Begell House Inc.
Multiphase Science and Technology
MST
0276-1459
21
1-2
2009
EXTRACTING INFORMATION FROM TIME SERIES DATA IN VERTICAL UPFLOW
1-12
10.1615/MultScienTechn.v21.i1-2.10
Ryuhei
Kaji
Process and Environmental Engineering Division, Faculty of Engineering, University of Nottingham, University Park, Nottingham NG7 2RD, United Kingdom
John H.
Hills
Process and Environmental Engineering Division, Faculty of Engineering, University of Nottingham, University Park, Nottingham NG7 2RD, United Kingdom
Barry J.
Azzopardi
Department of Chemical, Environmental and Mining Engineering, The University of Nottingham, University Park, Nottingham NG7 2RD, England
Data was obtained from eight ring-pair probes mounted along the length of a vertical 19-mm-diameter, 7-m-long pipe. This was provided with air and water at its base and the flow was allowed to develop. The ring-pair probes consisted of two stainless steel rings mounted flush with the pipe wall and a short distance apart. The conductivity between them was measured continuously. Careful calibration allowed the void fraction or film thickness to be obtained from this data. This presentation shows the type of information that can be extracted from the void fraction time series. For time series taken at conditions corresponding to slug flow, two thresholds were employed to separate Taylor bubble and liquid slug regions. The crests and troughs were detected through a change in the sign of the void fraction/time curve. Then the velocities of individual Taylor bubble were obtained by cross-correlating signals from two successive probes with selective fragments. The distribution of lengths of Taylor bubbles and liquid slugs is presented. In addition, careful examination of portions of the time series enabled the waves on the film surrounding the Taylor bubbles to be identified and quantified. These results show that there is a distribution of velocities of these small waves, with some travelling upward and some downward.
CFD PREDICTION OF BUBBLE BEHAVIOR IN TWO-DIMENSIONAL GAS-SOLID FLUIDIZED BEDS
13-24
10.1615/MultScienTechn.v21.i1-2.20
A.
Busciglio
Dipartimento di Ingegneria Chimica dei Processi e dei Materiali, Università degli Studi di Palermo, Palermo, Italy
G.
Vella
Istituto di Applicazioni e Impianti Nucleari Facolta di Ingegneria - Universita di Palermo Viale delle Scienze, 90128 Palermo, Italy
G.
Micale
Dipartimento di Ingegneria Chimica dei Processi e dei Materiali, Università degli Studi di Palermo, Palermo, Italy
L.
Rizzuti
Dipartimento di Ingegneria Chimica dei Processi e dei Materiali, Università degli Studi di Palermo, Palermo, Italy
bond fluctuation model
annealing
polymer melts
This work focuses on the computational fluid dynamics (CFD) simulation of a laboratory-scale, two-dimensional fluidized bed and the relevant experiments in order to validate the prediction capability of the adopted codes and models. Both experimental and computational quantitative data were analyzed by means of an original digital image analysis technique, allowing for coherent comparison of computational and experimental results. In particular, this work analyzes the capability of the CFD simulations in predicting the fluctuating behavior of bubbling fluidized beds by means of frequency analysis of bubble-related phenomena.
OIL-WATER FLOW IN HORIZONTAL PIPES: A FLOW PATTERN AND PRESSURE DROP STUDY
25-35
10.1615/MultScienTechn.v21.i1-2.30
Christophe
Conan
Institut Français du Pétrole, IFF, 1 ave Bois Préau, 92500 Rueil Malmaison, France
Sandrine
Decarre
1Institut Français du Pétrole, IFF, 1 ave Bois Préau, 92500 Rueil Malmaison, France
Olivier
Masbernat
Laboratoire de Génie Chimique, UMR 5503 CNRS/INPT/UPS, 31106 Toulouse Cedex 1, France
Alain
Liné
Laboratoire d'Ingénierie des Systèmes Biologiques et des Procédés, UMR INSA-CNRS 5504, UMR INSAINR 792, F-31077 Toulouse Cedex 4, France
This article presents an experimental study of a liquid-liquid dispersed/stratified flow in a horizontal pipe. All the flows studied contain a highly concentrated layer of oil drops (light phase) under which flows a continuous layer of aqueous phase (heavy phase). Depending on the dispersed phase concentration, a third layer of pure oil may appear at the top of the pipe. The impact of the oil injection system has been studied and it appears that it plays a role in the transition, being one of the systems favoring the coalescence of the oil droplets. The evolution of the pressure drop is discussed. It is systematically greater than that obtained for a single water flow, confirming observations made by Elseth (2001), Nadler and Mewes (1995), and Angeli (1996) for a steel pipe. A correlation between the pressure loss and the experimental conditions (mixture velocity and phase ratio) is proposed. Two classical models, usually used for gas-liquid stratified flows (Lockart and Martinelli) and for fully dispersed flows (homogeneous model), are also tested and adjusted to dispersed/stratified flows.
