Chemical Engineering Science 54 (1999) 1975}1995 Multiphase reactors } revisited Milorad P. Dudukovic!,*, Faical Larachi", Patrick L. Mills# !Chemical Reaction Engineering Laboratory (CREL), Washington University, St. Louis, MO 63130, USA "Department of Chemical Engineering, Laval University, Quebec, Canada G1K 7P4 #Du Pont Central Research, Wilmington, DE, 19880-0262, USA Abstract Multiphase reactors are found in diverse applications such as in manufacture of petroleum-based fuels and products, in production of commodity and specialty chemicals, pharmaceuticals, herbicides and pesticides, in re"ning of ores, in production of polymers and other materials, and in pollution abatement. In all such applications, the knowledge of #uid dynamic and transport parameters is necessary for development of appropriate reactor models and scale-up rules. The state of the art of our understanding of the phenomena occurring in three-phase reactors such as packed beds with two-phase #ow, slurry bubble columns and ebullated beds is summarized in this review. ( 1999 Elsevier Science Ltd. All rights reserved. Keywords: Trickle bed; Ebullated bed; Bubble column; Fluid dynamics 1. Introduction Processes based upon multiphase reactions occur in a broad range of application areas and form the basis for manufacture of a large variety of intermediate and consumer end-products. Some examples of multiphase reactor technology uses include: (1) the upgrading and conversion of petroleum feed stocks and intermediates; (2) the conversion of coal-derived chemicals or synthesis gas into fuels, hydrocarbons, and oxygenates; (3) the manufacture of bulk commodity chemicals that serve as monomers and other basic building blocks for higher chemicals and polymers; (4) the manufacture of pharmaceuticals or chemicals that are used in "ne and specialty chemical markets as drugs or pharmaceuticals; and (5) the conversion of undesired chemical or petroleum processing by-products into environmentally acceptable or recyclable products. An overview of the chemistry and process technology of these various application areas is provided in the monograph of Weissermel and Arpe (1993). The importance and contribution of the products *Corresponding autor. Fax: (314) 935-7211. E-mail address: dudu@wuche3. wustl.edu (M. P. Dudukovic) generated by the mutliphase reactor technology to the national economy of the United States is illustrated by the pie chart of Fig. 1. Due to lack of space, we are unable to discuss here the various emerging chemistries that demand multiphase reactor technology and will present such a discussion elsewhere. Instead, we focus here on three-phase multiphase reactors and attempt to describe our current understanding of them. This is becoming increasingly important for rapid commercialization of new technologies. The new paradigm of simultaneous catalyst and reactor development for new processes is becoming prevalent in modern chemical engineering (Villermaux, 1993; Lerou and Ng, 1996). To use this parallel approach, in addition to the "rm grasp of chemistry and catalysis, one needs to have a good knowledge of what various reactor types can and cannot do. Krishna and Sie (1994) advocated a simple but e!ective approach to multiphase reactor selection which examines the particle scale phenomena, phase contacting pattern and #ow, and the mixing pattern expected in a particular reactor from the point of view of their e!ect on the chemical pathways and energy requirements of the process under consideration. Such analysis can then guide the development of the catalyst with desirable properties and of the right size and shape to "t into the best reactor type. However, if 0009-2509/99/$ } see front matter ( 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 9 - 2 5 0 9 ( 9 8 ) 0 0 3 6 7 - 4 1976 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Fig. 1. Value generated by multiphase reactor technology. one is to attempt scale-up from laboratory scale to industrial scale, as the current economic climate increasingly demands, then one must assure either that the scale-up will be forgiving or one must have a profound and detailed understanding of the multiphase reactor that is being considered for scale-up. Hence, improved understanding of the #uid dynamics and transport processes in frequently used multiphase reactors is more important than ever for accomplishing large scale-up factors with con"dence. Lack of thorough understanding of the phenomena occurring in multiphase reactors can lead to disasters in scale-up or design. The price paid for such failures can ultimately be quite costly as the plant has to be used as a pilot or lab in seeking a way to improved performance. The need to quantify the performance of multiphase reactors leads to their modeling. A typical model of a multiphase reactor rests on the solution of the generic conservation equation (1) applied to species mass and energy of the system: A rate of output by phase i BA } rate of input by phase i BA net rate of tion used in modeling the reactor #ow pattern and mixing should be commensurate with the level of modeling used to understand the kinetics, i.e. species generation rate. Whenever that is not the case, the modeling e!ort yields less than maximum bene"ts. In addition to computation of molecular level events, which have become increasingly popular, the recent rapid advances in available software for computational #uid dynamics (e.g. CFDLIB, FLUENT, PHOENICS, FLOW 3D, FIDAP, etc.) make it possible to simulate the gross #ow patterns in large reactors. However, for multiphase #ows experimental veri"cation, at least via cold #ow models, is still needed due to the uncertainty of the closure forms used in the description of phase interaction terms. A review of the role of CFD in chemical reaction engineering appeared recently (Kuipers and Van Swaaij, 1997). It is clear from this review that two types of e!orts are encountered: (a) global system models, which typically provide the overall features of #ow in large reactors and are sometimes tied with various degrees of empiricism to transport and kinetics to describe reactor performance, and (b) detailed models that describe the phenomena on various scales from "rst principles. This second type of model cannot yet be implemented on multiphase reactor systems. In absence of detailed models for most multiphase reactor types and chemistries conducted in them, lower level models provide valuable tools in process development but still need experimental veri"cation. This brings us to the perennial problem in multiphase reactors which is that of scale-up, i.e., how to achieve the desired results in a large scale reactor based on observations made on the laboratory unit. All reaction engineers know that success of scale-up rests on our ability to understand and quantify the transport-kinetic interactions on a particle scale (or single eddy scale), interphase transport on particle and reactor scales, #ow pattern of each phase and phase contacting pattern and their changes with the changes in reactor scale and operating conditions. It is with the goal of providing such improved understanding of multiphase reactors that research on BA rate of BA rate of B ! interphase transport " generation ! accumulation . into phase i The sophistication of our reactor model depends at which level we treat the molecular, single eddy or catalyst particle, and reactor scale, as indicated in Table 1. Naturally, the more sophisticated the model the more expensive it is to develop and run. With that in mind, one simple rule should be followed. The level of sophistica- in phase i in phase i (1) #uid dynamics and transport in multiphase systems continues to be performed at an increased rate. We now consider how much progress has been made in the understanding of three-phase reactors. The importance of this topic is evident in the fact that three international conferences were held on M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 1977 Table 1 Levels of multiphase reactor modeling Modern reaction engineering requires handling phenomena over a multitude of scales: Molecular scale (kinetics) Eddy or particle scale (local transport phenomena) Reactor scale (#ow patterns, contacting and #ow regime) Possible level of description Molecular scale (rate forms) Strictly Mechanism Fundamental empirical based elementary D - - - - - - - - - - - - - - - - - - -D - - - - - - - - - - - - - - -D Eddy or particle scale transport Empirical Micromixing DNS CFD models D - - - - - - - - - - - - - - - - - - -D - - - - - - - - - - - - - - -D Empirical part Thiele modulus Rigorous of rate equation Reactor scale Ideal reactors Empirical Phenomenological CFD models models models D - - - - - - - - - - - - - - - - - - -D - - - - - - - - - - - - - - - D - - - - - - - - - - - - - - - - - - - - - D PFR, CSTR Axial dispersion Gas-Liquid-Solid Reactor Engineering; the "rst one in October of 1992 in Columbus, OH (Chem. Engng Sci. 47 (13/14) 1992), the second in Cambridge, England, in March, 1995 (Trans. IChemE. 73, 1995) and the third one in Osaka, Japan, in December, 1997 (Chem. Engng. Sci. 52, (21/22) 1997). There were also two highly interdisciplinary international symposia on Catalysis in Multiphase Reactors (I, Lyon, France, 7}9 December, 1994; II, Toulouse, France, 16}18 March, 1998) the proceedings of which appeared in the Journal of Applied Catalysis. Therefore, for additional information, and regarding multiphase reaction engineering topics that we do not manage to cover here, the reader is referred to the proceedings of ISCRE 13 and 14 published in Chemical Engng Sci. 49 (24A/B) and Vol 51(1/11), respectively, to ISCRE 15 and to the publications that resulted from the above-cited conferences. 2. Fixed beds with two-phase 6ow Packed-bed reactors processing gas and liquid reactants can operate in downward cocurrent two-phase #ow (trickle-bed reactors } TBR), in upward cocurrent #ow (packed-bubble columns } PBC) and in countercurrent #ow. The three modes of operation are illustrated in Figure 2 and the processes recently investigated in these reactor types are listed in Table 2. The importance of packed beds with two-phase #ow to the petroleum, petrochemical, chemical and other industries attracted nu- Fig. 2. Packed bed reactors for gas-liquid-solid catalyzed systems (from Mills and Chaudhari, 1997). (a) Trickle-bed with cocurrent down#ow. (b) Trickle-bed with countercurrent #ow. (c) Packed bed bubble #ow reactor with cocurrent up#ow. merous review papers. Among the more recent ones are the contributions by Zhukova et al. (1990), Gianetto and Specchia (1992), Martinez et al. (1994), Saroha and Nigam (1996) and Al-Dahhan et al. (1997). Here we mention only the newest results and "ndings that have been implemented in practice. 