Effects of varying fibre distribution in closed injection pultrusion modelling
Topic(s) : Manufacturing
Co-authors :
Nik POPPE (GERMANY), Georg ZEEB , Florian WITTEMANN (GERMANY), Michael WILHELM (GERMANY), Luise KÄRGER (GERMANY)Abstract :
Pultrusion is a continuous and highly automated process for manufacturing fibre-reinforced profiles of constant cross section. Pultruded profiles exhibit high specific mechanical properties. In closed injection pultrusion, the impregnation of fibres takes place in an almost closed compartment, the so-called injection and impregnation chamber (ii-box). For a given profile cross section, the design of the ii-box has to ensure complete impregnation of the rovings. Process simulations can give insight into the ii-box, from which recommendations for tool design can be derived. The present work underlines the importance of precise fibre distribution modelling in impregnation simulation.
The impregnation process can be modelled by the laws of fluid dynamics, namely Navier-Stokes equation and Darcy’s law for flow in porous media. The presented simulation method applies these fundamental laws within the software OpenFOAM 6, which is based on the Finite Volume Method [1]. The rovings are homogenised over volume, represented by a field of anisotropic permeability tensors [2].
Experimental trials for several cross sections have shown that the core of the profile cross section is the most challenging part. E.g., a rectangular profile cross section reveals a dry spot in the centre [3]. Simulations of the injection chamber with homogeneous and static fibre distribution don’t show a corresponding impregnation behaviour. Such simulations with a constant anisotropic permeability always lead to complete impregnation, also at the profile centre.
The ii-box is equipped with one injection point, located at the upper wall. Assuming that the whole cross section is homogeneously filled by fibres, the region with the lowest z-coordinates is impregnated by resin flowing from the injection point in negative z-direction (Fig. 1). Hence the centre should be impregnated before the flow front reaches the bottom face in this configuration. Since this is not the case in the mentioned trials, it can be concluded that the assumption of a homogeneously filled cross section is not met by processing conditions.
Inhomogeneities in fibre distribution may arise from capillary adhesion [4], uneven fibre infeed and the geometry of the ii-box, or be induced by the resin flow [5]. The specific fibre distribution in the ii-box is a priori unknown. Within this work, one hypothetical pattern of inhomogeneity is investigated: a surface fluid layer next to the cavity walls all around the fibre stack. It is shown how such a surface layer impacts the impregnation behaviour. The described fluid layer creates a main flow path around the fibres circumventing the centre. Thus the impregnation occurs from the periphery towards the centre, not in negative z-direction as before. This results in dry spots, though small, in the centre and matches qualitatively the occurrence of the real process (Fig. 2).
The impregnation process can be modelled by the laws of fluid dynamics, namely Navier-Stokes equation and Darcy’s law for flow in porous media. The presented simulation method applies these fundamental laws within the software OpenFOAM 6, which is based on the Finite Volume Method [1]. The rovings are homogenised over volume, represented by a field of anisotropic permeability tensors [2].
Experimental trials for several cross sections have shown that the core of the profile cross section is the most challenging part. E.g., a rectangular profile cross section reveals a dry spot in the centre [3]. Simulations of the injection chamber with homogeneous and static fibre distribution don’t show a corresponding impregnation behaviour. Such simulations with a constant anisotropic permeability always lead to complete impregnation, also at the profile centre.
The ii-box is equipped with one injection point, located at the upper wall. Assuming that the whole cross section is homogeneously filled by fibres, the region with the lowest z-coordinates is impregnated by resin flowing from the injection point in negative z-direction (Fig. 1). Hence the centre should be impregnated before the flow front reaches the bottom face in this configuration. Since this is not the case in the mentioned trials, it can be concluded that the assumption of a homogeneously filled cross section is not met by processing conditions.
Inhomogeneities in fibre distribution may arise from capillary adhesion [4], uneven fibre infeed and the geometry of the ii-box, or be induced by the resin flow [5]. The specific fibre distribution in the ii-box is a priori unknown. Within this work, one hypothetical pattern of inhomogeneity is investigated: a surface fluid layer next to the cavity walls all around the fibre stack. It is shown how such a surface layer impacts the impregnation behaviour. The described fluid layer creates a main flow path around the fibres circumventing the centre. Thus the impregnation occurs from the periphery towards the centre, not in negative z-direction as before. This results in dry spots, though small, in the centre and matches qualitatively the occurrence of the real process (Fig. 2).