Webinar – Elena Syerko – Towards robust and efficient numerical prediction of permeability of composite reinforcements
Elena SyerkoÉcole Centrale de Nantes, Nantes University, CNRS, GeMFrance
Permeability of preforms is a key parameter in Liquid Composite Molding (LCM) manufacturing processes. Its determination highly influences the entire chain of composites processing: from definition of process cycle times, injection strategies, to the prediction of defects in final parts. Experimental measurements of permeability are time-consuming and require specialized equipment. Furthermore, there is no experimental procedure to measure the permeability of the locally deformed due to draping fabric. This is, however, possible using a numerical computation approach. Hence, for the first time, an international benchmark exercise on the numerical prediction of permeability of fibrous reinforcements has been organized with the objectives to, firstly, identify possible sources of error of different methods used for computation, and to ultimately develop the guidelines for the robust prediction of permeability.
Under the name of Virtual Permeability Benchmark this benchmark is led by Ecole Centrale de Nantes (Nantes, France) together with Leibniz-Institut für Verbundwerkstoffe GmbH (Kaiserslautern, Germany) and with the support of KU Leuven (Belgium) and University of Delaware (USA). The concept of this benchmark study is to perform the computations based on real 3D images of textiles, which allows to treat an important feature of this class of materials – their high variability.
The first round of the benchmark consisted in the prediction of permeability at the scale of fibers. Although all participants were provided with the same fiber bundle image sample, the scatter of the computed results was between 14% and 24% since no conditions were imposed at this first stage in order to analyze the state-of-the-art of existing methods. This variation of results is nevertheless of the same order of magnitude as the results of the experimental permeability measurement benchmarks that had revealed at least 20% of variation. However, experimental measurement does not allow to precisely and efficiently analyze the influence on permeability of different structural parameters, such as, for example, fiber volume fraction. The experimental constraints motivated our developments of a scientific software for permeability computation named `PoroS’, which includes a set of numerical solvers specifically designed for such highly anisotropic and multi-scale materials as textile reinforcements.
The software calculates the saturated permeability based on real 3D images of the material micro/mesostructure. Alternatively, digital twins of the material can also be used as inputs to PoroS. The used full-field homogenization method allows the full 3D permeability tensor to be calculated using only the computed velocity fields, giving the advantage of reducing the number of degrees of freedom without having to compute the pressure field.
PoroS has participated in the first, as well as the second benchmark round. The focus of the latter was the permeability computation at the scale of a textile. The results obtained using PoroS for the first and second rounds fell within the main cluster of computed results of the benchmark. They also correlated well with the measured in the experimental benchmark values. The outcomes of the predictions by PoroS and of the benchmark studies should allow to gain insights into the robust and efficient numerical prediction of this key parameter of composites processing.
