Search

The hydrodynamics of the High Pressure Homogenizer has been investigated in order to increase the understanding of the emulsification process. Fragmentation of drops in the homogenizer is generally assumed to be caused by cavitation and/or turbulence. Both processes are investigated experimentally in order to find its location in the HPH valve region. Cavitation was visualized by investigating light scattered in a HPH valve model with optical access. Detailed measurements of the one phase flow turbulence were obtained by Particle Image Velocimetry (PIV) on a carefully scaled model. The effects of dispersed phase flow volume fraction on the continuous phase turbulence was studied with refractive index matched PIV. The experiments show cavitation being focused in the first half of the gap whereas turbulence intensities are very low inside the gap. The turbulence is most effective in the outlet chamber downstream of the narrow gap. This is even more evident for the turbulent eddies of sizes comparable to the drops that are well known to be most efficient for the fragmentation. Increasing the disperse phase volume fraction does not alter the conclusion of highest turbulence downstream in the gap, however, it leads to an increase in energy of large turbulent eddies and a decrease in the energy of small sale eddies. This would imply a relative increase of the strength of the turbulent viscous mechanism compared to the turbulent inertial mechanism when increasing the volume fraction of dispersed phase. When comparing these findings on with visualizations of drop break-up; turbulence rather than cavitation seems to be the dominant mechanism of fragmentation in these geometries.

Effects of various enzymes and extraction conditions on yield and molecular weight of ß-glucans extracted from two batches of commercial oat bran produced in Sweden are reported. Hot-water extraction with a thermostable a-amylase resulted in an extraction yield of ˜76% of the ß-glucans, while the high peak molecular weight was maintained (1.6 × 106). A subsequent protein hydrolysis significantly reduced the peak molecular weight of ß-glucans (by pancreatin to 908 × 103and by papain to 56 × 103). These results suggest that the protein hydrolyzing enzymes may not be pure enough for purifying ß-glucans. The isolation scheme consisted of removal of lipids with ethanol extraction, enzymatic digestion of starch with a-amylase, enzymatic digestion of protein using protease, centrifugation to remove insoluble material, removal of low molecular weight components using dialysis, precipitation of ß-glucans with ethanol, and air-drying.

Large eddy simulations (LES) of a stirred reactor with a liquid volume of 0.64 m3 were performed. The tank was stirred by either two standard Rushton impellers or two Scaba 6SRGT impellers. The LES can provide details of the flow field that cannot be obtained with so‐called Reynolds‐averaged equations and the corresponding models. The unsteady spatially filtered incompressible Navier‐Stokes equations were solved together with a transport equation for the concentration of an inert additive using a multigrid finite difference code with a staggered grid. The unresolved turbulent scales were modeled using a scale similarity model. The aim of this investigation has been to study the influence of impeller type on the flow structure and scalar transport. The results show that, compared at equal power input, the center plane velocities and the volume flow in the impeller stream do not differ much for the impeller types. Also, for the 6SRGT the periodic fluctuations associated with the blade passing is less pronounced.

The deposition of polystyrene latex particles of 0.46 μm diameter was studied in situusing a wall-jet cell in combination with total internal reflection microscopy. The particles were deposited onto an indium tin oxide surface in a laminar flow field for up to 13 h. The initial particle flux was found to be mass-transfer-controlled. It was shown that one period of the time-dependent deposition rate period was of first-order nature. The effective particle transfer coefficient during this period appeared to be correlated inversely to the wall shear rate. With the help of three characteristic empirical constants, one of them being a mass transfer coefficient, the overall deposition process was described by a model equation. The concentration dependency was elucidated using a Langmuir-type pseudoisotherm.

