Search

• A method for turbulent mass transfer solution using unclosed PDEs and sparse data. • Efficient concentration fields prediction under different boundary condition. • A deep learning method for handling observed data noise with sufficient accuracy. • Inverse computation of turbulent viscosity and mass diffusivity based on PINN. Turbulent mass transfer is widespread in chemical engineering processes including separation, reaction, and others. Traditionally, turbulent mass transfer processes can be simulated by computational fluid dynamics (CFD) methods. However, CFD methods require adequate turbulent models for the fluid flow and mass (and heat) transfer, which are normally difficult to develop for reliable solutions. Moreover, the CFD simulation is computationally intensive and hard to use in the cases like process optimization or analysis where the simulation needs to be called repeatedly. In this paper, physics-informed neural network (PINN) is developed and trained by unclosed mechanism model and sparse observation data of turbulent mass transfer process. The PINN method shows stronger generalization ability in solving the velocity, pressure and concentration fields in a turbulent mass transfer process than traditional deep neural network (DNN) method and turbulent Schmidt model. Under different boundary conditions, the PINN developed in the present paper can instantly predict the concentration distributions with sufficient accuracy and be used for inverse computation for estimating the turbulent viscosity and mass diffusivity as output results. The PINN is also capable of handling data noise by adjusting parameters, suggesting its potential in integrating experimental data. [ABSTRACT FROM AUTHOR]

[Display omitted] • The applicability of CFD-CGDEM in the simulation of packed bed is discussed. • The characteristics show dependence on scaled spring constant for CGDEM. • CGDEM with moderate scaling scheme is less dependent on spring constant. • The scale-up uncertainties for CFD-CGDEM increase with coarsen degree. The numerical modelling with computational fluid dynamics coupled with discrete element method (CFD-DEM) could be of high fidelity but rather time consuming, which leads to the development of coarse grain DEM (CFD-CGDEM). This paper describes the effects of scaling criteria for CFD-CGDEM on the modelling of multi-phase flow in packed bed. Four representative scaling criteria for particles contact interaction are discussed to analyze the packing structure and pressure drop and compare the relative deviation between CFD-CGDEM and CFD-DEM. It is found that when the spring constant in CFD-CGDEM is enlarged by α 3 or α 2 times or remains the same as that in original CFD-DEM, as α is defined as coarsen degree which is referred as the ratio between parcel size and particle size, a relatively large spring constant should be chosen to maintain the similar porosity and pressure drop. However, when the spring constant in CFD-CGDEM was scaled moderately by α times, the modelling results of CFD-CGDEM are comparable to data from CFD-DEM regardless of the value of spring constant. Furthermore, a quantitative analysis shows that the relative deviation in porosity and pressure drop between CFD-DEM and CFD-CGDEM will be enhanced as coarsen degree increases but the variation is insignificant with domain size for periodic boundary. The increase of parcel size could affect the simulation of porosity in packed bed and deviate the predicted hydrodynamics away from the CFD-DEM. In addition, the comparison of computational efficiency illustrates the speed up of CFD-CGDEM increases with the coarsen degree while the relative deviation also increases. This study provides instructive guidance for the application of CFD-CGDEM to model the multiphase flow in packed beds. [ABSTRACT FROM AUTHOR]

• A novel GPU-based DEM-CFD model was developed to simulate the gas–solid flow. • A grid-based approach was proposed to improve efficiency in simulating gas–solid systems with complex geometries. • The effects of tuyere angles on the raceway dynamics were analyzed. • The wear of tuyere for different tuyere angles was analyzed. The coupling of Computational Fluid Dynamics (CFD) and Discrete Element Model (DEM) is a powerful tool for simulating dense particulate systems, yet the conventional CFD-DEM has limits for systems with large particle numbers and complex geometry. This paper reports a novel GPU-based CFD-DEM model to simulate the gas–solid flow with large particle numbers and complex geometry. A novel coupling strategy between the CFD solver and DEM solver is developed, featuring high efficiency and stability. The developed model is validated against the experimental measurements, and its efficiency is compared to the previous CFD-DEM simulations. Then, for demonstration, the model is employed to simulate the dynamic behavior of gas–solid flow in the raceway in ironmaking blast furnaces by considering complex tuyere structure details and the huge particle numbers involved. This model allows to study the effect of the tuyere angle in terms of raceway formulation and tuyere erosion. The results show that the largest and most stable raceway volume can be reached at 5° downward tuyere, although the −5° tuyere nose experiences more wear than 10° downward tuyere. The model provides a cost-effective tool to overcome the longstanding challenge of simulating dense fluid-particle systems with huge particle numbers and complex geometry. [ABSTRACT FROM AUTHOR]

