Uncertainty analysis of variable stiffness laminates
CovenanteePolitecnico di Torino
Document typeMaster thesis
Rights accessOpen Access
Variable stiffness composites are fiber reinforced components manufactured by means of automated fiber placement technique. The fiber orientation angle varies within the tow path, providing the variable stiffness properties of the manufactured components. These non-conventional parts have shown better mechanical properties per unit-weight than many conventionally manufactured components. However, its fabrication process is subjected to imperfections which are prone to generate defects and undesired variability in the parts, decreasing the mechanical performance of the components. In order to avoid material waste or the application of very large safety coefficients, it is of great importance to understand the effect of manufacturing induced defects and uncertainty in variable stiffness composites. In this work, a multi-scale sensitivity analysis is performed, in which the buckling performance at a macro-scale level is studied for composite plates affected by uncertainty effects at a micro-scale level. For the study, the macro and micro-stress fields are obtained for the static solution of two variable stiffness composites. Finite element models and theories within the Carrera Unified Formulation framework are employed in the investigation. A layer-wise approach is applied in order to model each composite layer separately, introducing uncertainty effects in the ply level, at the meso-scale. Stochastic fields are generated to introduce variability into the fiber volume fraction at the micro-scale. This parameter affects the material properties assigned to the finite element mesh elements and are obtained from the unit cell problem resolution applying the mechanics of structure genome. The sensitivity study is performed via Monte Carlo analyses, carrying out many deterministic linearized buckling simulations in which the fiber volume fraction value within the plies is defined by a different random field. Distribution plots of the buckling analysis outputs are obtained and statistical indicators are calculated. At the conclusion of the analysis, polynomial chaos expansion models are proposed as a time saving alternative to the full Monte Carlo analysis.
SubjectsMaterials -- Mechanical properties -- Mathematical models, Laminated fabrics -- Analysis, Materials -- Propietats mecàniques -- Models matemàtics, Teixits i tèxtils plastificats -- Anàlisi
DegreeMÀSTER UNIVERSITARI EN ENGINYERIA INDUSTRIAL (Pla 2014)