2.1. Fluid dynamics An extensive review of hydrodynamic and transport parameters for two-phase #ow systems in packed beds appeared recently (Al-Dahhan et al., 1997) and there is no point in repeating here the numerous tables and references that were provided in that review. We attempt here to summarize the key "ndings that ought to be of importance to the research and plant engineer. 2.1.1. Flow regimes It is well known that trickle beds can and do operate in the variety of #ow regimes ranging from spray #ow (liquid drops and continuous gas #ow), trickle #ow (continuous gas phase and one directional liquid rivulets and some discontinuous liquid "lms), pulse #ow (intermittent passage of gas- and liquid-rich zones through the reactor) and downward bubble #ow (continuous liquid and dispersed gas #ow). Similarly, cocurrent up#ow packed bubble columns can experience the so-called homogeneous and heterogeneous bubble #ow, while the onset of #ooding is of great importance in countercurrent #ow operation. While the existence of the various #ow regimes has been proven and many criteria have been proposed to delineate the regime boundaries (see Al-Dahhan et al., 1997) none of them is yet entirely successful in accomplishing such a task (Wild et al. (1991). Attempts have been made by Larachi et al. (1993) for high-pressure operations and Burghardt and Bartelmus 1978 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Table 2 Some recent applications of three-phase reactions carried out in TBR/PBC Residuum and vacuum residuum desulfurization for the production of low-sulfur fuel oils (Meyers, 1996) Hydrodesulfurization of atmospheric gas oil (Chen and Tsai, 1997) Catalytic dewaxing of lubestock cuts to produce fuel or lube products for extremely cold conditions (Meyers, 1996) Sweetening of diesel, kerosene, jet fuels, heating oils (Meyers, 1996) Hydrodemetallization of residues (Trambouze, 1993; Euzen et al., 1993; Chen and Hsu, 1997) Hydrocracking for production of high-quality middle-distillate fuels (Meyers, 1996; Landau et al., 1998) Hydrodenitri"cation (Meyers, 1996) Isocracking for the production of isopara$n-rich naphtha (Meyers, 1996) Production of lubricating oils (Meyers, 1996) Selective synthesis of wax from syngas (Fan et al., 1997) Selective hydrogenation of 1,5,9-cyclododecatriene (StuK ber et al., 1996), hydrogenation of C -ole"ns (Vergel et al., 1995), naphthalene (Huang and 4 Kang, 1995), 3-hydroxypropanal (Valerius et al., 1996), acetophenone (Bergault et al., 1997), maleic anhydride (Herrmann and Emig, 1997), a-nitromethyl-2-furanmethanol (Khadilkar et al., 1998c), 2,4-dinitrotoluene (Rajashekharam et al., 1998), Dicyclopentadiene (Chou et al., 1997), glucose (Tukac, 1997), nitrotoluene (Westerterp et al., 1997), a-methylstyrene (McManus et al., 1993; Lange et al., 1994; Castellari and Haure, 1995; Frank, 1996) Synthesis of butynediol from acetylene and aqueous formaldehyde (Gianetto and Specchia, 1992) VOC bio-scrubbers (Dicks and Ottengraf, 1991; Alonso et al., 1997; Rihn, et al., 1997; Laurenzis et al., 1998; WuK bker et al., 1998; Sto!els et al., 1998), VOC chemical abatement in air pollution control (Cheng and Chuang, 1992) Hydration of propene (Westerterp and Wammes, 1992), 2-methyl-2-butene (Goto et al., 1993) Wet air oxidation of waste water and model pollutant e%uents: phenol (Fortuny et al., 1995; Pintar et al., 1997; Alejandre et al., 1998), substituted phenols (Tukac and Hanika, 1997, 1998), n-propanol (Mazzarino et al., 1994) Oxidation of SO (Haure et al., 1990b; Kiared and Zoulalian, 1992; Ravindra et al., 1997); Oxidation of glucose (Tahraoui et al., 1992), poly(a-ole"n) 2 lubricant (Koh and Butt, 1995). (1996) and Burghardt et al. (1996) for organic systems. A priori prediction of foaming also remains elusive. A number of useful observations were summarized by Al-Dahhan et al. (1997): the trickle-to-pulsing transition is a function of gas density so that high pressure operation with light gases like hydrogen can be simulated via heavier gases like nitrogen at a much lower pressure; higher gas density broadens the trickle #ow regime while higher liquid denisty makes it narrower; hydrophobic packing broadens the trickle #ow regime (Horowitz et al., 1997), while non-Newtonian #uids cause the transition to pulsing at lower velocities (Iliuta and Thyrion, 1997). Novel experimental techniques are allowing us to collect more precise #ow regime data in trickle beds. Noteworthy are the micro electrode sensors used to detect wall shear and to elucidate the local #ow regime (Rode et al., 1994, 1995; Lati" et al., 1992a, b). Smooth signals were characteristic of trickle #ow, whereas high-frequency, high-amplitude #uctuations were observed in dispersed bubble #ow and in liquid slugs during pulse #ow. Based on these measurements the conclusion is reached that pulsing #ow represents a hybrid of trickle #ow and dispersed bubble #ow. 2.1.2. Pressure drop and liquid holdup Recent correlations and semi-theoretical models for prediction of two-phase pressure drop and liquid holdup at high-pressure operation have also been recently summarized by Al-Dahhan et al. (1997). No method emerges as clearly superior to others but those based on semitheoretical and phenomenological models seem more reliable than strictly empirical correlations. The e!ect of elevated pressure mainly manifests itself via increased gas density. Hence, high-pressure operation can be successfully simulated with gases of higher molecular weight at lower pressures. The following qualitative observations emerge. At a given density, the two-phase pressure drop increases with gas and liquid mass #uxes, super"cial velocities and liquid viscosity. Liquid holdup increases with liquid mass #ux and super"cial velocity, and liquid viscosity, but decreases with increasingly gas mass #ux or super"cial velocity. Hydrodynamic hysteresis may occur at high pressure when the liquid is contaminated with impurities, e.g. an antifoam agent. However, for common single-component liquids or liquid mixtures consisting of similar components, hysteresis is not detected at high pressure. For very low gas velocities (; (1}2 cm/s) G liquid holdup is pressure insensitive and equals the value determined at atmospheric pressure. At given super"cial velocities as gas density is increased, pressure drop increases and liquid holdup decreases. When the pressures of gases of di!erent molecular weights are set to have equal densities, identical pressure drops occur for the same #uid throughputs (see Fig. 5c in Al-Dahhan et al.,1997). Liquid holdup in PBC in bubble #ow is greater than in TBR in trickle #ow, whereas in pulse #ow, they tend to be quite close in values. For design purposes, PBC and TBR can be treated as hydrodynamically similar in the pulse #ow regime (Yang et al., 1992a). Recently, more detailed information about liquid holdup and the nature of liquid #ow in trickle beds has become available due to the increased use of non-invasive M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 sophisticated measurement techniques. For example, Reinecke and Mewes (1996, 1997), Reinecke et al. (1998) and Schmitz et al. (1997) used capacitance tomography imaging to capture the transient pattern of liquid #ow in a trickle bed. Toye et al. (1994, 1996, 1997) utilized X-ray transmission tomography to capture two-phase #ow distribution in trickle beds. It is expected that in the near future additional studies of this type will provide su$cient information on two-phase #ow structure in trickle beds to allow for veri"cation of more detailed models of #ow. Non-invasive measurement techniques that can be utilized in multiphase #ows have recently been summarized both in a book edited by Chaouki et al. (1997a) and in an extensive review article (Chaouki et al., 1997b). 2.1.3. Gas}liquid interfacial areas and interphase mass transfer coezcients Correlations and models for predicting gas}liquid interfacial areas and volumetric gas}liquid and liquid}solid mass transfer coe$cients in PBC/TBR were also summarized by Al-Dahhan et al. (1997). The scarcity of gas-side volumetric mass transfer coe$cients is noteworthy; and to the best of our knowledge no experimental data on k a are available for high-pressure G operation. Gas}liquid and liquid}solid mass transfer involving non-Newtonian liquids is also sparcely addressed in the literature (Iliuta et al., 1997a; Iliuta and Thyrion, 1997b). Considering the large number of biochemical processes that utilize PBCs and TBRs, this gap in knowledge needs to be "lled. The overwhelming majority of gas}liquid mass transfer parameters in TBR/PBC are derived based on the so-called chemical methods. A signi"cant step forward was achieved when these methods were adapted to measure mass transfer in pressurized vessels (Oyevaar et al., 1990). Soda or potash carbonation, sul"te oxidation and amine carbonation are known to be coalescence inhibiting systems which may cause problems in assessing mass transfer parameters in the high interaction regimes. There is a need to implement new gas}liquid chemical methods using coalescing systems, such as hydrazine oxidation (Lara-Marquez et al., 1994) to study gas-liquid mass transfer in TBRs and PBCs. From the experimentally determined gas}liquid interfacial areas and liquid-side volumetric mass transfer coe$cients at elevated pressure (Lara-Marquez et al., 1992; Wild et al., 1992; StuK ber et al., 1996; Molga and Westerterp, 1997a,b; Larachi et al., 1997a; Larachi et al., 1998a), the following qualitative observations can be made: at a given gas density, gas}liquid interfacial areas and volumetric liquid-side mass transfer coe$cients increase as liquid and gas mass #uxes or super"cial velocities increase; both mass transfer parameters improve in TBR/PBC as gas density increases for given gas and liquid super"cial velocities. 