The high-pressure homogenizer (HPH) is, together with the rotor–stator mixer (RSM), the standard equipment for emulsification in many fields of chemical processing. Both give rise to intense turbulence which, in turn, gives rise to drop breakup. Previous investigations focus on average turbulent disruptive stress. However, turbulence is a stochastic phenomenon and drop breakup will be characterized by instantaneous stresses, or more specifically by the probability distribution of instantaneous turbulent stresses.
This study uses high-resolution particle image velocimetry (PIV) data to measure the probability distribution of turbulent stresses in the HPH. It is concluded that stress distributions approximately follow a lognormal model and that the skewness of the distributions decreases with increasing distance from the gap exit until a constant distribution shape is obtained at the position where the turbulence is fully developed. This converged stress distribution is similar to that obtained for RSMs in previous studies, suggesting that stress distribution shape is a general property. Moreover, large differences are observed when comparing these experimental stress distributions to the most widely used expression for describing this stochastic effect in fragmentation rate models. This indicates that the traditionally used fragmentation rate models can be fundamentally flawed, at least in relation to RSM and HPH emulsification.
Drop fragmentation in high intensity turbulent emulsification processing equipment-such as rotor-stator mixers (RSMs)-has traditionally been described in terms of a stress balance; between the stabilizing stress of the drop and the time-averaged turbulent stress at the most intense position of the flow. As shown in part 1 of this series, this approach is often a fruitful first approximation. However, the instantaneous local stress experienced by drops differs,from the time-averaged local stress due to hydrodynamics in general and the stochastic nature of a turbulent flow in particular. This study estimates the probability distribution of instantaneous turbulent stresses in an RSM from velocity fields obtained using particle image velocimetry. Results show that regions with low average stress still have a substantial probability of having instantaneously high stresses. This explains why low probability breakup is observed at these positions in visualization experiments. Results also show that the probability distribution of instantaneous stresses is approximately lognormal. The results are compared to two commonly used models for how to take the stochastic variations into account: the exponential decay model, and the multifractal emulsification model. It is concluded that both models predict reasonable distributions shapes but underestimate the width of the stress distribution.
Since the emulsification in the High-Pressure Homogenizer (HPH) is controlled by hydrodynamic forces, the turbulent flow field in the valve region is of significant interest. Computational Fluid Dynamics (CFD) simulations have been used for this in many studies. However, there are reasons to question if the utilized turbulence models, with their inherent assumptions and simplifications, could accurately describe the influential aspects of the flow.
This study compares CFD simulations using the methods from previous studies with experimental measurements in a model HPH valve. The results show that the region upstream of the gap can be described accurately regardless of turbulence model and that the gap region can be captured by using one of the more refined k−ε models. None of the studied turbulence models were able to describe the details of the highly turbulent region downstream of the gap. The obtained results are also discussed in relation to generalizability and limitations in using CFD simulations for understanding the emulsification in the HPH.
Particle image velocimetry is performed on a model of a high pressure homogenizer, scaled for qualitative similarity of the one phase turbulent flow field in a production scale homogenizer. Flow fields in gap entrance, gap and gap outlet chamber are obtained with high resolution. The measurements show gap flow development and formation of a turbulent wall adherent jet when exiting into the outlet chamber. Turbulent kinetic energy spectra show how the turbulent energy available for fragmentation is transported over distance along the jet centre axis.
The high resolution images are also used together with a Kolmogorov-Hinze theory framework for discussing drop fragmentation together with a direct evaluation of disruptive stresses from measurements. For the turbulent inertial mechanism large drops experience high fragmenting force close to eight gap heights downstream of the gap exit where as this occurs closer to 20 gap heights for smaller drops. The turbulent viscous mechanism is most efficient at a downstream distance of eight gap heights into the outlet chamber for all drops sizes.
The aim of this study was to find models for turbulent fragmenting forces in the high-pressure homogeniser from data available in Computational Fluid Dynamics (CFD) simulations with Reynolds Averaged Navier Stokes (RANS) turbulence models. In addition to the more common RANS k–ε turbulence models, a Multi-scale k–ε model was tested since experimental investigations of the geometry imply large differences in behaviour between turbulent eddies of different length-scales.
Empiric models for the driving hydrodynamic factors for turbulent fragmentation using the extra information given by multi-scale simulations were developed. These models are shown to give a more reasonable approximation of local fragmentation than models based on the previously used RANS k–ε models when comparing to hydrodynamic measurements in an experimental model.
