Half Wing N219 Aircraft Model Clean Configuration for Flutter Test On Low Speed Wind Tunnel

Flutter is a rapid self-feeding motion which is caused by the interaction of aerodynamic, structural and inertial forces. Flutter can cause major damage on aircraft structure which can lead to fatal accident in aviation. Several methods have been evolved to avoid the flutter phenomena occur during the flight envelope of aircraft design. On this study, method was developed by Indonesian Aerospace which consist of Finite Element Method (FEM) analysis, Ground Vibration Test (GVT), and Wind Tunnel Flutter Test (WTT). Based on the study, FEM have similar results toward to Wind Tunnel Flutter Test conjunction the clean configuration of N219 aircraft half wing model.


INTRODUCTION
Flutter is a rapid self-feeding motion which is caused by the interaction of aerodynamic, structural and inertial forces. Flutter can cause major damage on aircraft structure which can lead to fatal accident in aviation. The aims of this research are to validate and compare the results of wind tunnel test to Finite Element Method (FEM) and to earn the critical speed on flutter phenomena based on wind tunnel test. This research was conducted to avoid and ensure the flutter phenomena will not occur when flight test of aircraft is held.
In other paper, flutter suppression has been spotlighted because of its destructive nature (Westin et al, 2009). The materials use is suppressed, especially PZT materials. Instead of the passive suppression will increase the mass of aircraft. The result of the test is to measure of flexible wing design which is suppressed on materials choice. Piezoelectric transduction is used by active and passive flutter suppressions for highly flexible wing (Tsushima et al, 2017). By properly place the piezoelectric actuators and energy harvesters, it was possible to stabilize the wing, while extracting a certain amount energy, both of which would contribute to improving the performance of wing and aircraft.
Furthermore, the methods of flutter analysis of aeroelastic system includes modelling uncertainties is more efficient (Lokatt, 2017). Both of structural and aerodynamic uncertainties can have notable effect on the damping of the flutter modes. The method shows that main advantage is the ability to analyse the combine effect of structural and aerodynamic uncertainties. Scope: This paper consists of four sections. Section 1 presents scope of work, objectives as introduction. Section 2 briefly describes procedure of the Wind Tunnel Flutter and FEM analysis result. Section 3 2 1234567890 ''""

FEM Analysis
Based on method which is described in flow chart below on Figure 1 and then updated with GVT, mode shape and its natural frequency final design half wing model that is mentioned on Table 1.   Then, half wing flutter model will be tested on wind tunnel. The model attached on wind tunnel wall using moderate spring and soft spring. Therefore, spring shall be modeled on FEM and tested on GVT in order to get better accuracy.

Ground vibration test (GVT)
Ground Vibration Test (GVT) is useful for validating dynamic characteristics of test model from analytical model. The purpose of this GVT is to get structural modes parameters such as natural frequency, mode shapes, and damping. Those modes parameters from GVT result, will be used on flutter analysis and dynamic aeroelastomechanics system responses.
The main instruments that will be used for GVT exciter are accelerometer, and frequency analyzer as showed in Figure 2. The model should have fix supported at the wing root (Fixed on Tx,Ty, Tz, Rx, Ry & Rz). T is torsion which react to external force on model and R is translation force which react to external force on model that both of them is placed on wing root. While x, y, z are the direction based on axis. The excitation method that will be used to produce mechanical vibration on test model is a random excitation ranged up to 50 Hz. Responses on the observation points will be measured using transducer accelerometers. The signal coming from accelerometers and force transducer will amplified using amplifier and continue to be processed on the analyzer for getting its frequency response function. With curve fitting method then the modal parameters of the half wing flutter model can be achieved. The listing of Detail Accelerometer Locations as showed in Table 2.   [11]. The GVT was performed in one mass configurations that the mass configurations are described in Table 3.
The GVT result will be use for updating the structural dynamic model of the half wing flutter test model design, furthermore the updated structural dynamic model will be use for updating the flutter analysis result. Thus, the flutter speed reference will be use to compare wind tunnel flutter test result with flutter speed from updated flutter analysis result.

Finite Element Model of Half Wing Flutter Model Test N219
According to the boundary condition (support system) of the tested model, the analysis FEM for GVT needs to be modified to represent the support system. These support systems are connected to Beam combination Spring. Figure 3 is show the FEM of the Half Flutter Model for GVT.