Emulsification carried out in continuous devices offers a series of advantages over batch emulsion, such as better control of the droplet size distribution, reduced volume equipment, and lower operative costs. This paper investigates through Computational Fluid Dynamics simulations the emulsification process inside Sulzer Static Mixers. An analysis aimed to identify the most appropriate turbulence model, from a practical point of view, was performed, finding that the realizable k − ϵ model is more suitable than the well-known k − ω model. Moreover, an operative correlation linking the Sauter diameter to the main operating parameters, in a wide range of fluid properties and operating conditions, was developed. • Emulsification processes were modelled through a CFD approach. • The minimum number of bins to have reliable results was found. • The accuracy of different turbulence models was assessed. • The range of application of the Luo and Lehr breakage kernels was studied. • A correlation valid for a huge range of conditions was found. [ABSTRACT FROM AUTHOR]

[Display omitted] • Simulations of granular collapse scenarios to explore the impact of pile shape on granular material flow. • Innovative CFD integration with experimental results: Achieving accurate 3D modelling of granular flows. • Practical applications: Risk assessment insights for natural disasters using μ(I)-rheology simulations. In recent years, significant progress has been made in modelling granular flows using the μ I -rheology model, which connects the viscosity of a granular medium to the pressure and strain rate via a dimensionless quantity called the inertial number, I. This model allows treating the granular material as a non-Newtonian liquid with a yield stress, making it possible to model the flow using the continuum approach, which is less computationally expensive than discrete element methods. In this paper, we implement the μ I -rheology model in a computational fluid dynamics (CFD) code and couple it with the volume of fluid (VOF) interface tracking approach to model the three-dimensional (3D) flow of monodisperse granular materials. After validating the model using experimental data, we briefly describe a trial-and-error method for evaluating the material properties of powders via a simple collapse experiment. Then, employing the CFD model, we investigate the physics of the unchannelized collapse of granular materials and perform an energy budget analysis to demonstrate the different stages of the granular collapse. To further investigate the effect of the initial shape of the pile on the spreading dynamics, we run a campaign of 3D simulations. Our results show that the μ I -rheology model accurately reproduces the dynamics of the granular material during the collapse and can be used for risk assessment purposes in natural disasters. The findings from our simulations can also aid in developing preventative measures to minimize potential harm. [ABSTRACT FROM AUTHOR]

[Display omitted] • A novel down-flow jet loop reactor with helical sieve tapes was proposed. • Hydrodynamic behaviors of this novel reactor were studied with detailed CFD simulations. • Effect of geometry of the helical sieve tape on overall gas holdup was evaluated. • Gas holdup, turbulent kinetic energy and radial mixing were significantly increased. • Optimal design parameters of a helical sieve tape were presented based on CFD simulation. Hydrodynamics and mixing in a novel down-flow jet loop reactor (DFJLR) with an internal helical sieve tape (HST) were studied by computational fluid dynamics (CFD) simulations. A helical sieve tape inside the draft tube caused a significant portion of the gas bubbles to flow helically, which increased radial mixing, overall gas holdup, axial liquid velocity, and turbulent kinetic energy. Compared with a similar reactor without a helical sieve tape, the gas holdup in reactors with a helical tape or a helical sieve tape inside the draft tube was increased, respectively, by 50 % and 75 %. A helical sieve tape with an area ratio of 27 % and helical pitch length of 492 mm was efficient in increasing gas holdup while keeping a lower pressure drop. Data and designs presented in this work will give insight and further understanding for reactor scale-up as a function of geometry and flow patterns for industrial applications. [ABSTRACT FROM AUTHOR]

• O 2 interfacial transport in production scale mammalian STBR is evaluated numerically. • Film theory in tandem with temperature-dependent Henry's constant is applied. • Predicted k L a values validated experimentally against production scale STBR data. • Optimal operational matrix to cultivate mammalian cells in STBR is presented. Computational fluid dynamics (CFD) is widely used for the prediction of hydrodynamics and mass transfer in a bioreactor. In this paper, a four-way coupled 3-D CFD, population balance model, and species transport model along with temperature-dependent Henry's law has been developed and applied to evaluate hydrodynamics, discrete phase distribution, and interphase O 2 transport in a 300 L STR equipped with discrete sparger for production of monoclonal antibody therapeutics. Momentum, mass, and energy equations have been solved. The temperature dependency of Henry's constant has been studied via coupling the van't Hoff's equation. The proposed coupled CFD model has been successfully validated against experimental k L a data. The optimal conditions for the bioreactor operation were identified as 200 RPM and 4 LPM. For k L a, the film theory (Ranz-Marshall) yielded the best prediction for k L a in the range of aeration and agitation rates examined. [ABSTRACT FROM AUTHOR]

