Virtual Testing for Predicting Effect of Automated Fiber Placement Manufacturing Defects

Abstract

Actually aero-structures are preliminarily sized as designed and not as manufactured. Manufacturing issues and defects are mainly taken into account downstream in the development cycle when costly demonstrators are produced and analyzed. That approach leaves on one hand limited room to the designer to improve/modify the initial configuration and on the other hand it constrains manufacturing to respect strict engineering requirements introducing conservativeness. Automated manufacturing technologies as Automated Fiber Placement ensure high quality, repeatability and flexibility opening the way to new composite architectures and AFP advanced programming tools allow to simulate the as-manufactured parts to be accounted in preliminary sizing. However, unavoidable AFP singularities as gaps and overlaps together with fiber deviation respect to the nominal calculated orientation and their effects should be also managed in an early phase of the structure definition. The number of possible combinations between singularities is huge and limited physical test campaigns don’t permit to understand their influence (Knock Down Factors) and optimize their distribution. This will require increased predictive capabilities and efficient simulation methods to be validated for effect of AFP defect prediction. In this work, part of the IRT Saint-Exupery VITAL Project, a framework of virtual testing is developed combining advanced academic damage models, finite element models and surrogates for predicting the ‘’Effect of manufacturing defects’’ on the strength of composite laminates. Meso-scale damage models representing materials at ply level are developed by academic partner ONERA. A key aspect of this work is FE modelling of defects at coupon level, depending on the type of defect itself. Standard plain and open hole tensile tests with several combination of defects have been simulated using simple to complex FE models. Experimental data and tomography scans are referred for validation. Automatic generation of models contributes to the generation of a database and the modeling of a surrogate.