Logo Leibniz Universität Hannover
Logo Leibniz Universität Hannover
  • Zielgruppen
  • Suche
 

WP 4: Structural testing

In WP4, structural tests will be performed. The targets of WP4 are:

  • Experimentally evaluate the material damping and the structural damping properties
    for typical structural components
  • Validate models adopted during the simulations

WP leader: Haim Abramovich, Technion - Israel Institute of Technology

 

 

Figure 1: Stiffened flat panels after production
Figure 2a: NDT inspection of panels
Figure 2b: NDT inspection graph
Figure 3: Curing tool for cylindrical shell

WP 4 Results

Dynamic tests have been done within DAEDALOS on composite stiffened panels. The curved and flat panels were tested by BUT, POLIMI, RWTH and TECHNION, statically and dynamically until failure. Major conclusions from tests are summarized below.

 

BUT performed numerical and experimental dynamic analyses on flat composite panels delivered to BUT from AERNNOVA and LETOV. 

BUT performed FE simulations of buckling of carbon-reinforced composite panels using MSC.Patran/MSC.Dytran system. These simulations were done for two thicknesses of one prepreg layer, theoretic t = 0,125 [mm] and increased t = 0,132 [mm] (thickness measured on real panels). Other analyses were eigenfrequency analyses of panels in four configurations: Free panel, Fixed panel, Fixed panel with pre-loading 50 [kN] and Fixed panel with pre-loading 100 [kN]. 

Panels manufactured in accordance with drawing RD-002-2011 were prepared by AERNNOVA and LETOV. Panels underwent measurement of eigenfrequencies for following configurations: Free panel, Fixed panel, Fixed panel with pre-loading 50 [kN] and Fixed panel with pre-loading 100 [kN].  Dynamic buckling tests were performed for the panels. Forces at first buckling were measured using data from strain gauges. Deformation shapes were recorded by 3D ARAMIS system.  Results indicated that speed of loading (in the range of loading speeds considered – close to gust loading) does not have significant impact on force at first buckling.

 

POLIMI performed numerical and experimental dynamic analyses on the panels delivered to POLIMI from AAEM. Two types of investigations were performed: modal analysis and dynamic buckling analysis. Both of them were considered numerically and experimentally. The first one is mainly focused on the identification of modal damping to be used in further response analysis performed on structural components of the full aircraft. The second one is mainly directed to the identification of possible short duration dynamic loads able to activate the dynamic buckling phenomenon. 

Concerning modal analysis, the end tabs requested for load application and for the connection with the loading frame are not modelled in the finite element model. In this way, the mass difference between the numerical and experimental specimens is probably the main reason for which the numerical frequencies are higher than the experimental ones. For what concerns the measured modal damping, it must be pointed out that the variation along the applied load does not demonstrate any particular relation with the load itself. On the other hand, the limited scatter reported is due to the accuracy typically available using the standard modal identification tools.

Concerning the dynamic buckling, it must be underlined that while the numerical simulations were able to capture the phenomenon, the tests performed are not, due to the difficulty to obtain with the available equipment a very short excitation with the load level requested by the high strength of panels proposed by AAEM. Indeed, the MTS loading frame available at POLIMI allowed minimum load duration still in the range of static buckling.

 

RWTH results from the experimental modal analysis show the square of the natural frequencies for the first four vibration modes versus the in-plane compression load. There is a linear decrease of the square of the natural frequencies with increasing in-plane compression load. The variance of the experimental data with regard to the linear approximation (dashed lines) is small. The curves corresponding to the first three vibration modes show a similar slope, whereas the curve from the fourth mode is characterised by a higher gradient (Mode 4 is the first one with a shape consisting of two half-waves per panel length). Plus, the extension of the trendline corresponding to Mode 4 is the first one to intersect the horizontal axis, at a load of approximately 280 [kN]. 

The analysis of the frequency response functions measured shows that there is a good agreement between the natural frequencies predicted with Abaqus for the unloaded state and those gained experimentally. An overview of the natural frequencies for the first four vibration modes shows that the highest relative deviation corresponds to -3.7 %. Moreover, the good agreement between FE prediction and measurement  also holds for the mode shapes observed. 

In addition to the natural frequency and the mode shapes, the modal damping is determined from the frequency response functions. For every load step between 0 and 175 [kN], the maximum and minimum damping ratio is extracted. For the extraction of the modal damping ratios, LMS Test.Lab uses the half-power bandwidth method. This is a Single-Degree-of-Freedom based method which can be applied to Multiple-Degree-of-Freedom systems when the resonances or modes from the system are clearly separated from each other.  In the region between the maximum and minimum damping ratio measured and computed by Test.Lab  for the case of the modes 1, 2 and 4, the damping curves show a ‘hat-shaped’ progression for the compression loads considered (0-175 [kN]). For these three modes, the peaks in the frequency response functions are well pronounced and easily identifiable. In the case of  mode 3, and in particular for higher compression loads, the peak becomes indistinguisable in the neighborhood of mode 2 and 4. The half-power bandwidth method returns then values for the damping ratio which are too high.

 

The TECHNION results concern tests carried out for two panels.

Panel Dcom1: The first buckling load, as determined from the behavior of strains gauges was 16.19 [kN] while the collapse load was found to be 41.95 [kN]. The panel was excited using the NI impact hammer. One can notice a sharp bend in the graph of frequency squared vs. the axial compression load of the lower frequency f1 in the vicinity of the experimental first buckling load 16.19 [kN]. 

Panel DAL3: The first buckling load, which could not be determined by the strain gages readings as it was located at the upper right side of the panel, a region without strain gages, was 76 [kN] while the collapse load was found to be 100.7 [kN].  The straight edges of the panel were supported by vertical v-groove beam to simulate simply-supported boundary conditions. The panel was excited using the NI impact hammer. One can notice a sharp bend in the graph of frequency squared vs. the axial compression load of the lower frequency f1 in the vicinity of the experimental first buckling load 76 [kN].