·]J"fi"· European Research Community ~ -... 't;,JJ ~ On Flow, Turbulence -,,,-sr ... <;!·. -'!:1-'!f~.i· • And Combustion SIAMUF, Swedish Industrial Association for Multiphase Flows 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 !!est Practice Guidelines for Computational Fluid Dynamics of Dispersed Multiphase Flows Editors: Martin Sommerfeld, Berend van Wachem, Rene Oliemans Version 1, October 2008 Front picture: courtesy by Prof. Jos Derksen Copyright statement: All rights reserved; no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form by any means, electronic, mechanical, photocopying, recording or otherwise without .t. he prior written permission ofERCOFTAC . ISBN 978-91-633-3564-8 Contents: 1 Introduction .............................................................................................................. 2 2 Fundamentals ..........................................................................................................4 2.1 Classification of Multi-Phase Flows ........................................................................... 4 2.2 Integral characterization of Multi-Phase Flows .......................................................... 5 2.3 CFD requirements for industrial dispersed multiphase flows ...................................... 7 3 Forces acting on particles, droplets and bubbles ................................................... 11 3.1 Drag force ................................................................................................................. 12 3.2 Pressure gradient and Buoyancy force ....................................................................... 19 3 .3 Added mass force and Basset force ........................................................................... 19 3 .4 Body force ................................................................................................................ 20 3 .5 Saffman and Magnus lift forces ................................................................................ 21 3.6 Torque ...................................................................................................................... 24 3.7 Response time and Stokes number ............................................................................ 25 3 .8 Forces acting on bubbles ........................................................................................... 27 3 .9 Importance of the different forces ............................................................................. 31 4 Computational Multiphase Fluid Dynamics for dispersed flows ............................. 35 4.1 DNS resolving the particles ...................................................................................... 35 4.2 Discrete Particle Model (DPM) ................................................................................. 39 4.3 Eulerian-Lagrangian point-particle DNS, LES, and RANS ...................................... .47 4.4 Euler/Euler (Two-Fluid) method ............................................................................... 57 5 Specific Phenomena and modelling approaches ................................................... 64 5 .1 Particle-Wall Collisions ............................................................................................ 64 5.2 Inter-Particle Collisions ............................................................................................ 72 5 .3 Heat and mass transfer in droplets and sprays ........................................................... 81 5 .4 Open modelling issues .............................................................................................. 86 6 Sources of errors ................................................................................................... 91 7 Industrial examples for multiphase flows ............................................................... 94 7.1 Bubble column ......................................................................................................... 94 7 .2 Air Lift ..................................................................................................................... 97 7 .3 Cyclones ................................................................................................................. 100 7 .4 Stirred tanks ............................................................................................................ 102 7 .5 Fluidized bed .......................................................................................................... 103 7 .6 Summary of exemplary test cases ........................................................................... 112 8 Checklist of Best Practice Advice ........................................................................ 114 9 Suggestions for future developments .................................................................. 116 1O References ....................................................................................................... 