Schedule (years) Design Optimization Approach for FML Wing Structure Background The aerospace industry is gaining significant interest in the application of Fiber Metal Laminates (FMLs) for fatigue sensitive parts, such as lower wing and fuselage structures. FMLs are well-known for their outstanding fatigue and damage tolerance (F&DT) behavior. To assess the F&DT criteria for design, reliable prediction methods are needed. Consequently, many studies have been performed to develop methods to predict, for example, crack initiation, crack propagation and residual strength. The context of this research is to develop an optimization methodology for FMLs which includes all available prediction methods. Progress The CI criteria is both physically reversed as well as by obtaining the alternative relationship using the black box approach. The Metal Volume Fraction (MVF) was found to be the linking parameter between the design life and lay- up. The latter step between the MVF and lay-up should still be done with GA. However, this process is much faster, since the original model is not repeated during the search algorithm. For CG criteria, only an alternative relationship is obtained by fixing input parameters, since it includes iteration loops and, therefore, physically reversing is not possible. The optimization framework based on GA is also build around existing CI and CG methods. An initial comparison is done with reversing the prediction model. For CG, large performance gain is expected, since the iteration loops are eliminated. The optimization procedure is extended from a coupon to a complete wing optimization for the crack initiation criteria. Publications I. Şen, G. Delgrange, R.C. Alderliesten, R. Benedictus (2013) “Design optimization approach for crack initiation in FML structures”, 17 th ICAF conference Aim The aim of this study is to develop a design optimization approach for FML wing structures, which will be used to obtain a optimized wing skin tailored for a given geometry where the F&DT criteria are not limiting while design and manufacturing constraints are satisfied. For this, I am adapting existing prediction methods to fit it in the optimization toolbox. This adaptation is either achieved by creating an alternative relationship between the lay-up parameters and the F&DT criteria or an optimization algorithm build around the prediction method to find the optimal FML lay-up. Optimization Procedure The optimization procedure is given in the flow chart below. PhD Candidate: Ilhan Şen Department: ASM Section: Structural Integrity & Composites Supervisor: R. C. Alderliesten Promoter: R. Benedictus Start date: Cooperation: Constellium Aerospace Engineering Methodology Basically, two different approaches are considered for including the design criteria into the optimization tool: 1)Reversing the prediction method The aim of this approach is to reverse the prediction model in such a way that a direct relationship is derived between the lay-up and the design constraints. The standards approach would be physically reversing the prediction methods by simply reversing and rewriting the equations. However, due to complexity of prediction methods this approach is not always possible. Therefore, an alternative relationship is needed that describe the design parameters as function of lay-up parameters, while the reversal of this relationship could be used to obtain the required FML lay- up. This desired relationship can be derived in three ways: -Iterative (brute force)  Database filling with ‘look-up’ function -Black box using machine learning algorithms (i.e. neural networks)  Find correlation between input and output variable of prediction method -Problem limitation  Break methodology in parts and obtain alternative relationship for non-reversible parts. The latter two ways are used to reverse crack initiation (CI), crack growth (CG) and residual strength (RS) criteria. 2)Optimization algorithms around prediction methods This approach aims to build an optimization framework around the prediction model and to find an optimal lay-up using Genetic Algorithm (GA). This approach would be the robust approach and will be compared with the previous approach with respect to performance and accuracy. Future Work More effort will be put on the compatibility between lay- ups obtained for neighboring segments. This should include checks for manufacturing and design challenges. The remaining fatigue criteria will be incorporated in the tool together with static strength criteria. Furthermore, a special attention will be given to the validity of existing prediction methods for thick metal layers which are required to quickly fill the total laminate thickness for ‘thicker’ lower wing panels. Therefore, experiments are planned to verify these thicker lay-ups. Wing Optimization The optimization procedure considering the crack initiation criteria is finished. As an example an preliminary simulation has been performed along the lower wing panel for Glare-2A as material with fixed metal (0.5 mm) and fiber (0.133 mm) layer thickness in order to see the required number of metal layer which satisfy the design loads along the wing. Property Analysis An alternative relationship is found by analyzing the fatigue life for different lay-ups. The idea was to find a single equation with the design life as function of laminate stress and parameters.