1979 2.1.4. Catalyst wetting During the past couple of decades it has been established that incomplete catalyst utilization may occur, especially in the trickle #ow regime, and that it has two main causes. One is reactor scale liquid maldistribution that may leave certain portions of the bed poorly irrigated. Proper design of liquid distributors, operation with packing that assures needed minimal pressure drop, and redistribution of the liquid in quench boxes and other devices can take care of this problem. Large-scale CFD computations are helpful in establishing the e!ect of the bed voidage variation and of the presence of internals on gross liquid distribution. The other cause of incomplete catalyst utilization is particle scale incomplete external wetting. This results from the fact that at su$ciently low liquid mass velocity the liquid #ow available is insu$cient to cover all the catalyst particles with a continuous liquid "lm at all times. In a time-averaged sense the external surface of the particle is then only partially covered by the #owing liquid. Correlations and models developed for liquid}solid contacting e$ciency (de"ned as the fraction of the external catalyst area covered by the #owing liquid "lm) have been summarized and discussed by Al-Dahhan et al. (1997). The ability of the Al-Dahhan and Dudukovic's (1995) correlation, which is the extension of the work done by El-Hisnawi (1981) to high pressure, to properly predict catalyst wetting and, hence, catalyst e!ectiveness and reactor performance has been documented by a number of studies performed by di!erent investigators (Beaudry et al., 1987; Wu et al., 1996a; Khadilkar et al., 1996; Llano et al., 1997). At "xed liquid mass #ux, and at high gas velocities, contacting e$ciency improves noticeably with the increase in pressure. Increased pressure drop and liquid mass velocity lead to increased contacting e$ciency also. Hence, both liquid and gas velocity increase the contacting e$ciency at high pressures. In scale-up and scale-down of TBRs it is highly desirable to run laboratory reactors at the well de"ned state of catalyst wetting (often complete wetting) while matching the LHSV of the large units. Close to complete external catalyst wetting can be achieved in up#ow reactors, at the expense of much larger liquid holdup than in the commercial scale TBR. This may be undesirable if side reactions occur in the liquid phase or if gas}liquid mass transfer rate is impaired by larger liquid "lm resistance in the small unit. An alternative is to run a laboratory trickle bed where the voids among catalyst particles are "lled with "nes. If proper packing procedure is used (Al-Dahhan et al., 1995; Al-Dahhan and Dudukovic, 1996) a bed packed with the mixture of catalyst and "nes decouples the apparent kinetics from hydrodynamics, which is desired. Packed beds containing "nes perform then identically in up#ow and down#ow at the same set of mass velocities (Al-Dahhan and Dudukovic, 1996). 1980 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Fig. 3. Prediction of external liquid holdup in low and high interaction regime. In summary, we can say that in spite of considerable research, the #uid dynamic parameters in packed beds with two-phase #ow cannot be predicted with desired accuracy. An engineer attempting to evaluate the hydrodynamic parameters needed for design or scale-up, such as external liquid holdup, #ow regime and pressure drop, has to select a suitable correlation. By &&suitable'' one usually means a correlation that in its data base contains operating conditions and physical system properties that are the closest to the system of interest. Can one not do better now at the turn of the millenium and recommend the best universal correlation? The answer unfortunately is negative. The ability (or lack of it) of the currently available methods to predict the key #uid dynamic parameters is illustrated in Fig. 3, which is a parity plot of the 8000 external liquid holdup data, collected from various sources in both low and high gas-liquid interaction regimes, against the predictions of the appropriate form of the empirical Ellman (1988) and Ellman et al. (1990) correlation. The lack of success is self-evident. We chose Ellman's correlation as an illustration not because we believe it is inferior to others, but on the contrary, because it covers the broadest data base at elevated pressure and, hence, is expected to be among the better choices. Clearly, Fig. 3 indicates the need for a renewed e!ort to reach more predictability in evaluation of two-phase #ow packed beds hydrodynamic parameters. One approach is to increase our reliance on fundamental approaches and utilize improved computational power to solve the resulting more complex #ow models. The other (perhaps parallel approach pursued by one of the authors (F.L.)) is to utilize the advances in computers and neural networks to train a neural net model based on a huge set of available data (F.L. has accumulated over 30,000 data for the #uid dynamic parameters discussed above) and make predictions based on such a model. Two recent papers by Bensetiti et al. (1997) and Larachi et al. (1998b) illustrate the possibilities of such an approach (see also AndreH , 1997). These authors show that if one selects randomly about 60% of the available data, a neural net can be trained to achieve a remarkable "t of the training set. The advantage arises that when the neural net predictions are tested against the remaining 40% of the data very good agreement is found. This is illustrated in Fig. 4 for mass transfer coe$cient. Needless to say the classical correlations without neural nets provided the quality of "t observed for holdup in Fig. 3. 2.2. Comparison of upyow packed bubble columns (PBC) and down-ow trickle bed reactors (¹BR) When a "xed bed is chosen to process gas and liquid reactants the question whether to use up#ow or down#ow Fig. 4. Neural network based predictions of mass transfer coe$cients (a) Training set. (b) Comparison with other data. M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 operation is frequently asked. Liquid holdup is higher and liquid is typically the continuous phase in the former, while gas is the continuous phase in TBR and liquid holdup is lower. Goto and Mabuchi (1984) demonstrated that for the atmospheric pressure oxidation of ethanol in presence of carbonate, down#ow is superior at low gas and liquid velocities but up#ow should be chosen at high gas and liquid velocities. Beaudry et al. (1987) studied atmospheric pressure hydrogenation of a-methylstyrene in liquid solvents at high liquid reactant concentration in the feed and observed that down#ow performance is better than up#ow except at very high liquid reactant conversion. Mazzarino et al. (1989) observed higher rates in up#ow than in down#ow for ethanol oxidation and attributed the observed phenomenon to better e!ective wetting in up#ow. Liquid holdup measurements at elevated pressure using water/glycol as liquid with H , N , 2 2 CO as the gas phase by Larachi et al. (1991) indicate 2 that liquid saturation is much greater in up#ow than in downward #ow at all pressures (up to 5.1 MPa). LaraMarquez et al. (1992) studied the e!ect of pressure on up#ow and down#ow using chemical absorption, and concluded that the interfacial area and the liquid side mass transfer coe$cient increase with pressure in both cases. Goto et al. (1993) observed that down#ow is better than up#ow at atmospheric pressure (for hydration of ole"ns) and noted that the observed rates in down#ow were independent of gas velocity while those in up#ow were slightly dependent on it. In order to provide general guidance to practicing engineers as to which reactor type to choose, Khadilkar et al. (1996) examined all the previously reported studies. They concluded that most reaction systems can be classi"ed as being liquid reactant or gas reactant limited. The 1981 value of parameter c, wh ich represents the ratio of the liquid reactant #ux to the catalyst particle to the gas reactant #ux to the particle, scaled by the ratio of stoichiometric coe$cients, delineates these two categories. For cA1 the reaction can be considered gas reactant rate limited, while for c(1 it is the liquid reactant that is rate limiting. For liquid-limited reactions up#ow reactor should be preferred as it provides for complete catalyst wetting and for the fastest transport of the liquid reactant to the catalyst. For gas limited reactions, down#ow reactor, especially at partially wetted conditions, is to be preferred as it facilitates the transport of the gaseous reactant to the catalyst. Applying this criterion to the previously reported studies in the literature, the conclusions regarding the preferred mode of operation can be reached and are tabulated in the last column of Table 3. This agreed with all experimental observations except the one by Mazzarino et al. (1989) at low pressure and high liquid reactant concentration. This observation is suspect because the comparison between &&up#ow and down#ow'' performance was not executed with the same catalyst bed. To further illustrate the usefulness of the proposed criterion, Khadilkar et al. (1996) and Wu et al. (1996a) conducted an experimental study of hydrogenation of a-methylstyrene on the same catalyst bed using up#ow and down#ow mode of operation. By changing hydrogen pressure and feed a-methylstyrene concentration they were able to run the reaction as gas reactant limited (c"8.8 at high feed liquid reactant concentration and at atmospheric hydrogen pressure) and as liquid reactant limited (c"0.87 at high hydrogen pressure and low feed a-methylstyrene concentration). The experimental results con"rmed the predictions based on the value of c which indicate that down#ow is preferred for the gas limited reaction and up#ow for the liquid limited Table 3 Identi"cation of the limiting reactant for literature data Authors Reaction system Goto and Mabuchi (1984) Mills et al. (1987) Oxidation of ethanol in presence of carbonate Hydrogenation of alphamethylstyrene Mazzarino et al. (1989) I. Ethanol oxidation Operating conditionsa Low concentration and atmospheric pressure High concentration low pressure Gamma (c) Limiting reactant Preferred mode 314 Gas Down#ow 92 Gas Down#ow Liquid Up#ow Gas Down#ow Gas Down#ow Gas Down#ow Liquid Up#ow Low concentration and 0.51 atmospheric pressure II. Ethanol oxidation High concentration and low 17 atmospheric pressure Goto et al. (1993) Oxidation of ethanol in presence Atmospheric pressure 10300 of carbonate Khadilkar et al. (1996); I. Hydrogenation of alphaHigh concentration low pressure 8.8 Wu et al. (1996a) methylstyrene II. Hydrogenation of alphaLow concentration high pressure 0.87 methylstyrene aConcentration refers to liquid reactant feed concentration. 1982 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 one. Moreover, it was shown that when the bed is packed with "nes the di!erences between up#ow and down#ow disappear completely as transport e!ects in both modes of operation become identical (Wu et al., 1996b). 