117 © ERCOFTAC 2008 1 1 Introduction These Best Practice Guidelines for Computational Fluid Dynamics of turbulent dispersed multiphase flows are a follow-up of the previous ERCOFTAC Best Practice Guidelines for Industrial CFD and should be used in combination with it. The potential users are Master/PhD students from Academia who are embarking on a project on CFD of (wall-bounded) turbulent dispersed multiphase flows and engineers in industry, using CFD codes for design and de-bottle necking of multiphase flow equipment. Engineers in industry have the option of using RANS or LES as turbulence models, and an Euler/Euler or Euler/Lagrange approach. They are interested in the size distribution of the dispersed phase and the dispersion of particles in the complex turbulent flow field as function of space and time. They use the CFD code as a subroutine for their design parameter and scale-up studies, so the code should be fast and reliable. The Master/PhD students at Universities tend to use more sophisticated turbulence models (DNS and LES) for idealized flow geometries (homogeneous turbulence, channel flow, free jets, etc.) with point particles and detailed models to study the behaviour of isolated or a restricted number of particles (Interface tracking, Level set, VOF). Ideally these guidelines will enable the newcomer in industry to obtain insight into the technique of performing the computations and the degree of sophistication required for his problem. For example if one wants to establish for a Stirred Tank Reactor design the positioning of the stirrers on the shaft a RANS model is adequate for the turbulence. If, however, one wants to know the variation of turbulence intensity to determine the size distribution of drops injected one should use at least LES for such a complex turbulent flow field. So, these.Best Practice Guidelines have the aim of serving both stakeholders by supplying information on the methods that are available and their possibilities and limitations. To achieve that goal it was necessary to provide a survey of the relevant models available for turbulent dispersed multiphase flows, structured such that the user can identify which models serve his ambition best. In addition to assessing the complexity of having to deal with multiphase flows the user should ensure that solutions are provided for grids and time steps, which are fine enough so that numerical errors are minimized. For the multiphase flow applications one has to realize that the spatial and temporal distribution of the dispersed phase may require local grid refinements to achieve accurate solutions. This document starts by addressing in Chapter 2, entitled Fundamentals, the classification of multiphase flows and of dispersed flows in particular. Here an important distinction is that between the dilute flows, with a small number of particles present in the turbulent flow field, and the dense dispersed multiphase flows, commonly encountered in industrial applications. For the latter one will· have to account for turbulence modulation due to the presence of the particles (two-way coupling) and particle collisions. The last section of this chapter provides overviews of turbulence models for the carrier fluid (gas or liquid): Direct Numerical Simulation (DNS), Large Eddy Simulation (LES) and Reynolds Average Navier-Stokes (RANS) and models for computational turbulent dispersed multiphase flows. Chapter 3 can be used to define the forces on bubbles, drops and particles to be accounted for in practical applications. Although drag forces are quite common, many engineering flows ask for other particle forces as well. To facilitate the choice of force models needed in an engineering application a survey of the importance of different forces is given. Chapter 4 gives a survey of the Computational Multiphase Fluid Dynamics methods for turbulent dispersed flows. It ranges from very refined models to study the detailed flow around particles with · sophisticated turbulence models to the Euler/Euler RANS methods that are very popular to perform industrial multiphase flow calculations with a great number of particles with a broad range of sizes flowing in a turbulent flow field in a complex geometry. Due to the complexity of multiphase flows, LES is a better choice than RANS in some cases, even though it might only give you information on a smaller part of the system - it does give it more correctly. Moreover, as computer power increases, more and more LES calculations become affordable. © ERCOFfA C 2008 2 Chapter 5 addresses particle-wall collisions, inter-particle collisions and heat and mass transfer for droplets and sprays as specific phenomena and modelling approaches. For engineering flows it is important to realize that the proximity of the walls and their roughness will affect the nature of the turbulent flow field and the behaviour of the bubbles, drops and particles in that field. In Chapter 6 we summarize the error sources when performing computer simulations of turbulent dispersed multiphase flows in equipment. Engineering examples for multiphase flows for which successful simulations have been performed can be found in Chapter 7. This chapter concludes with a summary of exemplary test cases that can serve as a basis for model validation. This certainly is very important, and should be a key element in the selection of proper models for