CFD SIMULATIONS OF THE TURBULENT LIQUID-LIQUID FLOW IN A KENICS STATIC MIXER
37-50
10.1615/MultScienTechn.v21.i1-2.40
Zdzislaw
Jaworski
Szczecin University of Technology, Al. Piastow 42, 71-065 Szczecin, Poland
Halina
Murasiewicz
Szczecin University of Technology, Al. Piastow 42, 71-065 Szczecin, Poland
The study is a continuation of an earlier work presented at the previous MFIP conference, where initial modelling results of the turbulent flow of a two-phase, liquid-liquid mixture in a Kenics static mixer were reported. Since then, more advanced transient simulations have been performed using the large eddy simulation (LES) approach. The numerical modelling for the mixer comprised of 10 mixing inserts was carried out using the commercial Fluent 6.2.16 code, while the mixer geometry and the respective block-structured grid with about 900 K cells were generated in Gambit 2.0.4. The two-phase flow was modelled employing the Eulerian approach in the version of the "mixture model" available in the code. The simulations were performed for a Reynolds number of 10,000, with a volumetric ratio of 99% water to 1% oil. Three cases were considered, which differed by the density of the two phases. The numerical analysis led to determination of the local fluctuating velocities, vorticity, helicity of the mixture, and the distribution of the dispersed phase ratio in cross sections of the mixer. The mixture ratio and velocity distributions along the mixer length were averaged in time and compared to those obtained in the RANS approach. Significant changes were noticed and less centrifugal effect on the phase separation was simulated using LES.
PARTICLE IMAGE VELOCIMETRY, GAMMA DENSITOMETRY, AND PRESSURE MEASUREMENTS OF OIL-WATER FLOW
51-64
10.1615/MultScienTechn.v21.i1-2.50
W. A. S.
Kumara
Telemark University College, P. O. Box 203, N-3901, Porsgrunn, Norway
B. M.
Halvorsen
Telemark University College, P. O. Box 203, N-3901; Telemark R&D Centre (Tel-Tek), Kjolnes ring, N-3918, Porsgrunn, Norway
M. C.
Melaaen
Telemark University College, P. O. Box 203, N-3901; Telemark R&D Centre (Tel-Tek), Kjolnes ring, N-3918, Porsgrunn, Norway
Oil-water flow in a horizontal pipe was investigated using particle image velocimetry (PIV), gamma densitometry, and pressure measurements. The experimental activities were performed using the multiphase flow loop at Telemark University College, Porsgrunn, Norway. The multiphase flow loop consisted of a 15-m-long steel pipe with an inner diameter of 56 mm. Water (density 996 kg/m3, viscosity 1 mPa s) and Exxsol D60 oil (density 790 kg/m3, viscosity 1.6 mPa s) were used as test fluids. The experiments were performed at different mixture velocities and water cuts. Mixture velocity and water cut varied up to 1.06 m/s and 1.0, respectively. The time-averaged cross-sectional distributions of oil and water were measured using a traversable gamma densitometer. The pressure drop along the test section of the pipe was also measured. The flow regimes were determined by visual observations. The instantaneous local velocities were measured using PIV, and based on the instantaneous local velocities mean velocities and turbulence profiles were calculated. The highest root mean-squared velocity components of streamwise (U-rms) and wall normal (V-rms), and Reynolds stress values were observed close to the pipe wall due to large mean axial velocity gradients. A damping effect of the Reynolds stress was observed close to the oil-water interface due to stable density stratification. PIV measurements are compared with laser doppler anemometry (LDA) measurements that were performed on a similar experimental setup by Elseth et al. (2000). The measured mean axial velocity and turbulence profiles using PIV are observed to compare favorably with LDA measurements.