2.3. Unsteady-state operation of trickle-bed reactors The concept of using unsteady state operation to enhance performance is not new to the "eld of chemical engineering. In case of trickle-bed reactors, however, unsteady-state operation has been considered only in the past decade or so and several strategies such as modulation of #ow, composition, and activity have been suggested (Silveston, 1990). Modulation of #ow of gas or liquid is done to achieve the desired ratio of liquid and gaseous reactants on the catalyst as well as to allow a controlled exotherm (Gupta, 1985; Haure et al., 1990a; Lee et al., 1995). Modulation of composition can improve selectivity or control phase change by addition of inerts or products (Lange et al., 1994) or by injecting cold shots of gas (Yan, 1980). Modulation of activity is usually accomplished by an extra component, which can help catalyst regeneration and prevent build up of poisons or inhibitors in the catalyst (Chanchlani et al., 1994; Haure et al., 1990a). The experimental studies of unsteady-state operation in trickle-bed reactors are summarized in Table 4 and only key observations are brie#y discussed here. The terminology used is that the total time of one cycle is referred to as cycle time (or period, denoted as q) and the part of the cycle when modulation is active is referred to as the ON part (denoted by sq, where s is the fraction of total time corresponding to the ON part) and the rest of the cycle is the OFF part (corresponding to (1!s)q). Haure et al. (1990b) and Lee et al. (1995) studied periodic #ow modulation of water in SO oxidation to obtain 2 concentrated sulfuric acid from dilute SO gaseous 2 streams. They observed an enhancement in supply of SO and O to the catalyst during the OFF part of the 2 2 cycle, resulting in higher performance and temperature rise of 10}153C. They also observed that the reaction results in formation of SO which is adsorbed on the 3 catalyst until it is washed by the pulse of water during the ON part of the cycle, which results in concentrated sulfuric acid formation as well as restoration of the catalytic activity. Lange et al. (1994) experimentally investigated the hydrogenation of cyclohexene, and the hydrogenation of a-methylstyrene on Pd catalysts by manipulation of liquid feed concentration and feed rate, respectively. They used non-isothermal composition modulation of cyclohexene to control conversion and keep the reaction system from switching from a three-phase system to a two-phase one, and, designed their total cycle time based on this criterion. For the case of hydrogenation of a-methylstyrene under isothermal conditions, the authors observed maximum improvement at a cycle period of 8 min at cycle split of 0.5. The observed improvement (between 2 and 15%) was attributed to better wetting due to the liquid pulse which caused the removal of stagnant liquid. Castellari and Haure (1995) investigated the performance enhancement due to the large temperature rise during the OFF part of the cycle. They observed gas-phase reaction at semi-runaway conditions and a large enhancement resulting from the high gasphase reaction rates. Most of the studies reported in the open literature are for gas-limited conditions. They indicate that periodic operation under gas-limited conditions can ensure completely internally wetted catalyst pellets, provide direct access of gaseous reactant to the catalyst sites, replenishment of catalyst with liquid reactant, periodic removal of products by fresh liquid, and quenching of a predetermined rise in temperature. Under liquid-limited conditions, catalyst external wetting and liquid supply to the particles is crucial, and periodic operation can reduce and eliminate liquid maldistribution, ensure a completely irrigated bed, and, quench developing hotspots. Several industrial reactors are operated under liquid-limited Table 4 Literature studies on unsteady state operation of trickle beds L and G #ow rates Author(s) System studied Modulation strategy Haure et al. (1990a) SO oxidation 2 Lange et al. (1994) Cyclohexene hydrogenation a-MS hydrogenation Stegasov et al. (1994) SO oxidation 2 Lee et al. (1995) SO oxidation 2 Castellari and Haure (1995) a-MS hydrogenation Flow (non isothermal) < "0.03}1.75 mm/s L < "1}2 cm/s G Composition Q "80}250 ml/h L (non isothermal) Conc"5}100% Liquid #ow Q "0}300 ml/h, L (isothermal) Q "20 l/h G Model < "0.1}0.5 cm/s, L < "1.7}2.5 cm/s G Adiabatic #ow < "0.085}0.212 cm/s, L modulation S< "1000 h~1 G Non isothermal Q "2.27 ml/s L Q "900 ml/s G Cycle period (q) and split (p) % Enhancement q"10}80 min (p"0.1, 0}0.5) q"up to 30 min. (p"0.2}0.5) q"1}10 min (p"0.25}0.5) q"10}30 min. (p"0.1}0.5) q"up to 60 min (p"0.02}0.1) q"5 to 45 min (p"0.3}0.5) 30}50% 2}15% (temp rise"303C) Max"80% 400% (temp. rise "353C) M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 conditions at high pressure and su!er from maldistribution of the liquid reactants, which can cause externally dry or even internally dry catalyst pellets. At high liquid and gas mass velocities, in the pulsing #ow regime, a signi"cant improvement in catalyst wetting and e!ective removal of hot spots has been reported (Blok and Drinkenburg, 1982). However, achieving this regime is not always practical in industrial reactors due to large pressure drop and little control over the slugging process. Periodic #ow modulation, with a low base #ow and a periodic slug of very high liquid #ow, can improve catalyst utilization even at low mean liquid #ows (lower pressure drop) and still achieve temperature and #ow control due to arti"cially induced pulses (or slugs). No study has been reported in the open literature on liquidlimited reactions or on unsteady-state performance data of large reactors. At ISCRE 15, Khadilkar et al. (1998a) presented the "rst experimental data for the e!ect of #ow modulation on performance of a trickle-bed reactor for a liquid-limited reaction. The e!ect of the natural pulsing #ow regime as opposed to the trickle #ow regime on selectivity has also been investigated recently by Wu (1997). Some industrial processes do employ periodic localized quenching of hot spots by injection of cold #uids at selected axial locations along the reactor (Yan, 1980). 2.4. Modeling TBR performance Most of the trickle-bed reactor models reported in the literature considered isothermal operation and used either a pseudo-homogeneous approach (Collins et al., 1984; Kheshgi et al., 1992) or a heterogeneous model with plug #ow for gas and liquid phase (El-Hisnawi et al., 1981; Mills and Dudukovic, 1984; Hekmat and Vortmeyer, 1994; Rajashekharam et al., 1998). Some models accounted for liquid #ow non-uniformity and maldistribution by using an axial dispersion model (Chu and Ng, 1986). Most investigations dealt with hydrogenation or oxidation in pure or moderately concentrated organic or aqueous solutions (large excess of liquid reactant), and, hence, considered zero-order rate with respect to the liquid reactant concentration and "rst order with respect to dissolved gaseous reactant concentration. Liquid reactants/solvents were assumed to be non-volatile and gas phase assumed to be pure at constant partial pressure of the reacting gas. Thus, the primary model variables of interest have been the dissolved liquid-phase concentrations of the gaseous reactant and the conversion of the liquid-phase reactants. The key e!ect that was incorporated in most recent models was that of partial wetting and transport of gaseous reactant to dry external areas of the catalyst resulting in higher rates observed in most of the experimental data (ElHisnawi et al., 1981; Berruti et al., 1984; Ruzicka and Hanika, 1994). Some models considered non-isothermal e!ects and used a pseudo-homogeneous energy balance 1983 to solve for the temperature at any axial location (Yang and Li, 1992; Harold and Watson, 1993; Rajashekaram et al., 1998). Others considered evaporation e!ects by adding vapor-liquid equilibrium calculations and #ash units to simpli"ed pseudo-homogeneous or equilibrium model mass balance equations on the reactor scale (LaVopa and Satte"eld, 1988; Collins et al., 1984). Other approaches include a cell model (Sims et al., 1994), a cross-#ow model (Tsamatsoulis and Papayannakos, 1995) and some other models based on liquid #ow maldistribution (Funk et al., 1990) or stagnant liquid zones in the reactor (Rajashekharam et al., 1998). Table 5 summarizes the application of TBR/PBC models in interpretation of mainly laboratory-scale reactor performance. Pellet-scale reaction and di!usion have been studied by taking reactant limitation in account in simpler versions (Beaudry et al., 1987), and in the general case by considering partial internal wetting of pellets, resulting in gas and liquid-phase reaction zones, and solving for the gas}liquid interface by considering liquid inbibition, pore "ling and capillary condensation (Harold and Watson, 1993). Approximate solutions of the gas}solid catalyst level equations have also been veri"ed by numerical solution for non-linear kinetics (Lemco! et al., 1988). The earliest unsteady-state modeling used a plug-#ow equilibrium model for predicting the hot spot formation and movement during start-up of a trickle bed and investigated the e!ect of a gas/liquid quench stream axial position on the developing hot spot (Yan, 1980). Pseudotransient behavior was also modeled by considering similar equations (Warna and Salmi, 1996; Sundmacher and Ho!mann, 1994). Mass transfer terms are considered in extension of these models to predict periodic variation of temperature and concentration (Haure et al., 1990a; Stegasov et al., 1994). Spatial terms were dropped in some subcases of this model to study time variation of mass transfer coe$cients and enhancement in rates and selectivity for the model reaction system (Wu et al., 1995). Catalyst wetting e!ects during periodic operation (Gabarain et al., 1997a, b) were also studied with elimination of spatial terms in the model equations. This was done primarily to reduce computational complexity. Activity modulation was incorporated in recent transient models for optimizing the performance on the basis of catalyst activity (Yamada and Goto, 1997). The level of complexity and features available in the models in the literature are su$cient for evaluation of steady-state experiments in comparison of trickle beds and packed bubble columns as outlined previously. These models are still far from mimicking reality in industrial hydrocracking and hydrotreating applications due to three main shortcomings. They do not consider multicomponent transport and multiple reactions properly, do not account for change of phase (evaporation and condensation) and for its e!ect on holdup and velocities. An improved model for unsteady-state 1984 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Table 5 Application of TBR/PBC models to laboratory studies Reaction Rate analysis Model assumptions Source/reactor H O decomposition 2 2 Linear kinetics Sims et al.(1994)/TBR Hydrogenation of C -ole"ns 4 Hydrogenation of 3-hydroxy-propanal L-H kinetics L-H kinetics Hydrotreating of vaccum gas oil H O decomposition 2 2 Hydrogenation of a-Me-styrene L-H kinetics Linear kinetics L-H kinetics Selective hydrogenation of 1,5,9-cyclododecatriene SO oxidation 2 Phenol oxidation Linear kinetics SO oxidation 2 L-H kinetics Phenol biodegradation Toluene bioscrubbing Hydrogenation of a-Me-styrene Hydrogenation of acetophenone Haldane kinetics Monod kinetics Linear kinetics L-H kinetics Hydrogenation of unsaturated ketones in supercritical CO 2 Hydrogenation of 3-hydroxypropanal Power law kinetics Isothermal, partial wetting, 2-region cell reactor Isothermal, plug #ow Isothermal, plug #ow, partial wetting, heat balance Isotherm., plug #ow, partial wetting Isotherm., plug #ow, partial wetting Isothermal, plug #ow, partial wetting, high pressure Isothermal, axial dispersion, high pressure/temperature Isothermal, full wetting Isothermal, full wetting, plug #ow, high pressure/temperature Isothermal, partial wetting, axial dispersion, static-dynamic Isotherm., plug #ow, static-dynamic } Isothermal, plug #ow, partial wetting Non-isothermal, plug #ow, full wetting, high press./temp. Non-isothermal, plug #ow, full wetting Hydrogenation of 2,4-dinitrotoluene L-H kinetics Hydrogenation of a-nitromethyl2-furanmethanol Oxidation of substituted phenols Hydrodesulfurization of atmospheric residue Hydrogenation maleic anhydride L-H kinetics Linear kinetics L-H kinetics L-H kinetics Linear kinetics Power law kinetics L-H kinetics Non-isothermal, deactivation, partial wetting, plug #ow Non-isothermal, plug #ow, partial wetting, stagnant liquid Isothermal, plug #ow, partial wetting Isothermal, partial wetting Non-isothermal, plug #ow, deactivation, complete wetting Isotherm., axial dispersion, full wetting operation that removes many of the above de"ciencies has been developed and is presented at ISCRE 15 (Khadilkar et al., 1998b). 