the simulation of a practical multiphase flow situation. Of course the validation ought to be a necessary activity of the developers of commercial codes. It is recommended to use the ERCOFTAC data base on multiphase flow benchmarks, established over the years at workshops organized by ERCOFTAC's Special Interest Group on Dispersed Turbulent Two-Phase Flow for validation purposes (see www.ercoftac.org and www-mvt.iw.uni-halle.de). Also the Best Practice Guidelines for CFD Code validation for Reactor-Safety Applications (Menter 2002) is highly recommended and considered useful for other applications as well. A checklist of Best Practice Advice is given in Chapter 8. This chapter also contains a list of recommended simulation methods in industrial examples for turbulent dispersed multiphase flows. Chapter 9 deals with suggestions for future developments needed to enhance our capability to perform computer simulations faster and with a greater precision. Finally, Chapter 10 contains the references to the relevant literature. Acknowledgements A project like this is team work with sponsoring organizations and specialists willing to contribute. The work was carried out with the highly appreciated contributions from: Prof. Jos Derksen (University of Alberta), Dr. Muhamed Hadziabdic (International University of Sarajevo), Prof. Hans Kuipers and Dr. Niels Deen (Twente University), Prof. Rob Mudde (Delft University· of Technology), Prof. Dirk Roekaerts and Dr. Nijso Beishuizen (Delft University of Technology) and Prof. Alfredo Soldati and Dr. Christian Marchioli (University of Udine). The contents were critically reviewed by Prof. Charles Hirsch (NUMECA International), Dr. Chris Carey (ANSYS Northern Europe) and Prof. Mick Casey (Institut fiir Thermische Stromungsmaschinen, Stuttgart). The editors express their gratitude to them all. The project was sponsored by SIAMUF, the Swedish Industrial Association for Multiphase Flows. These Multiphase Best Practice Guidelines are dedicated to the late Prof. Rolf Karlsson (Vattenfall Utveckling and Chalmers University), on whose request we embarked on this ambitious project. He was a strong advocate of ERC O FT AC' s Best Practice Guidelines for Industrial Computational Fluid Dynamics and convinced us to accept to work on a multiphase version. Please note that this is the very first version of the Best Practice Guidelines for Multiphase Flows, which undoubtedly will be followed by subsequent versions, since the modelling of turbulent dispersed multiphase flows in engineering environments is a complex research field very much in the limelight of excellent researchers making rapid progress. ©ERC OF TA C 2008 3 2 Fundamentals The simultaneous presence of several different phases in external or internal flows such as gas, liquid and solid is found in daily life, environment and numerous industrial processes. These types of flows are termed multiphase flows, which may exist in different forms depending on the phase distribution. Examples are gas-liquid transportation, crude oil recovery, spray cans, sediment transport in rivers, pollutant transport in the atmosphere, cloud formation, fuel injection in engines, bubble column reactors and spray driers for food processing, to name only a few. This demonstrates the great importance of multiphase flows, which might occur even more frequently than single phase flows. As a result of the interaction between the different phases such flows are rather complicated and very difficult to describe theoretically. For the design and optimisation of such multiphase systems a detailed understanding of the interfacial transport phenomena is essential. For single-phase flows computational fluid dynamics (CFD) has already a long history and it is nowadays standard in the development of airplanes and cars using different commercially available CPD-tools. Due to the complex physics involved in multiphase flow the application of CFD in this area is rather young (probably 20 - 30 years). The different methods being used for the numerical calculation of multiphase flows will be summarised below. This chapter is devoted to the classification of multiphase flows, their characterisation by integral properties and the CFD requirements for industrial turbulent dispersed multiphase flows. 2.1 Classification of Multi-Phase Flows Multiphase flows may be encountered in various forms in industrial practice (see Figure 2.1), as for example, transient flows with a transition from pure liquid to a vapour flow as a result of external heating (e.g. heat pipe), separated flows (i.e. stratified flows, slug flows, or film flows), and dispersed two-phase flows where one phase is present in the form of particles, droplets, or bubbles dispersed in a continuous carrier phase (i.e. gas or liquid). In all these different types of multiphase flows also different interfacial transport mechanisms are relevant. Consequently also different numerical approaches have to be applied. Transient multiphase flows may be found in steam generators of boilers, where the heat addition results in the formation of dispersed bubbles that further grow in size and also will coalescence yielding large vapour slugs. Further