Figure 3 FEM Half Wing Flutter Model Test N219
The analysis FEM that is optimized and updated is the structure FEM described in Ref. 4. The updating is based on the GVT results described in Ref. 7. It must be remembered that frequencies and modes from GVT results actually include some damping of the tested Model.
The natural frequencies and modes of the structural FEM are obtained by solving the real 'undamped' eigenvalue equation (1). Where: [M] = generalized mass matrix size (ndof x ndof) [K] = generalized stiffness matrix size (ndof x ndof) A method that is used to establish the degree of correlation between the real eigenvectors from test and from the analysis model is visual examination by plotting and animation of the test and analytical modes. The test modes are related to the analytical modes that correspond.
The visual method evaluates the quality of modes correlation. In this document, the grades of modes correlation are divided into three categories. The first category is 'good', this category explains similar mode shape between test and analysis FEM. The second category is 'fair', this one declares the dominant modes are similar. Finally the third category is 'poor', this one affirms weak or no corresponding between the test and analytical FEM.
In In a simplistic sense the Bayesian parameter estimation procedure has some similarity to the minimization of error by the method of least squares. The former permits a balanced model updating approach by incorporating uncertainties in both test data and in modeling parameters. In this approach each 'piece' of test data has a confidence associated with it. Then the test data as a whole is weighed relative to the initial model as a whole. The Bayesian parameter estimation procedure is an iterative approach wherein as error is minimized. The error is written as: Where: E = a measure of the error between test data and analysis RT = target response quantities (i. e. test data) RA = responses from the analysis WR = weighting factors (confidences) for response PF = parameters for the final model PO = parameters of the original model WP = weighting factors (confidences) for the parameter wt = scalar weighting for the test data as a whole wp = scalar weighting for the model parameter as a whole The above equation is solved iteratively to find PF, the final model parameter that minimize the error.
Model response and test data can be static displacements, resonant frequencies and mode shapes, and element forces, stresses, and strains any quantity that can be measured or computed with MSC/NASTRAN. Model parameters consist of beam areas and inertias card entries. Weighting matrices are usually diagonal, and are scaled such that they are divided by the square of the initial value. The units for all computations are therefore, fractional changes rather than absolute changes to avoid numerical problem caused by a wide variance of units. The GVT results are considered as the baseline standard data. This means that the GVT data are assumed correct a priori. The relative error is defined as the discrepancy between the corresponding measures frequencies of GVT and the computed frequencies of the structural FEM. The minimum of the relative error in the corresponding frequencies of GVT and structural FEM implies the optimization of the member properties corresponding to the selected design variable in the optimization process. Considering the selection criteria described above, these design variable are the member properties I1, I2, and J of the equivalent beam replacements of the structural FEM. Recall the optimization process to minimize the relative error using MSC/NASTRAN SOL 200.

Wind Tunnel Flutter Test
The test model is valid to use on wind tunnel flutter test after confirmed through GVT and validation tests. Likewise the wind tunnel itself must be already fulfilled the wind tunnel safety requirement. The wind tunnel model flutter test will be carried out by sweeping the speed which started from the lowest speed up to flutter speed (if possible). The execution of test will be conducted into two steps sweep speeds are as follows:  The first step is to obtain preliminary data and as an exercise test procedure (Familiarization) prior to testing at higher and more critical speed (close to flutter speed).The speed sweepsof test will be conducted gradually started from The half wing model was placed in wind tunnel then the accelerometer will received response from the exciter. The response will be read by data acquisition system then forwarded to extract parameter software to determine flutter margin. Figure 5 is the model installation in test section of wind tunnel. The flutter margin cannot be resulted online due to delay of data transfer from data acquisition system to the extract parameter, therefore flutter speed from FEM analysis was used as speed limitation. The strip chart also cannot be used due to technical error in the software.
Data acquisition system forwarded data response to be analyzed by extract parameter software. The extract parameter was generated by Matlab coding will result speed, frequency, and damping data by curve fitting process. The extract parameter data for soft spring and moderate spring as showed in Figure 6.    There are 86% of the test modes that correspond to the analytical FEM Update modes.

RESULTS
The Figure7 and Figure 8 are the results of the NASTRAN calculation of half wing model of N219 aircraft.  Based on data result from extract parameter there was coupling between bending mode and torsion mode on moderate spring model, shown in Figure 9. During a coupling the damping value approached zero on moderate spring model, it can be seen in Figure 10.   For the soft spring model case, the data that was sent by data acquisition system cannot be processed properly in extract parameter due to spring property. The spring was very responsive therefore other frequency appeared.
When wind speed reached 24 m/s, half wing model vibrated without excitation, it showed that half model wing already unstable. Frequency of bending and torsion mode were difficult to be extracted thus frequency of coupling mode also hard to be determined. Figure 11 and Figure 12 are shows the frequency and damping of bending and torsion mode at soft spring model. Based on the V-f curve and V-g curve, there can be seen that there is an existence of coupling mode at speed of 24.0 m/s, yet it does not indicate the phenomenon of flutter. The noise made it difficult to extract the data, therefore this coupling mode was not necessarily caused by the frequency of bending or torsion mode. In order for the data extraction process to be more convenient, it is recommended that the data is filtered beforehand.

CONCLUSION
Wind tunnel flutter test represents the real flight condition, therefore its result was used to validate Finite Element Model analysis result. Comparison between FEM analysis and wind tunnel flutter test show at Table 6. In moderate spring case, flutter speed resulted by FEM analysis and wind tunnel flutter test showed nearly the same result. In soft spring case, FEM analysis resulted flutter speed at 26.75 m/s however in wind tunnel flutter test at speed 24.0 m/s the half wing structure already unstable. According to the comparison between FEM analysis and wind tunnel flutter test, can be concluded that the methodology of FEM analysis has been correctly validated.