[Display omitted] • Propane dehydrogenation reaction was analyzed with porous media CFD model. • Propane conversion was affected by non-uniform distributions. • Space velocity could influence the temperature uniformity. • The mixed packing can effectively suppress non-uniform distributions. Kinetic experiments can be significantly influenced by fluidic effects and transport limitations within reactors, particularly for strongly exo- and endo-thermic reactions. This paper describes the influence non-uniform physical fields, e.g. temperature, velcocity and species concentration, on the determination of kinetic parameters for endothermic systems with propane dehydrogenation (PDH) as a model process, which was analyzed using computational fluid dynamics (CFD) method based on the porous media model. Propane conversions calculated from 3-D porous media model at various conditions have discrepancies from the reported experimental data and 1-D numerical regression. This is due to the application of kinetic parameters obtained from 1-D plug flow assumption to the 3-D model with inhomogeneous temperature distributions. In the kinetic analysis, it is found that the low space velocity may lead to backward diffusion but can be a preference to minimize the temperature gradients within the catalyst bed. On the other hand, the commonly used higher space velocity may enlarge the temperature inhomogeneity. The sensitivity of temperature distribution with various reaction conditions is further analyzed. In addition, with respect to the packed bed properties, a high thermal conductivity and low loading of catalysts can reduce the non-uniformity of temperature distribution. Therefore, it is proposed that a mixed packing scheme with lower catalyst concentration and higher thermal conductivity could effectively suppress temperature gradients in kinetic analysis. This study exemplifies the necessity to consider the non-ideal distribution of physical fields for the kinetic study of reactions with strong thermal effects. [ABSTRACT FROM AUTHOR]

• Numerical evaluation of three channel configurations of a plate and shell heat exchangers. • Validated results for internal flow for various Reynolds number. • Analysis of pressure drop, friction factor and Nusselt number for three configurations. • Analysis of the correlations present in the literature for plate heat exchangers. The Plate and Shell Heat Exchanger (PSHE) is a variation of the Plate Heat Exchanger (PHE) with a more compact and robust geometry. This paper presents the analysis of the flow in the inner channel formed by a pair of plates of a PSHE employing computational fluid dynamics to determine the hydrodynamic and thermal characteristics for three Chevron angle configurations, 15° × 15° (High/High-HH), 45° × 45° (Low/Low-LL) and 15° × 45° (High/Low-HL). The ANSYS CFX software was employed with the application of the Standard k-ε and the SST (Shear-Stress-Transport) turbulence models for comparison. Experimental pressure drop data and flow streamlines were used to validate the numerical model and to analyze simplifications in the geometric model. Pressure drop, friction factor, and Nusselt number were evaluated for the three geometries studied. A local analysis of the parameters that describe the channels was also performed in this study. [ABSTRACT FROM AUTHOR]

• Discretisation method for the population balance is developed. • Method is mass conservative, applicable to arbitrary grid, efficient and accurate. • Applicable to any combination of aggregation, fragmentation, nucleation and growth. • Method is tested with reference solutions. • Very accurate results are obtained for both distribution and moments. In the present paper, we present a method for solving the population balance equation (PBE) with the complete range of kinetic processes included: namely aggregation, fragmentation, nucleation and growth. The method is based on the finite volume scheme and features guaranteed conservation of the first moment by construction, accurate prediction of the size distribution, applicability to an arbitrary non-uniform grid, robustness and computational efficiency which is instrumental for coupling with computational fluid dynamics (CFD). The treatment of aggregation is based on the previous work by Liu and Rigopoulos (2019). An analysis of the aggregation terms in the PBE is made, and the source of conservation error in finite element/volume methods is elucidated. It is subsequently shown how this error is overcome in the present method via a coordinate transformation applied to the aggregation birth double integral resulting from the application of the finite volume method. The contributions to the birth term are delineated and their corresponding death fluxes identified. An aggregation map is then constructed for mapping birth and death fluxes, thus allowing the finite volume method to operate in terms of fluxes and achieve conservation of mass. The method is then extended to fragmentation, for which a map is also constructed to represent the birth and death fluxes. In the implementation, the aggregation and fragmentation maps are pre-tabulated to allow fast computation. It is also shown how the method can be coupled with a total variation diminishing (TVD) scheme for the treatment of growth with minimal numerical diffusion. The method is validated with a number of test cases including analytical solutions and numerical solutions of the discrete PBE for aggregation (theoretical and free molecule/Brownian kernels), fragmentation, aggregation-fragmentation and aggregation-growth. In all cases, the method produces very accurate results, while also being computationally efficient due to the pre-tabulation of the maps and the simplicity of the algorithm carried out per time step. [ABSTRACT FROM AUTHOR]