The high-pressure homogenizer emulsification modelling framework by Håkansson et al. (2009a, Chemical Engineering Science 64, 2915–2925; 2009b. Food Hydrocolloids 23, 1177–1183), is further developed in this study. The model, including the simultaneous fragmentation of drops, coalescence of drops and kinetic adsorption of macromolecular emulsifiers, is improved with regard to two points. First, the transport of adsorbed emulsifier between drops of different sizes due to the fragmentation and coalescence of drops, is included using bivariate population balances. Second, the coupling of hydrodynamics to the emulsification model is improved using information from recent hydrodynamic investigations. The proposed framework is exemplified using a set of physically reasonable kernels and a production scale high-pressure homogenizer geometry, showing realistic emulsification results.
Despite large industrial relevance, the relation between rotor-stator geometry, hydrodynamics and drop breakup is poorly understood, partly since no methods for measuring the fragmenting stresses acting on drops have been established. This study attempts to bridge this gap by developing, applying and evaluating two approaches for estimating local turbulent stresses based on particle image velocimetry data: namely one traditional but indirect approach based on the dissipation rate of turbulent kinetic energy, and another more direct approach based on the spatial turbulent spectrum that has proven useful in other high-intensity emulsification processing. The approaches are evaluated in terms of validity of underlying assumptions, how they compare to breakup visualizations in the same geometry and with regard to the reliability of primary measurables. Results show three consistent regions of high stress in the rotor-stator region: in a plume extending into the stator-hole from the trailing edge, in the shear layers of the jet exiting the hole and in the macroscopic flow structure formed after the rotor blocks a stator hole. Following, a drop travelling along an average velocity flow field, the measurement predict disrupting stresses exceeding the stabilizing stress at the stator hole exit, at approximately the same position where drop breakup is observed in breakup visualizations. Both methods are therefore able to predict the most likely breakup positions. It is also concluded that both methods have limitations, and that average stress alone cannot describe all aspects of the fragmentation process in rotor-stator mixers. (C) 2017 Elsevier Ltd. All rights reserved.
A simulation model for emulsification in high pressure homogenization (HPH), based on a population balance approach, is developed assuming it to be controlled by three simultaneous processes; fragmentation, coalescence and adsorption of a macromolecular emulsifier. The aim is to investigate the implications of adding a set of models together; studying the effects of dynamics, size effects and process interactions.
For fragmentation, turbulent inertial and turbulent viscous forces are included using a dynamic model based on the Weber and Capillary number. It was extended to include a deformation time scale.
The rate of adsorption and coalescence is assumed to be controlled by the collision rate of macromolecular stabilizer and bare interface, modeled using convective and diffusive transport in turbulent flow.
By comparing simulation results to general trends found in the literature, it can be concluded that the models can reproduce the general HPH process well. By dividing the active region of emulsification in the homogenizing valve into discrete steps, the dynamic process could also be examined, indicating the homogenization process being composed of three stages with coalescence predominantly found in the last one.
Computational fluid dynamics (CFD) has been applied extensively for studying rotor-stator mixers (RSM) in the past, both as a design-tool and in modelling mixing and emulsification. Modelling is always a balance between accuracy and computational cost. The theoretically soundest methods (i.e. fully resolved transient simulations) have often been deemed unfeasible, and the majority of previously published studies use severe simplifications (i.e. k-ε models for turbulence and multiple reference frame for rotation). High quality experimental validation is in great need, but are rare, due to the lack of local fluid velocity measurement.
Experimental validations of CFD on RSMs have previously been provided using laser Doppler aneometry. This study provides the first validation using particle image velocimetry, allowing for substantially higher spatial resolution than with the previously used techniques. The objective of this study is to map the possibilities and limitations of these commonly used CFD modelling approaches for RSMs. Special emphasis is put on validating the dissiaption rate of turbulent kinetic energy (TKE). Despite being the parameter used for linking CFD to mixing or dispersion models, this has not been the subject of experimental validation in previous studies.
Based on the validations, a list of best practice recommendations are given (in terms of turbulence model, mesh resolution and rotation formulation). When adhering to these, the CFD model accurately captures power draw, flow number, and the detailed velocity field inside the region where mixing and dispersion takes place. The dissipation rate of TKE is captured qualitatively but underestimate experimental values. Implications in terms of limitations are discussed in detail, including estimations of accuracy implications for emulsification and mixing modelling.