2.5. Packed beds with countercurrent yow Conventional gas}liquid absorbers have traditionally operated in this mode in order to maximize the driving force for gas}liquid mass transfer. In multiphase reactors of this type precise estimates of liquid holdup, pressure drop and mass transfer coe$cients are di$cult to make because the extensive data banks, utilized by the correlations for these parameters, do not include data for the small porous catalyst packing used in packed bed reactors with two phase #ow. Qualitatively, of course, one knows that pressure drop and holdup are intimately related and that an increase in one leads to the increase in the other. Flooding by and large follows the Sherwood type of correlation but detailed and accurate predictions of holdup, pressure drop and #ooding conditions may be elusive on most catalyst packing of interest. In order to Vergel et al. (1995)/PBC Valerius et al. (1996)/TBR Korsten, Hofmann (1996)/TBR Wu et al. (1996a)/TBR Khadilkar et al. (1996)/TBR, PBC StuK ber et al. (1996)/ PBC Ravindra et al. (1997)/TBR Pintar et al. (1997)/TBR Iliuta and Iliuta (1997)/TBR, PBC Iliuta (1997)/TBR, PBC Alonso et al. (1997)/TBR Castellari et al. (1997)/TBR Bergault et al. (1997)/TBR Devetta et al. (1997)/TBR Zhu and Hofmann (1997)/TBR Rajashekharam et al. (1998)/TBR Khadilkar et al. (1998c)/TBR Jiang et al. (1998)/TBR Tukac and Hanika (1998)/TBR Lababidi et al. (1998)/TBR Herrmann, Emig (1998)/PBC lower the pressure drop, high voidage packing or packing with special characteristics is preferred. The possibility that countercurrent #ow packed beds will be implemented in re"nery operations provides a strong motivation for investigating new types of structural packing with low-pressure drops and good gas}liquid and liquid}solid contacting. Structural packing for countercurrent #ow containing three porosity levels was recently reported by Van Hasselt et al. (1997), while Sie and Lebens (1998) illustrated the application of monoliths. Both reactors featured low-pressure drop compared to randomly packed beds. Flow transients, pressure drop overshoots and pressure drop hysteresis in countercurrent packed beds was recently studied by Stanek and Jiriczny (1997), Jiriczny and Stanek (1996) and Wang et al. (1997), respectively. Iliuta et al. (1997b) compared hydrodynamic parameters in cocurrent and countercurrent #ow. The introduction of countercurrent #ow "xed-bed reactors in a number of re"ning operations is likely, either via re-design of existing reactors or by introduction of M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 new process technology. The goal is not improvement in reactant (hydrogen) mass transfer, which is not rate limiting, but enhanced removal of inhibitory by-products or in situ product separation. Dassori et al. (1998) have illustrated the advantage in hydrodesulfurization. ABB Lummus is marketing such a technology and other studies involving this concept have been reported. 2.6. Concluding remarks During the last decade or so our understanding of catalytic packed beds with two-phase #ow has improved considerably. These are now recognized as reactors of choice when large catalyst to liquid volume ratio is desired, and when plug #ow of both phases is to be preferred, when reaction rates are not overly high and catalyst deactivation is very slow or negligible. It has also been accepted that in trickle #ow both reactor scale maldistribution can occur as well as incomplete external wetting of particles. To combat the former phenomenon, liquid redistribution is needed or induced pulsing #ow. The phenomenon of incomplete external catalyst wetting is detrimental to liquid-limited reactors only. It is now also understood that for liquid limited reactions scale-up at constant LHSV is forgiving since it results in improved wetting e$ciency, and better catalyst utilization. For gas limited reactions such scale-up at constant LHSV can lead to very poor performance (Dudukovic, 1998) as the catalyst e!ectiveness factor drops with increased contacting e$ciency due to a reduction in the gas reactant supply. Hence, for gas-limited reactions constant LHSV and constant reactor height are required in order to maintain the same performance upon scale-up. This leads sometimes to undesirable pan-cake reactor geometry which can be a problem in achieving uniform liquid distribution and hence model based scale-up ought to be used. By addition of "nes to the laboratory catalyst beds #uid dynamics can be separated from kinetics and transfer of laboratory data to industrial practice becomes possible. For well-established liquid-reactant-limited processes scale-up and scale-down between laboratory reactors and large industrial units can be accomplished. The choice of up#ow vs. down#ow reactors can be based on rational considerations as to what is the limiting reactant at the operating conditions of interest. As already mentioned countercurrent #ow will become more prominent in the future in processes that su!er from by-product catalyst inhibition. The available correlations for important hydrodynamic parameters leave a lot to be desired. As the use of novel structural packing becomes more widespread it will become increasingly necessary to re-establish engineering type of correlations for such packing. It is hoped that fundamental approaches involving CFD and proper description of multiphase mass transfer will also be increasingly used. 1985 3. Reactors with moving catalyst 3.1. Bubble columns and slurry bubble columns Bubble columns and slurry bubble columns are used extensively in a variety of processes for hydrogenation, oxidation, chlorination, hydroformylation, cell growth, bioremediation, etc. Recently they have been identi"ed as reactors of choice for gas conversion (e.g. liquid phase methanol synthesis, Fischer}Tropsch synthesis, etc.) due to their excellent heat transfer characteristics. Fig. 5 schematically represents a typical bubble column reactor (minus the internals needed for heat transfer). Gas is sparged at the bottom of the column and the resulting buoyancy driven #ow creates strong liquid recirculation. Thus, as long as the liquid super"cial velocity is an order of magnitude smaller than that of the gas, it is the gas super"cial velocity that is the dominant variable which drives the #uid dynamics of the whole system, and whether the liquid is processed batch-wise or #ows cocurrently or countercurrently to the #ow of the gas is immaterial from the #uid dynamics point of view. Slurry particles, as long as they are small (typically less than 60 lm) follow liquid motion except perhaps at very high slurry loadings exceeding 20}30%. While in some applications bubbly #ow is practiced (typically gas super"cial velocities smaller than 2}3 cm/s) of current industrial interest is the churn-turbulent #ow (with gas super"cial Fig. 5. Schematic of a bubble column. 1986 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 velocities in excess of 10 cm/s up to the 30}50 cm/s range). 3.1.1. Fluid dynamics Recent advances in bubble column #uid dynamics have resulted from novel measurements and computational modeling e!orts. Hot-wire anemometry (HWA) was used by Menzel et al. (1990) successfully to map the velocity as well as the turbulent stress "eld in three dimensional (3D) bubble columns up to reasonable gas velocities of 8 cm/s. Yang et al. (1990) also measured the time-averaged gas and liquid velocity distributions in 3D columns. L.S. Fan introduced the use of particle image velocimetry (PIV) to 2D and 3D bubble columns in bubbly #ow (Tzeng et al., 1993; Reese et al., 1993, 1996; Reese and Fan, 1994, 1997a; Mudde et al., 1997). They mapped the instantaneous velocity and holdup "elds, as well as the turbulent stresses, in 2D columns and showed good comparison with the volume-of-#uid computational predictions. The same group also developed a #ow visualization experiment at high pressure and generated an extensive correlation for bubble rise velocity and size as a function of operating conditions. The group at Delft (GroK en et al., 1995, 1996) implemented a novel "ber optic probe for bubble columns for examination of bubble size and rise velocity and mapped via LDA (Laser Doppler Anemometry) the Reynolds stresses in a 3D column close to the wall. At the Chemical Reaction Engineering Laboratory (CREL) at Washington University (Devanathan et al., 1990, 1995; Devanathan, 1991; Moslemian et al., 1992; Yang et al., 1992b, 1993; Kumar et al., 1994, 1995a, b, 1997; Dudukovic et al., 1997) computer-automated radioactive particle tracking (CARPT) and computed tomography (CT) were implemented for complete mapping of the velocity and holdup "eld in bubble columns. CARPT allows us to map the Lagrangian tracer particle trajectories throughout the column, and from these trajectories determine instantaneous velocities, time averaged #ow patterns, turbulent stresses and turbulent kinetic energy due to measured #uctuating velocities. From CARPT data mixing parameters such as the eddy di!usivity tensor are also readily calculated. The principles of CARPT (also called radioactive particle tracking, RPT) have been reviewed in detail by Larachi et al. (1997b) and this will not be repeated here. The interested reader is directed to the above-cited chapter and to the many references within it or to the above papers related to CARPT. Very brie#y, in CARPT the position of the single radioactive particle is continuously monitored by a series of pre-calibrated scintillation detectors. The particle is made of the same size and mass as the particles in the system, if motion of solids is monitored in slurries or #uidized beds, or it is neutrally buoyant when tracing the liquid motion. It can be shown that motion up to frequencies of 20}30 Hz can be followed. The gamma ray tomography setup in CREL allows one to obtain time-averaged holdup-pro"les in column cross sections at desired elevations. The CARPT-CT combined setup provides unique capabilities for mapping the #ow "eld in the whole enclosure (column) for opaque systems when other techniques fail. The CARPT-CT data have provided a unique view of the time-averaged #ow "eld and gas holdup distribution in bubble columns. While in bubbly #ow at low gas super"cial velocities the radial gas holdup pro"le is almost #at (with somewhat more gas in the center), in churn turbulent #ow the gas holdup pro"le is almost parabolic. The non-uniform gas holdup pro"le drives liquid circulation and throughout most of the column, except in the distributor region and in the disengagement zone, the liquid rises in the center and falls by the walls. The instantaneous #ow patterns are complex and involve toroidal, swirling vortex structures. CARPT provides information on the turbulence intensity, the anisotropy of turbulence and axial and radial di!usivities (Devanathan et al., 1990; Degaleesan, 1997; Yang et al., 1992b, 1993). The CARPT-CT have been used to relate the axial dispersion coe$cient to the measured liquid recirculation and eddy di!usivities (Degaleesan, 1997; Degaleesan and Dudukovic, 1998). Based on the hydrodynamic behavior that the data reveal, a recycle with cross-#ow with dispersion model was developed and used successfully for interpretation of tracer data (Degaleesan et al., 1996). The ensemble averaged liquid velocities and eddy di!usivities determined by CARPT and time-averaged holdup pro"les obtained by CT were used in the convection-di!usion model to predict the residence time distribution of a liquid tracer (Degaleesan et al., 1997). 