Effect of steering limit constraints on the performance of variable stiffness laminates

• Published in: Computers & structures Vol. 196; pp. 94 - 111
• Main Authors: Peeters, Daniël M.J., Lozano, Gustavo Gonzalez, Abdalla, Mostafa M.
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 01.02.2018; Elsevier BV
• Subjects: Composite materials; Fiber placement; Finite element method; Laminates; Lamination parameters; Manufacturability; Manufacturing; Optimisation; Optimization; Retrieval; Stacking sequence (composite materials); Stiffness; Variable stiffness; Variable stiffness; Manufacturability; Optimisation; Lamination parameters
• ISSN: 0045-7949; 1879-2243
• DOI: 10.1016/j.compstruc.2017.11.002

•Optimise variable stiffness laminates starting from optimal stiffness distribution in a computational efficient way.•Post-process result to account for maximum size of gaps and overlaps.•New fibre angle retrieval step combined with fibre angle optimisation. A method to optimise the fibre angle distribution of variable stiffness laminates is proposed. The proposed method integrates a fibre angle retrieval step with a fibre angle optimisation procedure. A multi-level approximation approach is used in combination with the method of successive approximations. First, fibre angle retrieval is done by approximating the structural responses based on the optimal stiffness distribution found using lamination parameters. The full fibre angle optimisation is done by updating the approximations based on the current stacking sequence. Next, the actual fibre paths are optimised taking into account the actual size of a tow, and the maximum size of any gap or overlap appearing. The paths are smoothed out using CATIA, and finally spline curves are found that can be sent to a fibre placement machine for manufacturing. It is shown for a bucking optimisation with a stiffness constraint that the number of finite element analyses reduces significantly by starting the optimisation from the optimal stiffness distribution rather than from a user-specified stacking sequence. Next, it is shown that updating the approximations also leads to considerable improvements over fibre angle retrieval. Similar promising results are obtained for a stress optimisation problem.