Variation Simulation Customized for Composites by Including FEM Simulations of Orthotropic Lamina
The use of composite materials is increasing in many fields of production. Composite manufacturing, as well as other fields of production, suffers from uncertainties resulting in products that deviate from the specification. Geometry assurance is a common procedure used to keep the variations in product assemblies under control. However, the methods used to simulate the variation are not developed for composites. In this thesis, two methods are presented that address typical uncertainties within composite production.
The method presented in Paper I focuses on the variation of fiber orientation and ply thickness within fibrous laminae. Variation simulation for the fiber orientation and ply thickness parameters is combined with a traditional such method. The combined variation simulation is carried out so that it is possible to study the effects of including perturbations in these composite parameters.
In Paper II, a method that captures a special type of deviation common for composites, called spring-in, is presented. These deviations are seen especially in T-beam structures and occur during the curing step of production, i.e., hardening in an oven. A FEM thermal expansion simulation is performed on the anisotropic composite laminate as a part of the traditional variation simulation method. The curing temperature is one parameter, along with the standard geometric parameters, within the proposed method.
The two methods proposed are tested on subassemblies originating from automotive and aviation industry, respectively. Applying the method presented in Paper I to the test case gives a resulting variation where the variance is increased by a factor of 10%. No structural differences are seen. Hence, these results indicate that traditional variation simulation is sufficient with the inclusion of a correction factor for composites. The method presented in Paper II is a new contribution to the field of geometry assurance. In addition, the results show an increase by a factor 4 in the resulting variation for the test case between keeping the curing temperature fixed at nominal value and letting it vary.
Method of Influence Coefficients