3.1.2. CFD models The simplest one-dimensional model relates the gas holdup pro"le to the radial pro"le of the axial velocity in the fully developed #ow region. Kumar et al. (1995a) have shown that existing correlations for turbulent viscosity and mixing length yield inaccurate velocity predictions, given the gas holdup pro"le. Degaleesan et al. (1997) provided an improved approach to such predictions. Two-dimensional models for gas}liquid #ow in bubble columns have also been studied extensively. A recent review by Jakobsen et al. (1997) covers the pertinent literature well. Two approaches are basically used: the Euler}Euler formulation, based on the interpenetrating two-#uid model, and the Lagrange}Euler approach. In the former Navier}Stokes equations are ensemble averaged using the approach of Drew (1983). Expressions for all interphase interaction terms are then required, and these mainly consist of the models for the drag, lift and added mass force. Also a turbulence model is required for the liquid phase (and perhaps gas phase at higher pressures). The Lagrange}Euler method solves the original Navier}Stokes equations for the continuous phase, (the M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 density and viscosity of which are often modi"ed to account for the presence of the low volume fraction of the dispersed phase) and then solves for the motion of each bubble by applying Newton's second law to it where all the forces on the bubbles are calculated based on the local velocity patterns in the continuous phase. This approach, while it appears at "rst glance &&more fundamental'', hides in the di!erent realizations that appeared in the literature some additional tuning parameters (e.g. e!ective di!usivity for the dispersed phase, e!ective viscosity of the continuous phase, etc.). Both approaches have their ardent advocates, and each approach has its advantages and disadvantages. The Lagrange}Euler approach seems quite appealing for bubbly #ow, but even at those situations it has not been documented that it can handle coalescence-redispersion at increased volume fraction of the dispersed phase. It is rather dubious that the Lagrange}Euler approach can be used in churn-turbulent #ow where at very high gas holdup of 25}50% no individual bubbles preserve their identity for long and where liquid and gas essentially battle for the available space. The Euler}Euler interpenetrating two-#uid model seems much more attractive under those conditions; unfortunately it is not clear yet what are the appropriate forms to use for the drag, lift and virtual mass under such conditions. Appropriate models for multiphase turbulence also remain elusive. In the chemical reaction engineering literature it was Professor Svenden's group at Trondheim (Torvik and Svendsen, 1990; Jakobsen et al., 1997; Jakobsen, 1993) that were the "rst to develop steady-state Euler}Euler #uid dynamic 2D models for bubble columns. Such models show reasonable agreement with data for time-averaged axial velocity pro"les and somewhat less favorable agreement with radial holdup pro"les obtained in presumably axisymmetric 3D columns. They even tied the computed #ow "eld to predictions of reactor performance. Lapin and LuK bbert (1994) introduced the Lagrange}Euler description to the simulation of bubbly #ows in 3D columns and presented impressive transient velocity and holdup pro"les, which qualitatively compared well with observations, and also showed semiquantitative agreement with measured mean values. Sokolichin and Eigenberger (1994) used the direct solution of Navier}Stokes equations for the liquid and gas and presented reasonable agreement with selected experimental studies. Recently, Delnoij et al. (1997a}c) developed a more detailed model for dispersed gas}liquid two-phase #ow based on Euler}Lagrangian approach. All relevant forces (drag, virtual mass, lift and gravity) acting on the bubble are accounted for. Direct bubble}bubble interactions are also accounted for via an interaction model that resembles the collision approach followed in #uidized bed modeling. With this model Delnoij et al. (1997c) were able to simulate reasonably well the experimental observations of Becker et al. (1994), 1987 who monitored a gas plume created by a few clustered ori"ces at the bottom of a 2D column. In addition to the above-described methods, Tomiyama et al. (1993) used the volume of #uid method (which allows tracking of the gas-liquid interface) to analyze the shape and motion of a single rising bubble in liquid. Recently, Lin et al. (1996) applied the VOF to study the time dependent bubbly #ows at low gas holdup and compared their computational results with experimental data obtained with Particle Image Velocimetry. Several bubbles emanating from a small number of ori"ces were tracked by VOF and satisfactory agreement with experiments were reported. It should be mentioned, however, that most of the comparisons between CFD model predictions and data were qualitative or semi-quantitative in nature. Successful quantitative comparison of the time-averaged velocity pro"les based on 2D axisymmetric Euler}Euler model (CFDLIB of Los Alamos was used for computations) and 3D data obtained by CARPT was reached (Kumar et al., 1995b) but the model was not truly predictive as the assumed bubble size for drag computations and turbulent viscosity could be adjusted. Moreover, no amount of adjustments could reconcile the experimentally measured gas holdup pro"les via CT, which showed the customary maximum in the center, and the computed ones which indicate a peak in between the center and the wall but closer to the wall. Some were inclined to blame the 2D nature of the model for the inability to capture the spiraling gas plumes, and hence the correct gas holdup pro"les, others doubted the adequacy of the models used for drag, lift, virtual mass and turbulence. This issue remains unresolved. 3.1.3. Bubble size The treatment of bubble column #uid dynamics would not be complete without discussing the bubble size distribution. Based on the dynamic gas disengagement technique in 3D columns and visual observations in 2D columns, Krishna and his co-workers have advocated a bimodal bubble size distribution in churn-turbulent #ow (Krishna et al., 1993; Ellenberger and Krishna, 1994; Krishna and Ellenberger, 1996). However, it is suspected that dynamic disengagement does not capture the true distribution of bubble sizes, because of the fact that neither liquid circulation nor bubble coalescence and redispersion die out as the gas #ow is cut o!. Hence, no simple relationships exist between the rate of drop of the free surface of the gas}liquid dispersion and bubble sizes that are disengaging. Moreover, visual experiments at high pressure shed some doubts as to whether two classes of bubbles indeed exist at high pressure. This issue is important as it a!ects how the bubble column reactors are modeled and should be resolved. In summary, while advanced 2D and 3D models of bubble column two-phase #ows have been developed, 1988 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 experimental veri"cation is still needed. This is especially true of churn turbulent #ows. No fundamental model for mass transfer has yet been coupled successfully to the #ow models and reliable reactor performance predictions based on these models are not imminent. However, improved knowledge of the hydrodynamics is helping the practicing engineer develop improved phenomenological models for assessment of reactor performance. As far as experimental veri"cation is concerned, PIV, LDA, HWA are "ne tools for dilute dispersed #ow systems but in churn turbulent bubble columns one needs to rely on CARPT, gamma ray CT, X-ray tomography and possibly in the future on impedance tomography. 3.2. Three-phase -uidized-bed reactors Gas}liquid}solid #uidized-bed reactors are receiving considerable attention in research and process development. They are an o!-shoot of slurry bubble columns except that the particles are now su$ciently large that they behave as a distinct third phase. Besides their traditional applications in hydrotreating, Fischer}Tropsch synthesis, coal combustion, etc., three-phase #uidized beds or ebullated beds are also considered as viable options in the "elds of aerobic and anaerobic waste water treatment, as well as in the production of valuable substances by means of bacteria, fungi, animal and plant cells (Godia and Sola, 1995; Wright and Raper, 1996; SchuK gerl, 1997). 