Inverse modelling of compression moulding of sheet moulding compound using CFD

Sheet Moulding Compound (SMC) is primary used in compression moulding which is a common manufacturing process for composite materials. When manufacturing SMC, voids are likely to appear on the surfaces of the parts causing large defects after painting. The automotive industry often use SMC in their applications implying that the parts often must have a class A appearance. When defects are discovered on the surface of the parts they need to be rejected or repaired which is a costly process. The formation of voids is strongly related to the spatial distribution of the pressure. It is therefore interesting to reveal the pressure distribution during moulding. If the pressure distribution is known critical areas can be localized. The purpose of the work is to investigate if an inverse modelling approach together with Computational Fluid Dynamics (CFD) can be used to predict the pressure distribution during compression moulding of SMC. The geometry in focus is a circular plate with a radius of 100 mm placed between two heated mould halves. To be able to find the pressure distribution the viscosity as a function of time and spatial coordinate needs to be known since it relates the deformation rate to stresses and hence the pressure. The viscosity of the SMC is known to be dependent on many parameters such as temperature, shear rate, degree of curing, amount of fill-material and fibre orientation. Here, the viscosity is assumed to be a function of temperature only since it is difficult to derive an explicit expression for the viscosity as a function of all the parameters mentioned above. A way forward is instead to do measurements of the pressure during moulding in a simple geometry and at the same time do numerical simulations that are fitted, by adjusting parameters in the viscosity model, to the experimental results by inverse modelling. The results from the experiments show a rapid increase in pressure initially. After the pressure has reached its peak value it decreases almost linearly with time. The same behaviour is captured by the viscosity model used in the numerical simulations. By an optimization procedure it is possible to find the parameters in the viscosity model that gives the best fit between the experiments and the simulations. The optimization is performed at one spatial coordinate and the resulting pressure is similar to the one obtained in the experiments. The pressure however differs when compared at an other spatial coordinate. Comparisons between visualizations made previously of the in-mould flow and the simulations show however similar flow behaviour. In other words, the numerical model serve as a good start but needs further improvements in order to capture the true mould flow behaviour.Validerat; 20101217 (root