1 ECCM 2010 IV European Conference on Computational Mechanics Palais des Congrès, Paris, France, May 16-21, 2010 Towards an Algorithm for the Design of Robust Composite Structures M. Lee 1 , D. Kelly 1 and R. Degenhardt 2,3 1 School of Mechanical and Manufacturing Engineering, University of New South Wales, Sydney, Australia, merrill.lee@unsw.edu.au, d.kelly@unsw.edu.au 2 Institute of Composite Structures and Adaptive Systems, German Aerospace Center, Braunschweig, Germany, richard.degenhardt@dlr.de 3 Private University of Applied Sciences Göttingen, Stade, Germany The compression response of curved stiffened composite structures presents several challenges in both the postbuckling and collapse stages of response. In the completed European Commission 6th Framework Project COCOMAT [1] (Improved MATerial Exploitation at Safe Design of COmposite Airframe Structures by Accurate Simulation of COllapse), significant statistical variation in buckling behaviour and ultimate loading were encountered in the experimental work packages. The variations observed in the experimental results were not predicted in the finite element analyses that were done in the early stages of the project. It was recognised that there was a gap in knowledge about the effect of initial defects in the input variables of both the experimental and simulated panels. The initial defects observed included poor bond adhesion, quality of lamina layup, geometrical imperfections from the curing process, loading conditions, etc. A combination of these factors led to the varied postbuckling behaviour observed under experimental testing such as the collapse load and postbuckling mode shapes. Using stochastic analysis [2], where a variety of measured and assumed variations were introduced into the finite element models, the results observed in the experiments were captured (see figure 1). Figure 1: Out of plane displacement plots from finite elements and experiments At the completion of the COCOMAT project, a stochastic methodology to identify the impact of variation in input parameters on the response of stiffened composite panels and the development of Robust Indices to support the design of new panels was developed. The stochastic analysis included the generation of metamodels that allow quantification of the impact that the inputs have on the response using two first order variables, Influence and Sensitivity. These variables are then used to derive the Robust Indices [3] which measure the effect that defects and variations have on the response of structures. In Ref. [3], two stiffened composite fuselage panel designs were tested and the results were evaluated together with numerical models using finite elements. (a) Finite element simulations (b) Experimental results