3.2.1. Fluid dynamics With the advent of three-dimensional particle image velocimetry (3-DPIV) and radioactive particle tracking techniques (CARPT, RPT) in gas}liquid}solid #ows, it has become possible to map the 2D and 3D full-"eld of the instantaneous and time-averaged phase holdup and velocity distributions, and to capture more quantitatively the phenomena, such as emulsion vortices and hindered swirling large bubbles, that occur deep in the reactor remote from its walls, etc. The "rst adaptation of PIV to three-phase #uidized beds was reported by Fan and co-workers (Chen and Fan, 1992), which was followed, after further improvements of the technique (Chen et al., 1994; Reese et al., 1995; Reese and Fan, 1997b), by re"ned qualitative and quantitative descriptions of the freeboard region in terms of three-phase velocity "elds, bubble-size, gas and liquid holdup distributions, and slip velocities. A radioactive particle tracing technique, more convenient for probing dense emulsions, was employed by Chaouki and co-workers, Dudukovic and co-workers and Larachi (Larachi et al., 1995a,b, 1996; Limtrakul, 1996) to measure the 3-D Lagrangian movement of the solids in dense three-phase #uidized beds without draft tubes. CARPT measurements were utilized to quantify the mechanisms of the solids motion, to evaluate and model the solids mixing and circulation times and to map the time-averaged Eulerian full #ow velocity vectors and turbulence "elds. A draft tube clearly intensi"es the magnitude of the axial average solids velocities due to the extinction of the turbulent radial transport at the radius of the draft tube, but also because of the additional outward spill-over of the solids towards the annulus right above the draft tube. In the standard #uidized bed, the solids mean #ow evolves clockwise in a 3D toroidal recirculation cell; whereas the draft tube brings about a twostage vertical clockwise rotational #ow pattern of the solids, fast in the bottom stage and slow in the upper stage. Identi"cation of the hydrodynamic regimes has been attempted based on visual observation, wall pressure #uctuations, and bubble sizes (Wild and Poncin, 1996; Fan, 1989) and time-series conductivity probe signals (Briens et al., 1996). However, predicting the #ow regime in three-phase #uidization is hampered by the complex dependence of #ow regimes upon column diameter, distributor type, settled bed height, particle density, geometry, and wettability, coalescence inhibition of the liquid, etc. (Bigot, 1990; Nacef, 1991; Nore, 1992; Nore et al., 1992). Bejar et al. (1992) derived a #ow chart suitable for fermentation media in three-phase #uidization to distinguish the dispersed bubble #ow from the coalesced bubble #ow regimes with Ca-alginate or carrageenan immobilizing particles. Zhang et al. (1997), by using a two-element conductivity probe, provided a re"ned discrimination of #ow patterns in three-phase #uidized beds and arrived at seven #ow regimes: dispersed-, discrete-, coalesced-bubble #ow, slug #ow, bridging #ow, churn #ow, and annular #ow. They also proposed a set of correlations to predict changeover between these di!erent regimes. The following rules of thumb regarding #ow regimes in three-phase #uidization emerge (Nacef, 1991; Nore, 1992; Cassanello et al., 1995; Wild and Poncin, 1996). In bed inventories made up of small/dense particles ()1 mm) and light particles (density)1700 kg/m3), only the coalesced bubble #ow regime is most likely to occur. In those cases, the #ow regime can be coerced to the dispersed bubble #ow regime by adding large and light bubble breakers (Kim and Kim, 1990). The dispersed bubble #ow prevails at low gas velocity and high liquid velocity, in bed inventories of large particles (*3}4 mm), whereas coalesced bubble #ow dominates for low liquid and/or high gas velocities. Slug #ow occurs in small-diameter columns ((0.1 m) at high gas velocity ('0.1 m/s). Further complications in #ow regimes arise when non-wettable particles are #uidized, Tsutsumi et al. (1991) thus identi"ed aggregative #uidization at moderate velocities, and dispersed #uidization at higher velocities. 3.2.2. Minimum yuidization velocity, porosity, phase holdups Minimum #uidization velocity and phase holdups can only be estimated based on empirical correlations. Nacef M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 1989 Table 6 Impact of operating conditions on the phase holdups in three-phase #uidization Increase in B E!ect on P Super"cial liquid velocity Super"cial gas velocity Particle diameter Particle density Liquid density Liquid viscosity Pressure Coalescence inhibition Distribution quality Bed expansion Gas holdup Liquid holdup Increase Increase Decrease Decrease Increase Increase N/A Increase Increase No change Increase N/C Decrease Decrease Decrease Increase Increase Increase Increase Decrease N/C N/C N/C Increase Decrease Decrease No change N/C: no clear cut; N/A: not available (1991) and Zhang et al. (1995) provide correlations for minimum #uidization velocity, while Han et al. (1990) and Nore (1992) present correlations for bed expansion and liquid holdup. The impact of the change in various operating or process variables on phase holdups in three phase #uidization is illustrated in Table 6 (Wild and Poncin, 1996; Luo et al., 1997). 3.2.3. Bed contraction/expansion Bed contraction, a phenomenon peculiar to threephase #uidized beds, occurs when su$cient liquid is sucked up in the bubble wakes to starve signi"cantly the liquid #ow in the emulsion phase; as a result the bed contracts. The following rules were drawn based on experimental observations of bed contraction/expansion (Han et al., 1990; Nacef, 1991; Wild and Poncin, 1996; Jiang et al., 1997). With moderately viscous liquids and for particles with size below 2.5 mm, bubble coalescence is promoted and bed contraction is likely to occur; larger particles ('2.5 mm) tend to promote bubble break-up and bed expansion increases with increasing gas velocity. For highly viscous liquids, bed contraction and bubble coalescence occur regardless of particle size. Badly designed distributors promote bed contraction even for large size particles. High pressure/temperature reduces the extent of bed contraction as a result of reduction in bubble size. 3.2.4. Heat and mass transfer Heat and mass transfer in three-phase #uidization seem to depend on many parameters in a very complex manner (Tang and Fan, 1990; Kim et al., 1990; Kang et al., 1991; Del Pozo et al., 1992; Nore et al., 1992; Kim and Kang, 1997; Luo et al. 1997). Wall to bed, as well as immersed heater-to-bed, heat transfer coe$cients are reported. In general, the heat transfer coe$cient in threephase #uidized beds increases with gas/liquid super"cial velocities, size and density of particles, column diameter, thermal conductivity and heat capacity of the liquid; whereas it decreases with liquid dynamic viscosity. The gas}liquid volumetric liquid-side mass transfer coe$cient increases with #uid throughputs, size and density of particles; it decreases with increasing surface tension and dynamic viscosity of the liquid, and solids holdup for light particles. Bubble breakers improve mass transfer; mismatch to verticality of the column may improve or deteriorate the gas}liquid mass transfer. There are no data available on heat transfer at high temperature, on the impact of coalescence inhibitors, quality of gas}liquid initial distribution, liquid surface tension and density. Recent correlations and models developed for the prediction of the various heat and mass transfer coe$cients for three-phase #uidized beds are discussed thoroughly in Kim and Kang (1997). 3.2.5. High-pressure operation Despite the fact that high-pressure and high-temperature operations are most often encountered in industrial three-phase #uidization practice, the paucity of studies relevant to these conditions is notorious. Only some papers on high-pressure/temperature three-phase #uidized beds (up to 15.6 MPa and 943C) have been published by Fan and co-workers (Jiang et al., 1992, 1997; Luo et al., 1997). The consequences of increased pressure and temperature on hydrodynamic and heat transfer parameters of three-phase #uidized beds can be summarized as follows: The transition between the dispersed bubble #ow and the coalesced bubble #ow regimes is moved with increased pressure towards higher gas super"cial velocities. As pressure increases up to 6 MPa, the transition velocity and gas holdup is increased; beyond this value, the transition velocity nearly levels o!. Gas velocity at the inception of the coalesced bubble #ow regime increases with liquid super"cial velocity and particle diameter. 3.3. Concluding remarks It is fair to say that the knowledge base for reactors with moving catalyst is even less complete than for "xed-bed 1990 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 reactors. The scale-up procedures are prone to more uncertainty and it is not possible in general to relate via simple scale-up rules the performance of laboratory size units to large-scale reactors. Careful investigation of kinetics in another reactor type coupled with cold #ow and CFD models of the large units such as risers, ebullated beds, bubble columns is usually the preferred route in process development. In these reactor types both improved scale-up procedures and utilization of CFD have an important role to play. Clearly, much more work based on fundamental approaches remains to be done. 4. Final remarks Our intent was to provide a more systematic review that includes two-phase systems such as packed beds, #uidized beds and risers as well as other frequently used reactor types such as stirred tanks for gas}liquid and liquid}solid operation. In addition, it is important to access the state-of-the-art of unconventional reactors, such as monoliths for two-phase processing, and reactors that combine separation and reaction, such as chromatographic reactors, catalytic distillation columns or rotating packed beds. While all of this has been prepared, due to space limitations it could not be included in this review. We will attempt to publish the whole comprehensive chapter elsewhere. This review as presented, attempted to summarize as to what is known about the #ow patterns, #uid dynamic parameters and transport phenomena in some commonly used three-phase reactors. This information is needed in reactor modeling or scale-up for any particular process. Four important areas were not discussed in detail. First, although we have indicated that the improved understanding of #uid dynamics in multiphase reactors can only be reached by non-invasive experimental means, and that such data are essential for veri"cation of computational models, we have not reviewed the available experimental techniques. This was omitted since two of the authors (M.P.D. and F. L.) have recently co-authored with Professor Chaouki an extensive review dedicated to this very topic (Chaouki et al., 1997b). In addition, a book has been edited on the subject that summarizes all the available techniques (Chaouki et al., 1997a). Second, while the importance of computational #uid dynamic models for multiphase reactors is stressed throughout this review, no attempt was made to systematically summarize this vast "eld in view of the recent comprehensive review by Kuipers and van Swaaij (1997). Third, we have not had the space to discuss process chemistries and kinetic modeling. In order to limit the size of this review, we had to focus on description of #ow patterns and transport. One of the authors (P.L.M.) has recently discussed the process chemistries and kinetic modeling e!ects of some processes of the pharmaceutical (Mills and Chaudhari, 1997) and specialty industries (Mills et al., 1992). This brings us to the "nal, and