Global behaviour of a composite stiffened panel in buckling. Part 2: Experimental investigation
Introduction
A representative test of a panel can be specified in terms of a compressive [1], [2] or shear loading state [3]. A pressurised test at the panel level is more difficult to assess since it needs a combined loading [4]. Several studies can be found concerning a compressive post-buckling test of co-cured or co-bonded stiffened panels, manufactured in prepregs. It is known that opposite buckling waves – symmetric or antisymmetric – are the areas, where delamination initiates [5], [6], leading to the loss of connection between skin and stiffeners [1]. In the case of excellent bonding, severe stress concentrations may cause delaminations and fracture in the skin or in the stiffener flanges at these locations [1].
This paper is the experimental investigation of a previous work which was the numerical modelling of this panel [7]. It aims to show the particular behaviour of a panel manufactured by a resin infusion process [8]. Numerical models consider the failure initiation anywhere through the thickness, rather than the damage initiates at the interface. A conventional testing machine can highlight the decohesion phenomenon in the skin–stiffener interface, by a uni-axial compressive test that makes the panel buckles. No strain gages are used because the displacement fields serve at correlating the numerical and experimental results through the use of stereo-DIC techniques, being tried and tested at the laboratory for more than 10 years.
Before the test, a newly made model that takes into account a layup modification of the stiffeners has led to redefine the load at collapse Fmax = 127.026 kN for a shortening UX = 2.40 mm. This model is defined as the reference model. The load at collapse will be used to set-up the test. Geometric imperfections of the skins are not considered since they have an almost negligible effect on the load at collapse [9]. The side stiffeners provide enough rigidity so that the panel borders do not buckle [7].
The critical buckling load Fcr (corresponding to skin local buckling) has been recalculated by the linear buckling analysis [7] on the reference model. It equals 90.4 kN, which is close to the end of the linear behaviour in the non linear analysis, where Fcr is approximately 90 kN. Now, numerical results are concordant with the buckling mechanism [10]: from the critical buckling load, the skin local buckling causes a load redistribution in the stiffeners until collapse, and the behaviour becomes nonlinear. Thus linear models cannot be used anymore.
These mechanisms given by numerical results need to be validated by experiment. First of all, the experimental test is set-up. The panel and the tool placement is carefully checked so that the panel is properly lined up with the load direction of the testing machine. Then numerical and experimental results are correlated through the out-of-plane displacement field (UZ) measured by two stereo-DIC measurement systems that are positioned on each side of the panel. X-ray CT and US techniques are used to analyse the failure zones on the panel surface and through the thickness of the stiffener flanges, allowing numerical and experimental results to be corroborated.
Section snippets
Experimental techniques
Experimental methods mainly use optical measurement systems from GOM® Software. Using Tritop, a photogrammetric software, distances are checked between primitive surfaces defined from reference points on the test panel and on the tool. ATOS is a digitising software that can complete Tritop information, it is used to make digitisation of the test panel, and more particularly of the resin blocks at both ends of the test panel.
Finally, ARAMIS is a measurement system more specifically dedicated to
Numerical and experimental correlation
The test is achieved after the loading on the testing machine has been projected (Table 1) from the reference model.
The real relative longitudinal displacement (shortening) is calculated by Aramis using two reference points located below the central stiffener on the skin side, at the intersection between the panel skin and the resin blocks. These reference points must be defined along the central stiffener for the shortening to be correlated to the numerical curve. Fig. 7 shows the experimental
Conclusions
Numerical models have correctly predicted the experimental post-buckling test of this composite fuselage representative panel. The numerical approach previously followed [7] is validated.
Two experimental load drops are observed. The first load drop corresponds to Ω1 collapse with a failure at its intersections, confirmed by tomographic measurements. No stiffener moves apart the skin at the first load drop.
The second load drop is the Ω2 and Ω3 delamination, initiated by crippling, also
References (12)
- et al.
Buckling and postbuckling of stringer stiffened fibre composite curved panels – tests and computations
Compos Struct
(2006) - et al.
Panel stiffener debonding analysis using a shell/3d modeling technique
Compos Sci Technol
(2009) - et al.
Buckling and postbuckling of composite structures
Composites
(1995) - et al.
Postbuckling behaviour of a blade-stiffened composite panel loaded in uniaxial compression
Compos Part A: Appl Sci Manuf
(2000) - et al.
Global behaviour of a composite stiffened panel in buckling. Part 1: Numerical modelling
Compos Struct
(2011) - et al.
Design and analysis of stiffened composite panels including post-buckling and collapse
Comput Struct
(2008)
Cited by (37)
Experimental and numerical study of bending-induced buckling of stiffened composite plate assemblies
2022, Composites Part B: EngineeringReview of digital image correlation application to large-scale composite structure testing
2021, Composite StructuresBuckling detection of an omega-stiffened aircraft composite panel using distributed fibre optic sensors
2018, Thin-Walled StructuresInvestigation on impact damage evolution under fatigue load and shear-after-impact-fatigue (SAIF) behaviors of stiffened composite panels
2017, International Journal of Fatigue