An axisymmetric model for the meat analog production process in a Couette Cell based device

Abstract

During the last years the demand for vegetarian products has increased. A subgroup of these vegetarian products consists of meat analogs, which are products that resemble meat in its functionality and are prepared in a similar fashion. One of the companies producing these meat analogs is Rival Foods. Rival Foods has been working on a production process based on a Couette cell, in which the dough is sheared in an annular region between two concentric cylinders. This allows them to create highly fibrous products with a thickness of roughly 3cm. Other production processes such as extrusion cooking are unable to achieve this combination of structure and thickness. In upscaling the production process, preferably a larger gap width between the cylindrical surfaces of the cell is preferred, because the thickness of the product is a unique selling point. Larger gap widths lead to greater temperature inhomogeneity and gradients, which negatively impacts product quality. It is currently not possible to accurately measure the temperature profile throughout the cell. Therefore, in this thesis a model has been developed in OpenFOAM which calculates the temperature profile with a small number of material parameters and process conditions as input. The model assumes the ingredient mixture behaves as a temperature dependent power law fluid. Rheological measurements have been performed to quantify these temperature dependent power law parameters. To study the influence of the viscous dissipation, preheat temperature, mixture density, and product thickness on the temperature field during processing, multiple simulations have been performed. The simulations used a time step of 0.0001s, for which the temperature, velocity, viscosity, and viscous dissipation were not yet fully converged. The principal flow in the Couette cell geometry was in the direction of rotation of the inner cylinder as expected and had a velocity with order of magnitude e-01 m/s. Besides the principal flow a secondary flow pattern has been found as well, consisting of vortices which had a velocity with order of magnitude e-03 m/s. These secondary velocity components were responsible for additional advection of heat in the simulation which resulted in a different temperature profile than expected. Since the Taylor number was well below the critical Taylor number, the existence of Taylor vortices could be excluded. After a further refinement of the time step the vortices disappeared. Running the full simulations with this time step would take months per simulation and was therefore not an option. It was concluded OpenFOAM had trouble simulating power law fluids with high viscosities.