[Display omitted] • Mixing in CIJ mixers using fluids with viscosity ratios from 2 to 9. • PLIF images and CFD simulations were used to characterize the flow. • Effective mixing occurs when the jets are in balance at the mixing chamber axis. • The necessary condition for mixing is the Reynolds number of the more viscous fluid. The industrial applications of Confined Impinging Jets (CIJ) mixers typically involve the mixing of two dissimilar fluids. Previous fundamental studies on the flow dynamics in CIJ mixers have only reported mixing fluids with a viscosity ratio up to 2 in mixers with similar nozzle diameters. In this paper, the onset of a self-sustainable chaotic flow regime was visualised for mixing of dissimilar fluids with a viscosity ratio range from 2 to 9 and for chambers with different injector diameters. Experimental and numerical results show that CIJ mixing chambers are able to efficiently mix fluids with different viscosities. The two sufficient conditions to effectively mix in a chaotic flow regime are the operation above the critical Reynolds number defined by the more viscous fluid R e M V , c > 150 and the balance of jets that makes them imping at the mixing chamber axis. [ABSTRACT FROM AUTHOR]

[Display omitted] • CFD was important to verify that the C-HSSFs and I-HSSFs presented a piston type flow without a short circuit and with a dead zone close to the base. • The dead zone volume found was not significant to suggest changes in the HSSF design. • Velocity profiles were similar in continuous and intermittent flow filters and did not influence the hydrodynamic behavior of HSSFs. • Operator filters under the same flow regime but with different diameters showed similar trajectories of velocity flow lines. This paper analysed geometries and flows in household slow sand filters (HSSFs) through computational fluid dynamics (CFD) to evaluate the hydrodynamics of filters. Four HSSFs in full scale were studied with a capacity to produce 48 L.d-1 each with diameters of 190 mm and 250 mm and with a filter medium depth of 0.5 m. The hydrodynamics of the four mathematical models of the HSSFs was validated using the experimental residence time distribution (RTD). The Kruskal Wallis non-parametric test indicates that the results of the experimental RTDs and the simulated RTDs did not show significant differences, therefore the mathematical models represent the physical models, which requires a detailed 3D analysis of the flow in the filters. The results showed that the HSSFs have a flow close to plug flow reactor. Internal recirculation zones were not found, nor short-circuit ones; however, dead zones were verified at the base of the filter with volumes below 3% when compared to the total volume of the filtering and draining layers, not necessary to make changes in geometry. The results of the computational simulation showed that the continuous filters had a smaller velocity variation and the filters with smaller diameter presented a reduction in dead zones when compared to the filters with larger diameters operated in the same flow regime. The study focused on the hydraulic aspects of HSSFs, but it is noteworthy that the choice of the type of operation to be adopted by families that use this type of treatment depends on studies that assess the efficiency of water treatment filters that were not considered in this work. [ABSTRACT FROM AUTHOR]

• Application of deep learning to predict the flow behavior of fluidized beds. • Introduction of a novel combination of encoder-decoder convolutional neural network and CFD. • The ability of combined deep learning- CFD approaches to predict the complex behavior of fluidized beds. In this paper, the capability of combined use of computational fluid dynamics (CFD) and data-based deep learning to predict fluidized beds' complex behavior without solving transport equations is being examined. A convolutional neural network (CNN) is trained to anticipate fluidized bed volume fraction contours based on the numerical simulations' results and data-based machine learning. The trained CNN receives the first ten frames from the CFD as input and predicts the next frame. This process continues until all the required frames are obtained. The results show CNN's superior spatial learning capability and how its combination with CFD can reduce the required computational power without compromising accuracy. [ABSTRACT FROM AUTHOR]

[Display omitted] • An analytical model for bed dimension of a FCVD process is proposed. • The proposed model defines relationships among process variables. • The proposed model can help manipulate process variables within their limits. • The proposed model can serve as a guideline for scale-up of a FCVD column. The present paper introduces a simple analytical model to predict the bed dimension of a continuous FBCVD (fluidized bed chemical vapour deposition) process. The expressions for column diameter and bed height were determined based on the target production rate, yield, conversion of the feed gas and gas volume change in the system. The model proposed in this article suggests that the diameter of an FBCVD column has to be reduced to enhance the conversion at a given production rate. To avoid linear expansion of the column diameter upon scale-up, the conversion or the superficial velocity of the feed gas has to increase, which results in an increase in the bed height. On the basis of the proposed model, a reasonable preliminary column dimension can be outlined when only a limited number of process parameters are available. CPFD (computational particle fluid dynamics) simulation results are presented to verify the model. [ABSTRACT FROM AUTHOR]