MODELLING OF LIQUID DISPERSION IN TRICKLE-BED REACTORS: CAPILLARY PRESSURE GRADIENTS AND MECHANICAL DISPERSION
65-79
10.1615/MultScienTechn.v21.i1-2.60
Katja
Lappalainen
Helsinki University of Technology, Chemical Engineering and Plant Design, P.O.B. 6100, FIN-02015 HUT, Finland
Ville
Alopaeus
Helsinki University of Technology, Chemical Engineering and Plant Design, P.O.B. 6100, FIN-02015 HUT, Finland
Mikko
Manninen
VTT Technical Research Centre of Finland, P.O.B. 1000, FI-02044 VTT, Finland
Sirpa
Kallio
VTT Technical Research Centre of Finland Ltd
Modelling is one of the most significant prospective tools for design and analysis of trickle-bed reactors. Unfortunately, current hydrodynamic models, developed on laboratory experiments, often work poorly in industrial scale. Therefore, physically, more authentic models are required in which the small-scale phenomena are separated from the large-scale phenomena. This would improve the scale-up of the model and consequently, its applicability to industrial-scale reactors. One of the small-scale phenomena lacking from the current models is radial distribution of liquid. It has not been considered in the model development, although it is commonly thought that liquid flow is radially more uniform in industrial than in laboratory scale. Here, models for liquid distribution, caused by capillary pressure gradients and mechanical dispersion, are suggested and the outline of the implementation of these models to CFD programs is presented. Laboratory experiments and CFD simulations of the experimental setup are performed to gain better understanding about liquid radial distribution. The physical validity of the presented models is assessed on the consistency between the experimental and the modelled liquid flow profiles.
VOLUME FLOW RATE MEASUREMENT IN VERTICAL OIL-IN-WATER PIPE FLOW USING ELECTRICAL IMPEDANCE TOMOGRAPHY AND A LOCAL PROBE
81-93
10.1615/MultScienTechn.v21.i1-2.70
Hua
Li
Institute of Particle Science and Engineering, University of Leeds, Leeds LS2 9JT, United Kingdom
Mi
Wang
Institute of Particle Science and Engineering, University of Leeds, Leeds LS2 9JT, United Kingdom
Ying-Xiang
Wu
Institute of Mechanics, Chinese Academy of Sciences, Beijing 100080, China
Gary
Lucas
School of Computing and Engineering, University of Huddersfield, Huddersfield HDl 3DH, United Kingdom
This paper presents the use of a high-performance dual-plane electrical impedance tomography (EIT) system and a local dual-sensor conductance probe to measure the vertical upward cocurrent oil-in-water pipe flows. Experiments were carried on a flow loop with a transparent 2.5-m-long, 80-mm inner diameter test section using kerosene and tap water. The flow conditions were predominantly of the dispersed type, with nonslip oil volume fractions of 9.1, 16.7, and 23.1%, respectively, and with two groups of different mixture velocities. A sensitivity coefficient back-projection algorithm was adopted to reconstruct the flow distributions from the EIT measurement data, and then the oil in situ volume fraction was calculated based on a Maxwell relationship with temperature compensation. The oil velocity distribution was obtained using a pixel-to-pixel cross-correlation method. A local intrusive conductance probe was adopted to supply a reference measurement of oil volume fraction and velocity profiles. The oil volume fraction and velocity distributions from the two techniques were compared and good agreement was found. A further calculation of the water velocity distributions and flow rates was implemented through the drift flux approach and the results were analyzed and discussed.
EFFECT OF DRAG MODELING ON THE PREDICTION OF CRITICAL REGIME TRANSITIONS IN AGITATED GAS-LIQUID REACTORS WITH BUBBLE SIZE DISTRIBUTION MODELING
95-106
10.1615/MultScienTechn.v21.i1-2.80
Miriam
Petitti
Department of Materials Science and Chemical Engineering, Politecnico di Torino, 10129 Torino, Italy
Daniele L.
