Modelling the Extrusion of a Hard-Metal Paste

Paste extrusion is a bulk forming technique used widely in the ceramics, food, metal and polymer industries, amongst others. This dissertation develops (i) measurements of the friction and bulk constitutive behaviour of pastes and (ii) numerical simulations of paste extrusion using these measurements as inputs. Two pastes were used: microcrystalline cellulose, calcium carbonate and water (MCC/CaCO3) and a commercial tungsten carbide-cobalt-based paste (WC-Co). The friction rig developed by Bryan, Rough and Wilson (2018) was used to determine the frictional behaviour of MCC/CaCO3 paste and to separate pressure- and velocity-dependence of the wall shear stress, O!. A new device with temperature control was constructed to test WC-Co paste. O! measurements were fitted to models proposed by Benbow and Bridgwater (1995): both pastes showed strong velocity dependence of O!, MCC/CaCO3 exhibited negligible pressure dependence, and WC-Co observable but small pressure dependence. For MCC/CaCO3 paste, these were compared to alternative friction measurements made using ram extrusion apparatus. Finite element simulations of extrusion were performed using ABAQUS® with an Arbitrary Lagrangian-Eulerian (ALE) formulation and custom friction subroutines. Steady-state velocity profiles and the pressure drop through square-entry geometries were predicted and extended to the extrusion of MCC/CaCO3 through conical-entry and 3D geometries, and WC-Co with a rate-dependent constitutive model. Manual adjustment of input parameters was required to obtain good agreement between experimental and simulated forces. The resulting MCC/CaCO3 parameters gave reasonable prediction of conical-entry experiments. Extrusion dies with ridges along their length were studied. An analytical model was developed using an upper bound method following the work of Khoddam et al. (2011a, b). Benbow-Bridgwater (1995) constitutive and friction parameters were used. Experiments were performed with 3D printed and machined dies for MCC/CaCO3 and WC-Co, respectively. For both pastes, the analytical model underpredicted the measured forces and parameter adjustment was required to improve the fit. This was attributed to greater surface roughness within the dies. A coupled Eulerian-Lagrangian (CEL) simulation was run in ABAQUS® with frictionless contact. This predicted comparable circumferential and vertical die land velocity components to the analytical model. Plastic deformation was observed along the die land, leading to slightly higher measured stresses and estimated pressure drops. The CEL method also led to a loss in paste-die wall contact and non-uniform paste volume fraction in the die land, creating difficulties testing frictional conditions