The aim of this research project is to improve the description of thick-walled filament-wound lightweight structures by taking into account the 3D stress state of the textile reinforcement structure, in particular by mapping the material-processstructure interactions. The description of these interactions is based on a machine-learning approach (GEP-FC: Gene Expression Programming with Free Coefficients), which allows simultaneous symbolic and numerical regression analyses. Based on these advanced models, rule sets are derived and implemented into an efficient holistic optimization strategy.
An important focus for the extension of existing modelling methods is the exact representation of the filament winding process under consideration of relevant technological and material interrelationships of the fibre-plastic structure, caused, among other things, by local changes of single layer parameters (thicknesses, orientations, direction-dependent stiffness and strengths), in order to be able to more accurately represent the structural-mechanical behaviour of filament-wound high-pressure vessels with regard to an improved failure prediction.
Although the models derived in this way describe the structural behavior with sufficient accuracy, they are too inefficient for direct structural optimization, primarily due to the expected non-convex objective function, which also covers a strength evaluation of various failure modes. For a structural optimization, therefore, corresponding model reductions have to be carried out and basic rule sets have to be derived, on the basis of which an optimization of the textile reinforcement structure with regard to various objectives (mass reduction, manufacturing specifications, resources) of the high-pressure vessel is carried out. With the help of such a design strategy, the enormous number of variants can be managed and the complex dependencies of the relevant parameters can be taken into account.
Within the project, extensive experimental investigations as well as numerical and analytical simulations and multi-criteria optimizations using stochastic and deterministic methods are carried out.
In addition to a better understanding of the relationships between material, process and structural parameters of filament-wound fibre-reinforced plastic composites, this research project makes a significant contribution to the integrated efficient design of continuous fibre-reinforced high-performance components. Fundamental interactions are being investigated, e.g. for the safe high-pressure storage of hydrogen, the results of which can be implemented in further industrial applications. Furthermore, the modelling and optimisation methods developed can be applied to related manufacturing processes and provide impulses for adjacent research fields of lightweight construction, simulation and production technology.