A method of treating data acquired with the planar laser‐induced fluorescence technique has been developed to visualize the topology of two‐dimensional concentration fields and to describe the dynamics of the coherent mixing structures identified. This method is based on a conditional binary transformation of the local concentration data, combined with a joint probability calculation. The methodology has been used to investigate the mixing in a stirred tank, at two injection port locations (in the bulk and in the impeller stream region). With bulk injection, a “folding phenomenon” of the coherent mixing structure was detected. Away from this port, large‐scale spatially periodic motion was identified, with a characteristic time of oscillation of the order of 2 to 3 s. With injection in the impeller stream region, no spatial instabilities of the coherent structure were detected. Local oscillations of the coherent mixing structure were found both on short and long time‐scales (i.e., ˜1 and ˜80 s).

Concentration gradients in industrial fed‐batch bioreactors due to imperfect mixing and biological consumption of nutrients result in a drastic decline in performance. Three advanced methods for the quantitative description of the mixing process in stirred, aerated tanks of industrial size are presented and commented on in terms of adequacy and reliability with regard to the microenvironmental approach, which gives biological and engineering factors the same importance.

Mathematical and numerical modeling of the complex interplay between compaction, flow, and hydrodynamic dispersion in a chromatographic column under steady state conditions has been examined. The mathematical model was based on pure elastic analysis and Darcy flow. A two-dimensional control volume approach was applied to solve the coupled nonlinear equations describing the flow and compaction. The two-dimensional equation describing the hydrodynamic dispersion was solved using an explicit control volume algorithm. Numerical diffusion was eliminated by using a window of moving points following the concentration front.

The shear viscosity of three doughs of different wheat cultivars mixed to a farinograph level of 500 BU was measured at low shear rates as a function of the shear deformation using a cone-and-plate viscometer. Cyanoacrylate adhesive was used to attach the dough samples to the instrument surfaces to eliminate wall slip. Flours used were Dragon, Kosack, and a fodder wheat. A distinct difference was observed between the viscosities of the different flour cultivars. The strongest dough (Dragon), with the highest protein content and a good resistance in the farinograph, had the highest maximum viscosity. The doughs showed distinct strain hardening, more pronounced for the strong doughs. Maximum viscosity was obtained at a strain of ˜4, almost independent of the shear rate, but at higher values for stronger doughs (5 for Dragon, 4 for Kosack, and 3.5 for fodder wheat). The maximum was most pronounced for well-mixed doughs after resting. The viscosity and its variation with strain may be used as a measure of quality; a higher viscosity and a maximum occurring at high strains indicating good quality (related to the farinogram). The viscosity gradually decreased at higher strains. Apparent viscosity increases with strain and reaches a maximum value at a common strain, which suggests the presence of entangled molecules. The increase of maximum viscosity with increase in mixing also supports this theory. Resting the dough increases the maximum viscosity, which suggests the formation of new cross-links in the nonequilibrium entangled network during resting.

The effect of adsorbed whey protein, β-lactoglobulin, has been investigated on the attachment of polystyrene latex particles to an indium tin oxide (ITO) surface and the subsequent release in anionic surfactant SDS solution and distilled–deionized water at pH 6.0. Experiments were carried out using a wall-jet flow cell and particle attachment was measuredin situusing the technique of evanescent wave microscopy. The deposition rate of particles increased as predicted up to a shear rate of ∼1000 s−1, for deposition at a diffusion-limited rate. There was a reduction in the rate at higher shear rates indicating a decrease in sticking efficiency. As the shear rate increased, the ITO surface became saturated more quickly due predominantly to blocking of the surface by deposited particles. The presence of adsorbed β-lactoglobulin on the ITO surface caused a large reduction in the subsequent deposition rate of protein-coated particles. This was due to an increase in electrostatic repulsion. Differences were found in both the extent of removal and in the release (cleaning) kinetics of particles in SDS and in distilled–deionized water for the different particle–protein–ITO surface conditions investigated. Release of particles was also independent of the shear rate. Results were interpreted by considering the roles of protein replacement and elution which occurs in SDS solutions; >90% removal of protein-coated particles from a coated ITO surface was observed in SDS when both processes play a role. This compared to 55% removal in distilled–deionized water where they were considered negligible.