evaporation results in annular two-phase flow with small droplets being dispersed in the core of the pipe. Stratified and slug flows are for example found in transportation pipes for crude oil recovery, where the observed flow regime also depends on the orientation of the pipeline, e.g. horizontal, vertical or inclined. Dispersed two-phase flows are encountered in numerous technical and industrial processes, as for example in particle technology (i.e. production and transportation of solid particles), chemical engineering, and biotechnology. Dispersed two-phase flows may be classified in terms of the different phases being present as summarised in Table 2.1 together with some of the most important industrial processes. Commonly dispersed two-phase flows are separated in two flow regimes. In dilute dispersed systems the spacing between the particles is rather large, so that a direct interaction between the particles is rare and fluid dynamic forces are governing particle transport. On the other hand dense dispersed systems are those where the inter-particle spacing is comparatively low (i.e. smaller than about 10 particle diameters). Under such conditions the transport of the particles is dominated by collisions between them as for example in a fluidised bed. Additionally, numerous processes may involve more than two phases (i.e. multiphase flows), as for example in a spray scrubber where droplets and solid particles are dispersed in a gas flow and the aim is to collect the particles by the droplets. Another example is a bubble column reactor with catalyst particles. © ERCOFfA C 2008 4 .. • a) 0 0 00 0 0 0 0 ~ 0 00 0 0 00 0 C0)(c))( J0 0 O0o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ° 0 0 0 0 0 b) ........... . . . Gas Gas Gas Liquid • c) 00 0o Q·O0 0o o0. o0o o 0' ~0Q (0~'o 0 0 0 )o(00 3 0000 000 O 00O 00 o 0oo Qo·0o oo0 0 o c00 )0 00 00 0 00 0 0 0 0 0 0 0 0 0 0000 0o G·o0a 0s 00 0 00 0 000 Iii Liquid Gas Bubbles Liquid Droplets or Solid Particles Figure 2 .1: Different regimes of two-phase flows, a) transient two-phase flow, b) separated two-phase flow, c) dispersed two-phase flow. Continuous/Dispersed industrial/technical application Phase Gas-solid flows pneumatic conveying, particle separation in cyclones and filters, fluidised beds Liquid-solid flows hydraulic conveying, liquid-solid separation, particle dispersion in stirred vessels Gas-droplet flows spray drying, spray cooling, spray painting, spray scrubbers Liquid-droplet flows mixing of immiscible liquids, liquid-liquid extraction Liquid-gas flows bubble columns, aeration of swage water, flotation - Table 2 .1: Summary of two-phase flow systems and important industrial and technical processes. 2.2 Integral characterization of Multi-Phase Flows For the characterisation of dispersed two-phase flows different integral properties are used, which are briefly summarised below. The volume fraction of the dispersed phase is the volume occupied by the particles in a unit volume. Hence this property is given by: I•i VP; ___ a=~; v (1) p where Vis the total volume, N; is the number of all particles in the size fraction i, having the particle © ERCOFfA C 2008 5 -t-. d3 7[ volume VP; = Throughout this manuscript subscript P will be used to label solid or liquid (bubble) particles while F will be used to label fluid phase. The particle diameter d P; in this context is the volume-equivalent diameter of a sphere. In case multiple phases are present one can define a volume fraction for each phase. Since the sum of the volume fraction of the dispersed phases and the continuous phase is unity, the continuous phase volume fraction is: (2) The bulk density or concentration of the dispersed phase is the mass of particles per unit volume and hence given by: (3) Correspondingly, the bulk density of the continuous phase is: (4) The sum of both bulk densities is called mixture density: (5) Often the particle concentration is also expressed by the number of particles per unit volume, as for example in clean-room technology: NP n =- (6) P V Especially in gas-solid flows for example in pneumatic conveying the mass loading is frequently used, which is defined as the total mass flux of the dispersed phase to that of the fluid: (7) where up and uF are bulk velocities of disperse phase and fluid respectively. The mass flux is also quite often used and is defined as the mass of particles flowing through a unit area per unit time. It should be noted that the mass flux is a vector quantity (i.e. a mass flux can be defined for each velocity direction), which also can be defined as a local property. The proximity of particles in a two-phase flow system may be estimated from the inter-particle spacing, which however can be only determined for regular arrangements of the particles. For a cubic arrangement the inter-particle spacing, i.e. the distance between the centres of particles L, is obtained from: ( JJ/3 6:p ~ (8) = For a volume fraction of 1 % the spacing is 3.74 diameters and for 10 % only 1.74. Hence, for such high volume fractions the particles cannot be treated to move isolated, since fluid dynamic interactions become of importance. In many practical flujd-particle systems