Friction factor improved correlations for laminar and turbulent gas–liquid flow in horizontal pipelines

We develop improved correlations for two-phase flow friction factor that consider the effect of the relative velocity of the phases, based on a database that includes 2560 gas–liquid flow experiments in horizontal pipes. The database includes a wide range of operational conditions and fluid properties for two-phase friction factor correlations. We classify the experiments by liquid holdup ranges to obtain composite analytical expressions for two-phase friction factor vs. the Reynolds number by fitting logistic dose curves to the experimental data with. We compute the liquid holdup values used to classify the experimental data using correlations proposed previously. The Reynolds number is based on the mixture velocity and the liquid kinematic viscosity. The Fanning friction factor for gas–liquid is defined in term of the mixture velocity and density. Additionally, we sort the experimental data by flow regime and obtain the two-phase friction factor improved correlations for dispersed bubble, slug, stratified and annular flow for different holdup ranges. We report error estimates for the predicted vs. measured friction factor together with standard deviation for each correlation. The accuracy of the correlations developed in this study is compared with that of other 21 correlations and models widely available in the specialized literature. Since different authors use different definitions for friction factors and Reynolds numbers, we present comparisons of the predicted pressure drop for each and every data point in the database. In most cases our correlations predict the pressure drop with much greater accuracy than those presented by previous authors.     

Gas—liquid annular flow under microgravity conditions: a temporal linear stability study

A temporal linear stability study was performed for a gas—liquid annular flow configuration under microgravity conditions. Data used to validate the modeling includes that generated by Texas A&M as well as all the other known data in two-phase flow under reduced gravity conditions. Following a discussion of theoretical considerations on the growth rates of different instabilities, it is shown that given the fluid properties, pipe diameter and phasic flow rates, one can predict with a high level of confidence the flow regime in the pipe. Acceptable confidence levels (80%) are achieved when one differentiates between slug, slug—annular, and annular flow. Higher confidence levels (90%) are found when one differentiates between slug and annular flow by merging the annular and slug—annular categories.     

A separated flow model for predicting two-phase pressure drop and evaporative heat transfer for vertical annular flow

A separated flow model has been developed that is applicable to vertical annular two-phase flow in the purely convective heat transfer regime. Conservation of mass, momentum, and energy are used to solve for the liquid film thickness, pressure drop, and heat transfer coefficient. Closure relationships are specified for the interfacial friction factor, liquid film eddy-viscosity, turbulent Prandtl number, and entrainment rate. Although separated flow models have been reported previously, their use has been limited, because they were tested over a limited range of flow and thermal conditions. The unique feature of this model is that it has been tested and calibrated against a vast array of two-phase pressure drop and heat transfer data, which include upflow, downflow, and microgravity flow conditions. The agreements between the measured and predicted pressure drops and heat transfer coefficients are, on average, better or comparable to the most reliable empirical correlations. This separated flow model is demonstrated to be a reliable and practical predictive tool for computing two-phase pressure drop and heat transfer rates. All of the datasets have been obtained from the open literature.     

Spectrum of density fluctuations in a particlefluid system—II. Polydisperse spheres

This paper presents an extension of the analysis shown in Part I to a polydisperse particle-fluid system. The density autocorrelation is shown to be a function of two quantities, a generalized Overlap function for which an analytical expression is derived, and the radial distribution function (RDF). In Fourier transform space, the density spectrum again appears to be a strong function of the mean particle size, and secondarily the mean particle separation distance. One unusual result is previously observed oscillations in the density spectrum of a monodisperse system of particles are severely dampened or even eliminated in the polydisperse case, depending on the width of the particle size distribution. Apparently contributions from different particle correlations interfere with each other, thereby reducing the coherent oscillations seen in the monodisperse particle-fluid system. Furthermore at large wavenumbers, the spectrum decays with a −2 power-law, independent of the shape of the particle size distribution. This behavior can be traced to the Overlap function which controls the behavior of the spectrum beyond the first peak. Remarkably the −2 power-law spectrum is determined by the shape of the particles (i.e. spheres) rather than their spatial distribution (RDF).

The effect of an asymptotically large pressure gradient on the correlation of several important higher-order moments is revisited for the polydisperse system. The relatively simple relationships developed for the monodisperse system are lost in the polydisperse case because particles of different sizes will be influenced differently by an applied pressure gradient. The result is moments that are of different order in velocity can no longer be related to each other (as they were in the monodisperse system), even in this idealized flow. A more comprehensive understanding of this phenomenon can only be achieved through direct numerical simulation or experiment.     