Marchisio
Politecnico di Torino, Department of Materials Science and Chemical Engineering, 10129 Torino, Italy
Marco
Vanni
Dept. of Materials Science and Chemical Engineering, Politecnico di Torino, C.soDuca degli, Abruzzi, 24,10129 Torino
Giancarlo
Baldi
Dept. of Materials Science and Chemical Engineering, Politecnico di Torino, C.soDuca degli, Abruzzi, 24,10129 Torino
Nicola
Mancini
ENI R&M, 20097 S. Donato Milanese, Italy
Fabrizio
Podenzani
Eni R&M, Via Maritano 26 20097 San Donato Mil. (MI), Italy
A four baffled gas-liquid reactor, agitated by a Rushton turbine, has been modeled in a wide range of operating conditions (mixing intensities and gas flow numbers) by using a Eulerian multifluid approach coupled with a population balance model to describe the evolution of the bubble size distribution. In particular, the work has focused on the role played by drag force, calculated by resorting to the Tomiyama correlation and the Bakker correction for the slip turbulent reduction, on the predictions of fluid-dynamics regime transitions and of the structure assumed by the gas phase near the turbine blades. This investigation was carried out under very different operating conditions, also assessing the ability of the model to predict global data such as the overall gas hold-up and power number. Simulations were carried out via the commercial computational fluid dynamics code Fluent, and both the drag and the population balance model were implemented through user-defined functions and subroutines. Comparisons with correlations based on experimental data and directly with experimental data for the bubble size distribution, also at quite high gas hold-ups, showed that the Bakker correction for the slip turbulent reduction, when implemented with the standard constant values, underestimates the overall drag force. In order to improve agreement with experimental data, new constant values are proposed.
EXPERIMENTAL INVESTIGATION OF THREE-PHASE OIL-WATER-AIR FLOW THROUGH A PIPELINE
107-122
10.1615/MultScienTechn.v21.i1-2.90
Pietro
Poesio
Department of Mechanical and Industrial Engineering, University of Brescia, Italy
Giorgio
Sotgia
Dipartimento di Energetica, Politecnico di Milano, Milano Italy
Domenico
Strazza
Università degli Studi di Brescia, Dipartimento di Ingegneria Meccanica e Industriale, Via Branze 38, 25123 Brescia, Italy
A three-phase flow of oil, water, and gas through a pipeline often occurs in industry. It is important to study such flows in order to be able to design a three-phase flow pipeline. To that purpose, experiments with oil, water, and air in a 28- and a 40-mm i.d. glass pipe were carried out to derive a new data set for a three-phase flow in a horizontal pipeline. First, we investigated the pressure drop, showing the influence of air injection on the two-phase flow reference flow pattern. We found a strong link between the qualitative behavior of the three-phase pressure drop reduction factor and the two-phase flow reference flow pattern. Finally, we studied the dynamic characteristics of the elongated air bubbles, analyzing both the bubble frequency and velocity.
FROM ELEMENTARY PROCESSES TO THE NUMERICAL PREDICTION OF INDUSTRIAL PARTICLE-LADEN FLOWS
123-140
10.1615/MultScienTechn.v21.i1-2.100
Martin
Sommerfeld
PAI+ Group, Department of Energetics & Mechanics, Universidad Autónoma de Occidente, Cali, Colombia
Santiago
Lain
PAI+ Group, Department of Energetics & Mechanics, Universidad Autónoma de Occidente, Cali, Colombia
In particle-laden flows, the overall flow structure and the relevant process parameters, e.g., pressure drop or separation efficiency, are strongly affected by the elementary processes occurring on the scale of the particles. Therefore, a detailed modeling of these microscale phenomena is required when anticipating reliable numerical predictions. Here the Euler/Lagrange approach was further developed in order to calculate confined particle-laden flows in pneumatic conveying lines and gas cyclones. Special emphasis is placed on the influence of particle-wall collisions and wall roughness as well as interparticle collisions with possible agglomeration on the developing two-phase flow structure and the resulting process parameters. The models and the numerical method were validated based on the pressure drop measured along a 6-m horizontal channel. The agreement was found to be excellent for different particle sizes, mass loading, and wall roughness. The numerical predictions of a horizontal pipe flow revealed that due to a wall roughness-induced focusing of particle trajectories toward the core of the pipe, a secondary flow in the pipe cross section develops. Moreover, it was found that the additional pressure drop due to the particles in the pipe flow was higher than that in the channel due to the different wall collision behavior. The numerical calculations of particle separation in a gas cyclone revealed the importance of a detailed modeling of interparticle collisions and particle agglomeration on the resulting grade efficiency curves. Agglomeration improves the separation of fine particles substantially.
