Probabilistic process simulation to predict process induced distortions of a composite frame

Abstract

Within this work, an enhanced simulation strategy for predicting process-induced distortions and residual stresses of composite structures is developed, while taking into account probabilistic variations. The simulation strategy is demonstrated for a carbon fibre-reinforced frame of an aircraft fuselage. Based on a sequentially coupled temperature-displacement analysis, temperature and degree of cure distributions as well as stresses, strains, and distortions are determined. For the description of the reaction kinetic, a phenomenological-based model considering chemical and diffusion-controlled reactions is introduced. The mechanical material behavior is described by a viscoelastic material model depending on temperature, time, and degree of cure. Corresponding process-induced distortions and occurring residual stresses are calculated and analyzed. The results of the process-induced distortions are used to derive a compensated tooling design. To analyze the effect of statistical and systematic deviations of underlying process and material parameters, a procedure for a probabilistic process simulation is created. Based on a first sensitivity analysis, a design of experiment is performed over a specified parameter range. From this surrogate, models are derived to enable a full but efficient probabilistic analysis of spring-in angles and residual stresses at multiple frame positions. The effect of process parameter variations and fluctuation of inherent material properties on process-induced distortions is assessed. Moreover, the influence of different process cycles is shown. This allows to determine the expected range of the final part distortion and to derive the resulting tolerances during assembly. For validation of the probabilistic process simulation, several frames are manufactured at different process conditions, and frame geometries are measured by digital image correlation technique. The resulting part thicknesses, spring-in angles, and global frame distortions are determined. Finally, a comparison of measurements and probabilistic predictions is given, and results are evaluated with respect to main sensitivities.