Spectrum of density fluctuations in a particle-fluid system— i. monodisperse spheres

A method for calculating the density autocorrelation ′(x)′(x + r) for a homogeneous particle-fluid system in both physical and Fourier transform space has been developed. The density autocorrelation was related to two quantities, the Overlap function which is defined as the volume of intersection of two spheres as a function of the separation distance and the radial distribution function (RDF) of the particles. In dimensionless co-ordinates, the parameter that characterizes the density autocorrelation is the volume fraction of particles, ₁, , or equivalently the dimensionless mean separation distance (normalized by the particle diameter), . For an isotropic randomly distributed system of particles, the density autocorrelation was observed to oscillate with the correlation distance r, with a wavelength that was proportional to λ. The Fourier transform of the autocorrelation likewise oscillated with the wavenumber k, however the effect of changes in the particle volume fraction was limited to the first peak only. Subsequent peaks were more closely associated with the Overlap function.

The results for the density autocorrelation were extended to a particle-fluid system which experienced an asymptotically large pressure gradient. This initially produced a uniform relative motion between the two fields. In this limit, other higher-order moments such as the Reynolds stress can be related to the density autocorrelation in a straightforward manner. Moreover the spectral shapes of all moments collapse onto the density autocorrelation spectrum in this limit. It was pointed out that the uniform relative motion will eventually become unstable because of hydrodynamic forces on the particles induced by the relative motion. This effect was estimated by introducing a mildly attractive force into the RDF. The results demonstrated that the induced hydrodynamic force promoted a shift in the density spectrum toward small k (large scale) indicating an alternative mechanism for growth in the integral length scale.     

Comparison of spectral and finite element methods applied to the study of the core‐annular flow in an undulating tube

A Galerkin/finite element and a pseudo‐spectral method, in conjunction with the primitive (velocity‐pressure) and streamfunction‐vorticity formulations, are tested for solving the two‐phase flow in a tube, which has a periodically varying, circular cross section. Two immiscible, incompressible, Newtonian fluids are arranged so that one of them is around the axis of the tube (core fluid) and the other one surrounds it (annular fluid). The physical and flow parameters are such that the interface between the two fluids remains continuous and single‐valued. This arrangement is usually referred to as Core‐Annular flow. A non‐orthogonal mapping is used to transform the uneven tube shape and the unknown, time dependent interface to fixed, cylindrical surfaces. With both methods and formulations, steady states are calculated first using the Newton–Raphson method. The most dangerous eigenvalues of the related linear stability problem are calculated using the Arnoldi method, and dynamic simulations are carried out using the implicit Euler method. It is shown that with a smooth tube shape the pseudo‐spectral method exhibits exponential convergence, whereas the finite element method exhibits algebraic convergence, albeit of higher order than expected from the relevant theory. Thus the former method, especially when coupled with the streamfunction‐vorticity formulation, is much more efficient. The finite element method becomes more advantageous when the tube shape contains a cusp, in which case the convergence rate of the pseudo‐spectral method deteriorates exhibiting algebraic convergence with the number of the axial spectral modes, whereas the convergence rate of the finite element method remains unaffected. Copyright © 2002 John Wiley & Sons, Ltd.     

Use of lagrangian methods to describe drop deposition and distribution in horizontal gas—liquid annular flows

Drop distribution and deposition in horizontal gas—liquid annular flow is described by a diffusion model, which views the concentration field as the result of dispersion from a distribution of sources. Drops originating from a wall source are considered to diffuse in a field of homogeneous turbulence, while simultaneously being swept downward by the gravitational field. Deposition is assumed to be controlled by two mechanisms operating in parallel, and boundary conditions are derived which correctly satisfy conservation of mass. This analysis for an instantaneous source is shown to be equivalent to considering diffusion in a coordinate system moving with the settling velocity of the particles. The results are found to be useful for understanding droplet distribution and deposition.     

Analysis of high speed non‐equilibrium chemically reacting gas flows. Part I. Physical modeling and sensitivity studies

In this part, a theoretical model for high speed flow of chemically reacting gases out of thermal and chemical equilibrium is presented. The main features of the physical model are discussed together with details for a new form of the kinetic rate coefficients for non‐equilibrium flows and presentation of a two‐layer radiation model used for a plasma torch problem. This model is implemented in a new hybrid finite volume/finite element scheme, which is developed in Part II. Results from this physical model are compared with experiments and other results in the literature for an arcjet and non‐equilibrium nozzle test case. Sensitivity studies are included for the nozzle problem to simulate the influence of the rate coefficients. Copyright © 2000 John Wiley & Sons, Ltd.     

