillation naturelle, décrits ici, seront
utilisés pour les analyses futures des phénomènes aériens élastiques pour confirmer que cet avion n’est pas soumis au
battement dans le cadre de la vitesse prévue du vol.

Mots clés: avion, avion à hélice, avion école, oscillations naturelles, analyse modale, analyse numérique, méthode des
éléments finis