however, the particle volume fraction is much lower. Consider for example a gas-solid flow (particle density pp = 2500 kg/m3, gas density of PF = © ERCOFTAC 2008 6 1.18 kg/m3) with a mass loading of one (i.e. 17 = 1) and assume no slip between the phases, then the volume fraction is about 0.05 % (i.e. ap = 5 · 104). This results in an inter-particle spacing of about 10 particle diameters, hence, under such a condition a fluid dynamic interaction may be neglected. In industrial bubble columns the gas volume fraction can have values of 40 % or even more. This yields an inter-bubble spacing of 1.1 bubble diameters. This will result in such a highly turbulent flow in a large collision rate and hence bubble coalescence will occur. inter-particle spacing LI dp 100 10 1 I I Dilute Dispersed Dense Dispersed Two-Phase Flow Two-Phase Flow l,l' ~ ~,/ U' One-Way Two-Way Four-Way Coupling Coupling Coupling I 1E -8 1E-7 1E-6 1E -5 1E-4 1E-3 0.01 0.1 volume fraction[-] Figure 2 .2: Regimes of dispersed two-phase flows as a junction ofp article volume fraction. A classification of dispersed two-phase flows with regard to the importance of interaction mechanisms was provided by Elghobashi (1994). Generally it is distinguished between dilute and dense two-phase flows as mentioned above (Figure 2.2). A two-phase system may be regarded as dilute for volume fractions up to ap; 10-3 (i.e. L/dp"" 8). In this regime the influence of the particle phase on the fluid flow may be neglected for ap < 10-6 (i.e. L/dp"" 80). For higher volume fractions the influence of the particles on the fluid flow, which is often referred to as two-way coupling, needs to be accounted for. In the dense regime (i.e. for ap > 10-3) additionally inter-particle interactions (i.e. collisions and fluid dynamic interactions between particles) become of importance. Hence, this regime is characterised by the so called four-way coupling. 2.3 CFD requirements for industrial dispersed multiphase flows Engineers in industry use CPD for turbulent dispersed multiphase flow problems as a tool to examine complex large scale systems for scale-up from laboratory to conditions in practice. It should enable them to develop design and operational criteria for their process equipment. The fluid dynamics of the dispersed multiphase flow systems is only an element of their more complex operational system for which many additional parameters (such as heat and mass transfer, throughput variation, turn-down ratio and emergency shut-downs). This means that an engineer expects to be able to produce accurate wall bounded turbulent flow fields with proper spatial and time-dependent distributions of the phases. He needs fast, reliable and accurate calculations and guidance to handle the enormous amount of simulation data. Prior to performing the dispersed multiphase flow simulations the engineer has to determine both the turbulence model, with its related closure models and wall functions, and the closure relations for the transport equations of the particles. Table 2.2 displays typical ranges for particle volume fractions and particle sizes in bubble columns, stirred tank reactors, fluidized beds and a cyclone. © ERCOFTAC 2008 7 = Bubble column ap 0.10 - 0.40 = dp 1-30mm = Stirred tank reactor ap 0.02-0.40 = dp 500 µm = Fluidized Bed ap 0.10-0.40 = dp 50-500 µm = Cyclone ap 0.001 = dp 1-50 µm Table 2.2 Particle volume fractions and sizes in industrial equipment. 400 ... >- t::: en • z • w ~ 300 lz- w + ...J .. + ::i aIl:l 200 x I X ::i I l- I + I " ~ I I 1lz..s:1J JOO )(II ~ ~ t.+ <;t I =~ ::c: u ..:.+ + ~ •o -~oL.._._.....-...... .;.....i~..._. ...............~ :--................' """"'"!7--'-_._. .................- :-' QOOOI 0.001 QOI 0.1 dp/ '• Figure 2.3 The effect ofp articles on turbulence intensity. When we compare the volume fractions with those shown in Figure 2.2 it is clear that for such industrial equipment two-way coupling and particle collisions are the rule rather than the exception. Even for the most dilute flow example, the cyclone, the particles are expected to affect the turbulence of the continuous phase (two-way coupling). Depending on particle size in comparison to a characteristic length scale of the turbulent flow field in which they move turbulence can be suppressed or enhanced as is shown in Figure 2.2, the well known turbulence modulation results collected on the basis of experimental observations by Gore and Crowe (1989). An additional complication for the flow of particles in wall-bounded systems is that particles are not homogeneously distributed, but may concentrate in local area~ of the flow field. In particular for riser flow, the accumulation of particles in the wall region (known as segregatfon) may have a negative effect on the mass transfer, hence on the efficiency of conversion processes. Answers to the questions on which particle forces, turbulence models and particle-wall collisions, inter particle collisions and heat and mass transfer aspects can be found in the subsequent chapters 3-5 of this document. Regarding the choice of a turbulence model for the carrier fluid (see also Loth 2000) there are three options as shown in Table 2.3. © ERCOFTA C 2008 8