Injection and Storage of CO<Subscript>2</Subscript> in Deep Saline Aquifers: Analytical Solution for CO<Subscript>2</Subscript> Plume Evolution During Injection

Injection of fluids into deep saline aquifers is practiced in several industrial activities, and is being considered as part of a possible mitigation strategy to reduce anthropogenic emissions of carbon dioxide into the atmosphere. Injection of CO₂ into deep saline aquifers involves CO₂ as a supercritical fluid that is less dense and less viscous than the resident formation water. These fluid properties lead to gravity override and possible viscous fingering. With relatively mild assumptions regarding fluid properties and displacement patterns, an analytical solution may be derived to describe the space–time evolution of the CO₂ plume. The solution uses arguments of energy minimization, and reduces to a simple radial form of the Buckley–Leverett solution for conditions of viscous domination. In order to test the applicability of the analytical solution to the CO₂ injection problem, we consider a wide range of subsurface conditions, characteristic of sedimentary basins around the world, that are expected to apply to possible CO₂ injection scenarios. For comparison, we run numerical simulations with an industry standard simulator, and show that the new analytical solution matches a full numerical solution for the entire range of CO₂ injection scenarios considered. The analytical solution provides a tool to estimate practical quantities associated with CO₂ injection, including maximum spatial extent of a plume and the shape of the overriding less-dense CO₂ front.     

Liquid single- and gas—liquid two-phase flows of magnetic fluid in the presence of a high non-uniform parallel magnetic field

The effect of a non-uniform parallel high magnetic field on flow control characteristics is investigated experimentally for a magnetic fluid single-phase flow and an air—magnetic fluid two-phase flow in a vertical channel. It is found that as the magnetic field strength is increased, the friction factor of the single-phase flow increases significantly. For the two-phase flow, the friction pressure loss and the head pressure loss, which is smaller than the friction loss, are negligibly small compared with the magnetic pressure loss. In the case where air is injected 27.9d upstream from the maximum magnetic field, the air flow is blocked by the magnetic force in the entrance of the magnetic field, which leads to increases in both local void fraction and pressure drop there. In the case where air is injected 1.43d downstream from the maximum magnetic field, the air flow is accelerated, resulting in a decrease in void fraction and an increase in pressure rise. In the latter case and under the present range of experimental conditions, the magnetic pumping head reaches 0.02 MPa at the highest, and the maximum circulation flow rate reaches twice as high as non-magnetically driven flow rate.     

An investigation of void fraction in liquid slugs for horizontal and inclined gas—liquid pipe flow

The void fraction in liquid slugs has been determined for air—water fiow in horizontal and near-horizontal pipes by a newly-developed conductance probe technique. A semi-empirical correlation has been developed and compared with the present measurements and available data. This correlation predicts reasonably well the observed effects of diameter, inclination and physical properties.     

Heat transfer measurements and correlations for air–water flow of different flow patterns in a horizontal pipe

Heat transfer coefficients were measured and new correlations were developed for two-phase, two-component (air and water) heat transfer in a horizontal pipe for different flow patterns. Flow patterns were observed in a transparent circular pipe using an air–water mixture. Visual identification of the flow patterns was supplemented with photographic data, and the results were plotted on the flow regime map proposed by Taitel and Dukler and agreed quite well with each other. A two-phase heat transfer experimental setup was built for this study and a total of 150 two-phase heat transfer data with different flow patterns were obtained under a uniform wall heat flux boundary condition. For these data, the superficial Reynolds number ranged from 640 to 35,500 for the liquid and from 540 to 21,200 for the gas. Our previously developed robust two-phase heat transfer correlation for a vertical pipe with modified constants predicted the horizontal pipe air–water heat transfer experimental data with very good accuracy. Overall the proposed correlations predicted the data with a mean deviation of 1.0% and an rms deviation of 12%.     

Nucleation and flashing in nozzles—2. Comparison with experiments using a five-equation model for vapor void development

A quasi-one-dimensional, five-equation, homogeneous, nonequilibrium model has been developed and utilized on a microcomputer to calculate the behavior of flowing, initially subcooled, flashing water systems. Equations for mixture and vapor mass conservation, mixture momentum conservation, liquid energy conservation and bubble transport were discretized and linearized semi-implicitly, and solved using a successive iteration Newton method. Closure was obtained through simple constitutive equations for friction and spherical bubble growth, and a new nucleation model for wall nucleation in small nozzles combined with an existing model for bulk nucleation in large geometries to obtain the thermal nonequilibrium between phases. The model described was applied to choked nozzle flow with subcooled water inlets based on specified inlet conditions of pressure and temperature, and vanishing inlet void fraction and bubble number density. Good qualitative and quantitative agreement with the experiment confirms the adequacy of the nucleation models in determining both the initial size and number density of nuclei, and indicates that mechanical nonequilibrium between phases is not an important factor in these flows. It is shown that bulk nucleation becomes important as the volume-to-surface ratio of the geometry is increased.     

