(505f) Development of a Simulation Method to Study Liquid Phase Maldistribution During the Squeeze Flow of Highly Filled Particulate Pastes

Particulate pastes are widely used to manufacture products such as foods, agrochemicals, pharmaceuticals, ceramics, and solder pastes using common processing techniques such as ram extrusion and injection moulding. These materials are often classified as soft solids and feature a high volume fraction of particulates in a liquid binder. Their rheology is therefore strongly influenced by interparticle contacts, rather than collisions, and exhibits complex yield stress behaviour and wall slip. Furthermore, several types of paste-specific flaw may develop during processing, as discussed by Benbow and Bridgwater (1993, p. 21).

One such flaw is liquid phase maldistribution (or migration), henceforth abbreviated to LPM. When stresses are applied to a paste during ram extrusion, a direct consequence of the multi-phase nature of pastes is that the applied stress is split between the load-bearing particulate network (the solids skeleton) and the liquid present in the gaps (pores) between the particles. These pores form an interconnected network that is permeable to the liquid. If the applied stress varies spatially across the paste, pore liquid pressure gradients develop within the pore network and drive flow of the liquid relative to the solids skeleton. LPM will thus result in an inhomogeneous distribution of liquid. Since the rheological properties of the paste are strongly dependent by liquid content, flow patterns are modified and dead zones may result. In extreme cases of LPM, flow will cease completely, causing damage to the extruder and loss of production.

Detailed and reliable simulation of paste extrusion processes is required to improve design of extrusion dies but can also be used to study flaws such as LPM. The aim of the latter simulations is to link the severity of LPM to the operating conditions and the composition (or formulation) of the paste: the onset of LPM occurs when the relative velocity of the local pore liquid is large compared to the absolute velocity of the solids, so that the liquid has time to redistribute while the paste passes through a high stress region, i.e. the die entry zone in a ram extruder. The relative velocity of the liquid phase to the solids skeleton is a function of the (local) pore liquid pressure gradient and the permeability of the solids skeleton. The latter parameter is determined by the properties of the particles (absolute particle size, size distribution, shape, roughness and deformability), the skeleton (solids volume fraction), liquid (rheology, liquid volume and air volume fractions for unsaturated pastes), and chemical interactions between the solids and the liquid.

Models for paste flow incorporating LPM do exist for ram extrusion, e.g. Rough et al. (2002), and for squeeze flow, e.g. (Poitou and Racineux 2001), Sherwood (2002), Kolenda et al. (2003) and Roussel et al. (2003). However these models are usually one-dimensional. Two-dimensional modelling approaches are well established in the field of soil mechanics and the challenge is to apply these approaches to simulate paste extrusion processes where the strains are much larger than normally arise in soil problems. The squeeze flow test is a relatively simple experimental procedure that shows promise in measuring some of the mechanical constitutive parameters of pastes that are required to calibrate soil mechanics models.

This work investigates the accuracy of this procedure for a model paste and discusses the physical and computational difficulties that arise. The Drucker-Prager constitutive relationship was used to describe the solids skeleton during squeeze flow simulations of a saturated viscoplastic paste comprising glass ballotini and an aqueous glucose liquid binder. The liquid is modelled as Newtonian and incompressible, and Darcy's law is used to represent the interactions between the two phases. The interaction between the paste and the plates was assumed to be frictionless. These simulations were implemented in a two-dimensional axisymmetric finite element model using the ABAQUS/Standard FEM package (Simulia, Rhode Island, USA). A series of simulations have been performed in order to investigate the effect of solids loading and squeeze flow velocity (deformation rate) on paste performance during squeeze flow.

Benbow, J. J. and J. Bridgwater (1993). Paste Flow and Extrusion. Oxford, Clarendon Press: 19-21.

Kolenda, F., P. Retana, et al. (2003). "Identification of rheological parameters by the squeezing test." Powder Technology 130(1-3): 56-62.

Poitou, A. and G. Racineux (2001). "A squeezing experiment showing binder migration in concentrated suspensions." Journal of Rheology 45(3): 609-626.

Rough, S. L., D. I. Wilson, et al. (2002). "A model describing liquid phase migration within an extruding microcrystalline cellulose paste." Chemical Engineering Research and Design 80(7): 701-714.

Roussel, N., C. Lanos, et al. (2003). "Induced heterogeneity in saturated flowing granular media." Powder Technology 138(1): 68-72.

Sherwood, J. D. (2002). "Liquid-solid relative motion during squeeze flow of pastes." Journal of Non-Newtonian Fluid Mechanics 104(1): 1-32.