HYBRID MULTIPHASE-FLOW SIMULATION OF BUBBLE-DRIVEN FLOW IN COMPLEX GEOMETRY USING AN IMMERSED BOUNDARY APPROACH
141-155
10.1615/MultScienTechn.v21.i1-2.110
Masahiro
Tanaka
Graduate School of Engineering, Kobe University, Rokkodai, Nada, Kobe 657, Japan
Kosuke
Hayashi
Department of Mechanical Engineering, Graduate School Engineering, Kobe
University, 1-1 Rokkodai, Nada, Kobe, Hyogo, 657-8501 Japan
The purpose of this study is to develop a practical numerical method for predicting multiphase flows in industrial systems of large scale. The practical method should be equipped with two functions. One is the ability of dealing with complex geometry of a given practical system without time-consuming grid generation procedures. The other is to simulate flows in large complex systems without using a high number of computational cells. Therefore, a numerical method based on the combination of a multifluid and interface tracking method and an immersed boundary method is proposed in this study. The hybrid combination of the multifluid and interface tracking methods enables us to simulate multiphase flows with various scales and various phases, and the immersed boundary method makes it possible to satisfy the two requirements. Several numerical simulations are carried out to demonstrate the potential of the proposed method. Comparisons between measured and predicted bubbly flows around single obstacles prove that the proposed method gives reasonable predictions for the interaction between bubbly flow and structures. Simulation of a bubble column with complex structure demonstrates its applicability to large industrial systems.
A NEW METHOD OF MEASURING TWO-PHASE MASS FLOW RATES IN A VENTURI
157-168
10.1615/MultScienTechn.v21.i1-2.120
He
Peixiang
Faculty of Mechanical Engineering, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Cees W. M.
van der Geld
Department of Mechanical Engineering, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Claudio
Alimonti
Al University “La Sapienza” of Rome, Italy
Julio Cesar
Passos
Departamento de Enga. Mecânica-LEPTEN/Boiling-Centro Tecnológico, Universidade Federal de Santa Catarina, Campus Universitário, 88.040-900 Florianópolis-SC, Brazil
Metering of the individual flow rates of gas and liquid in a multicomponent flow is of great importance for the oil industry. A convenient, nonintrusive way of measuring these is the registration and analysis of pressure drops over parts of a venturi. Commercially available venturi-based measuring equipment is costly because it also measures the void fraction. This paper presents a method to deduce the individual mass flow rates of air and water from pressure drop ratios and fluctuations in pressure drops. Not one but two pressure drops are used and not only time-averaged values of pressure drops are utilized. As a proof-of-principle, prediction results for a horizontal and vertical venturi are compared with measurements for void fractions up to 80%. Residual errors are quantified and the effect of variation of equipment and of slip correlation is shown to be negligible. At relatively low cost a good predictive capacity of individual mass flow rates is obtained.
DIRECT NUMERICAL SIMULATION OF AN INDIVIDUAL FIBER IN AN ARBITRARY FLOW FIELD−AN IMPLICIT IMMERSED BOUNDARY METHOD
169-183
10.1615/MultScienTechn.v21.i1-2.130
Srdjan
Sasic
Department of Applied Mechanics, Chalmers, 412 96 Göteborg, Sweden
Berend G. M.
van Wachem
Formerly at the Department of Mechanical Engineering, Imperial College London, SW7 2AZ, UK; Lehrstuhl für Technische Thermodynamik, Otto-von-Guericke-Universität Magdeburg, 39106 Germany
An immersed boundary method for three-dimensional, time-dependent flows is presented in this work and applied to simulating the behavior of an individual fiber in various flow regimes. The fiber is placed in a periodic box and has either a fixed position or is allowed to move freely (including translation and rotation) through the domain. The immersed boundary method is used to match the fluid velocity with the velocity of the interface of the fiber by mirroring the velocity field along the normal of the local triangulated immersed boundary segment to guarantee that the fluid accurately takes into account the presence of an immersed body. As a result of the procedure, there is a fictitious velocity field inside the immersed boundary, mirroring the boundary layer. The method applied is second-order accurate for the drag on the fiber and is intended to be used for fully resolving the flow field around arbitrary moving bodies immersed in a fluid. The immersed boundary method is employed on a selection of different fiber shapes, aiming at predicting the behavior of real fibers in realistic flow situations. A grid refinement study and study of the influence of the size of the periodic box used for the simulations are carried out. It is shown by grid refinement that the simulations performed here are truly direct numerical simulation (DNS). The force exerted by the fluid on a fiber is directly calculated by integrating the pressure and viscous forces over the objects immersed. The resulting coarse-grained drag and lift force functions can be employed by calculations of fluid-fiber flows on a larger scale (e.g., Eulerian-Eulerian simulations of air-fiber flows).