A consistent splitting scheme for unsteady incompressible viscous flows I. Dirichlet boundary condition and applications

A well‐recognized approach for handling the incompressibility constraint by operating directly on the discretized Navier–Stokes equations is used to obtain the decoupling of the pressure from the velocity field. By following the current developments by Guermond and Shen, the possibilities of obtaining accurate pressure and reducing boundary‐layer effect for the pressure are analysed. The present study mainly reports the numerical solutions of an unsteady Navier–Stokes problem based on the so‐called consistent splitting scheme (J. Comput. Phys. 2003; 192 :262–276). At the same time the Dirichlet boundary value conditions are considered. The accuracy of the method is carefully examined against the exact solution for an unsteady flow physics problem in a simply connected domain. The effectiveness is illustrated viz. several computations of 2D double lid‐driven cavity problems. Copyright © 2005 John Wiley & Sons, Ltd.     

Multi-scale HPC system for multi-scale discrete simulation—Development and application of a supercomputer with 1 Petaflops peak performance in single precision

A supercomputer with 1.0 Petaflops peak performance in single precision, designed and established by Institute of Process Engineering, Chinese Academy of Sciences, is introduced in this brief communication. A designing philosophy utilizing the similarity between hardware, software and the problems to be solved is embodied, based on the multi-scale method and discrete simulation approaches developed at Institute of Process Engineering (IPE) and implemented in a graphic processing unit (GPU)-based hybrid computing mode. The preliminary applications of this machine in areas of multi-phase flow, molecular dynamics and so on are reported, demonstrating the supercomputer as a paradigm of green computation in new architecture.     

Multiphase Thermohaline Convection in the Earth’s Crust: I. A New Finite Element – Finite Volume Solution Technique Combined With a New Equation of State for NaCl–H2O

We present a new finite element – finite volume (FEFV) method combined with a realistic equation of state for NaCl–H₂O to model fluid convection driven by temperature and salinity gradients. This method can deal with the nonlinear variations in fluid properties, separation of a saline fluid into a high-density, high-salinity brine phase and low-density, low-salinity vapor phase well above the critical point of pure H₂O, and geometrically complex geological structures. Similar to the well-known implicit pressure explicit saturation formulation, this approach decouples the governing equations. We formulate a fluid pressure equation that is solved using an implicit finite element method. We derive the fluid velocities from the updated pressure field and employ them in a higher-order, mass conserving finite volume formulation to solve hyperbolic parts of the conservation laws. The parabolic parts are solved by finite element methods. This FEFV method provides for geometric flexibility and numerical efficiency. The equation of state for NaCl–H₂O is valid from 0 to 750°C, 0 to 4000 bar, and 0–100 wt.% NaCl. This allows the simulation of thermohaline convection in high-temperature and high-pressure environments, such as continental or oceanic hydrothermal systems where phase separation is common.     

Development of a highly accurate interpolation method for mesh‐free flow simulations I. Integration of gridless,particle and CIP methods

A simple, efficient, flexible and accurate interpolation method, CIVA, is introduced for use with mesh‐free methods for flow simulations. The method enables mesh‐free cubic interpolation with the local co‐ordinate system, such as volume and area co‐ordinates, by utilizing the concept of the CIP scheme and allows the development of new highly accurate mesh‐free methods. The mesh‐free methods integrate the gridless, particle and CIP methods since they have flexibility in the treatment of moving calculation points. For achieving high accuracy with the CIVA method, it is also important to correctly evaluate particle movement. The improvement of the evaluating algorithm is another objective of this study. The validity of the algorithms is confirmed by applying them to the convection and convection–diffusion problems. Since the CIVA‐based mesh‐free methods enable flexible, efficient and accurate fluid simulation, they make it possible to perform highly accurate simulations of many kinds of problems that involve complicated geometries and phenomena. Copyright © 1999 John Wiley & Sons, Ltd.     

Internal wave reflections and transmissions arising from a non-uniform mesh. Part I: An analysis for the Crank–Nicolson linear finite element scheme

This is the first of a series of three related papers dealing with some of the consequences of non-uniform meshes in a numerical model. In this paper the accuracy of the Crank–Nicolson linear finite element scheme, which is applied to the linear shallow water equations, is examined in the context of a single abrupt change in nodal spacing. The (in)accuracy is quantified in terms of reflection and transmission coefficients. An incident wave impinging on the interface between two regions with different nodal spacings is shown to give rise to no reflected waves and two transmitted waves. The analysis is verified using three different wavelengths (2Δx, 4Δx 8Δx) in three ‘hot-start’ numerical experiments with a mesh expansion factor of 2 and three experiments with a mesh contraction factor of 1/2. An energy flux analysis based on the concept of group velocity shows that energy is conserved across the interface.