arguably most important, area that was not covered in our review. That is the art and science of experimental multiphase reactors. From the process development point of view it is most important to have microreactors that are well instrumented in which mixing and contacting patterns are well characterized. Rapid evaluation of various catalysts is then followed by direct scale-up to large units with the help of CFD and cold #ow models. In our opinion, it is this area of currently available and current developments in the laboratory multiphase reactors that merits the attention of a review dedicated to that topic alone. Finally, the area of reactor safety, runaway prevention and control is essential to proper and safe usage of multiphase reactors and needs to be reviewed in the future. This is often more e!ectively done in the context of a speci"c process type rather than in a generic sense. We hope that the present review will provide the reader with the overall view of where do we currently stand with respect to our knowledge of various multiphase reactor types and as to what needs to be done to improve the state-of-the-art. Acknowledgements We would like to thank the CREL graduate students who have contributed to this review, in particular S. Roy, Y. Jiang and M. Khadilkar. References Al-Dahhan, M.H., Wu, Y.X., & Dudukovic, M.P. (1995). Reproducible technique for packing laboratory-scale trickle-bed reactors with a mixture of catalyst and "nes. I&EC Res., 34, 741. Al-Dahhan, M.H., & Dudukovic, M.P. (1995) Catalyst wetting e$ciency in trickle-bed reactors at high pressure. Chem. Engng Sci., 50, 2377. Al-Dahhan, M.H., & Dudukovic, M.P. (1996). Catalyst bed dilution for improving catalyst wetting in laboratory trickle-bed reactors. A.I.Ch.E. J., 42, 2594. Al-Dahhan, M.H., Larachi, F., Dudukovic, M. P., & Laurent, A. (1997). High-pressure trickle-bed reactors: A review. I&EC Res., 36, 3292. Al-Dahhan, M.H., Khadilkar, M.R. Wu, Y., & Dudukovic, M.P. (1998). Prediction of pressure drop and liquid holdup in high-pressure trickle-bed reactors. I&EC Res., 37, 793. Alejandre, A., Medina, F., Fortuny, A., Salagre, P., & Sueiras, J.E. (1998). Characterisation of copper catalysts and activity for the oxidation of phenol aqueous solutions, Appl. Catal. B: Environ., 16, 53. Alonso, C., Suidan, M.T., Sorial, G.A., Smith, F.L., Biswas, P., Smith, P.J., & Brenner, R.C. (1997). Gas treatment in trickle-bed bio"lters: Biomass, how much is enough. Biotechnol. BioEngng, 54, 583. AndreH , A. (1997). Construction de banques de donneH es sur les eH changes de matieH re dans un reH acteur triphasique aH co-courant vers le bas. DeH veloppement d'une correH lation neuronale. M.Sc., Institut National Polytechnique de Lorraine, Nancy, France. M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Beaudry, E.G., Dudukovic, M.P., & Mills P.L. (1987). Trickle-bed reactors: Liquid di!usional e!ects in a gas limited reaction. A.I.Ch.E. J., 33, 1435. Becker, S., Sokolichin, A., & Eigenberger, G. (1994). Gas-liquid #ow in bubble columns and loop reactors, Part II comparison of detailed experiments and #ow simulations. Chem. Engng Sci., 49, 5747. Bejar, P., Casas, C., Godia, G., & Sola, C. (1992). The in#uence of physical properties on the operation of a three-phase #uidized-bed fermentor with yeast cells immobilized in Ca-alginate, Appl. Biochem. Biotechnol., 34/35, 467. Bensetiti, Z., Larachi, F., Grandjean, B.P.A., & Wild, G. (1997). Liquid saturation in cocurrent up#ow "xed-bed reactors: A state-of-the-art correlation. Chem. Engng Sci., 52, 4239. Bergault, I., Rajashekharam, M.V., Chaudhari, R.V., Schweich, D., & Delmas, H. (1997). Modeling and comparison of acetophenone hydrogenation in trickle-bed and slurry airlift reactors. Chem. Engng Sci., 52, 4033. Berruti, F., Hudgins, R.R., Rhodes, E., & Sicardi, S. (1984). Oxidation of sulfur-dioxide in a trickle bed reactor: A study of reactor modeling. Can. J. Chem. Engng, 62, 644. Bigot, V. (1990) ED tude des paramètres caracteH ristiques d'un reH acteur à lit -uidiseH triphasique: reH tention des phases, cartes d'eH coulement du gaz et transfert de matière gaz-liquide. Ph.D. thesis, Institut National des Sciences AppliqueH es, Toulouse, France. Blok, J.R., & Drinkenburg, A.A.H. (1982). Hydrodynamics and mass transfer in pulsing trickle-bed columns. Chem. Engng J., 25, 89. Briens, C.L., Hudson, C., & Briens, L.A. (1996). Rapid characterization of #ow regimes in multiphase reactors through box-counting dimensions with an embedding dimension of two. Chem. Engng J., 64, 169. Burghardt, A., & Bartelmus, G. (1996). Hydrodynamics and mass transfer in three-phase cocurrent reactors. Chem. Engng Sci., 51, 2733. Burghardt, A., Bartelmus, G., & Stasiak, M. (1996). Hydrodynamika wspolpradowych reaktorow trojfazowych ze stalym zlozem czesc II. Wery"kacja parametrow modelu hydrodynamicznego. Inz. Chem. Proc., 17, 175. Cassanello, M., Larachi, F., Marie, M.-N., Guy, C., & Chaouki, J. (1995). Experimental characterization of the solid phase chaotic dynamics in three-phase #uidization. I&EC Res., 34, 2971. Castellari, A., & Haure, P.M. (1995). Experimental study of the periodic operation of a trickle-bed reactor. A.I.Ch.E. J., 41, 1593. Castellari, A.T., Cechini, J.O., Gabarain, L.J., & Haure, P.M. (1997). Gas-phase reaction in a trickle-bed reactor operated at low liquid #ow rates. A.I.Ch.E. J., 47, 1813. Chanchlani, K.G., Hudgins, R.R., & Silveston, P.L. (1994). Methanol synthesis under periodic operation: An experimental investigation. Can. J. Chem. Engng, 72, 657. Chaouki, J., Larachi, F., & Dudukovic, M.P. (1997a). Non-invasive monitoring of multiphase -ows. Amsterdam: Elsevier. Chaouki, J., Larachi, F., & Dudukovic, M.P. (1997b). Noninvasive tomographic and velocimetric monitoring of multiphase #ows. I&EC Res., 36, 4476. Chen, R.C., & Fan, L.-S. (1992). Particle image velocimetry for characterizing the #ow structure in three-dimensional gas-liquid-solid #uidized beds. Chem. Engng Sci., 47, 3615. Chen, R.C., Reese, J., & Fan, L.-S. (1994). Flow structure in a threedimensional bubble column and three-phase #uidized bed. A.I.Ch.E. J., 40, 1093. Chen, Y.W., & Hsu, W.C. (1997). Hydrodemetalation of residue oil over CoMo/alumina-aluminum phosphate catalysts in a trickle-bed reactor. I&EC Res., 36, 2521. Chen, Y.W., & Tsai, M.C. (1997). Hydrodrodesulfurization of atmospheric gas oil over NiMo/ aluminum borate catalysts in a tricklebed reactor. I&EC Res., 36, 2521. Cheng, S., & Chuang, K.T. (1992). Simultaneous methanol removal and destruction from wastewater in a trickle-bed reactor. Can. J. Chem. Engng, 70, 727. Chon, S.K. (1987). U. S. Patent 4 696 772. 1991 Chou, S.H., Chen, S.C., Tan, C.S., & Wang, W.H. (1997). Hydrogenation of dicyclopentadiene in a trickle-bed reactor. J. Chinese Inst. Chem. Engrs, 28, 175. Chu, C.F., & Ng, K.M. (1986). Liquid dispersion in trickle bed reactors with gas-liquid cocurrent down#ow, Paper 10p, AIChE Annual Meeting, Miami Beach, FL. Collins, G.M., Hess, R.K., Hook, B.D., & Ackgerman, A. (1984). Paper 31p, AIChE Annual Meeting, San Fransisco, CA. Dassori, C.G., (1998). Three phase reactor modeling with signi"cant backmixing in the liquid phase using modi"ed cell model (MCM). Chem. Engng. Sci., 22 (Suppl), S679. Degaleesan, S. (1997). Fluid dynamic measurements and modeling of liquid mixing in bubble columns. D.Sc. thesis, Washington University, St. Louis. Degaleesan, S., & Dudukovic, M.P. (1998). Liquid Backmixing in Bubble Columns and the Axial Dispersion Coe$cient. A.I.Ch.E. J. (in print). Degaleesan, S., Dudukovic, M.P., Toseland B.A., & Bhatt, B.L. (1996a). ¹racer studies of the ¸a Porte AFD; reactor during methanol synthesis. 4th Quarterly Report, DOE Contract DE-FC 2295 PC 95051. Degaleesan, S., Dudukovic, M.P., Toseland, B.A., & Bhatt, B.L. (1997). A two-compartment convection-di!usion model for slurry bubble column reactors. I&EC Res., 36, 4670. Degaleesan, S., Roy, S., Kumar, S.B., & Dudukovic, M.P. (1996b). Liquid mixing based on convection and turbulent dispersion in bubble columns. Chem. Engng Sci., 51, 1967. Del Pozo, M., Briens, C.L., & Wild, G. (1992). E!ect of column inclination on the performance of three-phase #uidized beds. A.I.Ch.E. J., 38, 1206. Delnoij, E., Kuipers, J.A.M., & van Swaaij, W.P.M. (1997a). Computational #uid dynamics applied to gas-liquid contactors. Chem. Engng Sci., 52, 1. Delnoij, E., Kuipers, J.A.M., & van Swaaij, W.P.M. (1997b). Dynamic simulation of gas-liquid two-phase #ow: E!ect of column aspect ratio on the #ow structure. Chem. Engng Sci., 52, 3759. Delnoij, E., Lammers, F.A., Kuipers, J.A.M., & van Swaaij, W.P.M. (1997c). Dynamic simulation of dispersed gas}liquid two-phase #ow using a discrete bubble model. Chem. Engng Sci. 52, 1429. Devanathan, N. (1991). Investigation of liquid hydrodynamics in bubble columns via computer automated radioactive particle tracking. D.Sc. thesis, Washington University, St. Louis. Devanathan, N., Dudukovic, M.P., Lapin, A., & LuK bbert, A. (1995). Chaotic #ow in bubble column reactors. Chem. Engng Sci., 50, 2661. Devanathan, N., Moslemian, D., & Dudukovic, M.P. (1990). Flow mapping in bubble columns using CARPT, Chem. Engng Sci., 45, 2285. Devetta, L., Canu, P., Bertucco, A., & Steiner, K. (1997). Modeling of a trickle-bed reactor for a catalytic hydrogenation in supercritical CO . Chem. Engng Sci., 52, 4163. 2 Dicks, R.M., & Ottengraf, S.P.P. (1991). Veri"cation of a simpli"ed model for removal of dichloromethane from waste gases using a biological trickling "lter. Bioprocess Engng, 6, 131. Drew, D.A. (1983). Mathematical modeling of two-phase #ow. Ann. Rev. Fluid Mech., 15, 261. Dudukovic, M.P. (1998). Trends in catalytic reaction engineering. Proc. 2nd Int. Symp. Catal. and Multiphase Reactors. (pp. 13}15). Toulouse, France. Dudukovic, M.P., Degaleesan, S., Gupta, P., & Kumar, S.B. (1997). Fluid dynamics in churn-turbulent bubble columns: Measurements and modeling. Proc. ASME Fluids Engineering Division Summer Meeting Vol. 244 (3517), Vancouver, Canada. El-Hisnawi, A.A. (1981). ¹racer and reaction studies in trickle-bed reactors. D.Sc. dissertation, Washington University, St-Louis, MO, USA. El-Hisnawi, A.A., Dudukovic, M.P., & Mills, P.L. (1981). Trickle-bed reactors: Dynamic tracer tests, reaction studies and modeling of reactor performance. ACS Symp. Ser., 196, 421. 1992 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Ellenberger, J., & Krishna, R. (1994). A uni"ed approach to the scale-up of gas-solid #uidized bed and gas-liquid bubble column reactors. Chem. Engng Sci., 24B, 5391. Ellman, M.J. (1988). Caracte& ristiques des re& acteurs triphasiques a% lit ,xe fonctionnant a% co-courant vers le bas. Ph.D. dissertation, Institut National Polytechnique de Lorraine, Nancy, France. Ellman, M.J., Midoux, N., Laurent, A., & Charpentier, J.C. (1988). A New, improved pressure drop correlation for trickle-bed reactors. Chem. Engng Sci., 43, 2201. Euzen, J.P., Trambouze, P., & Wauquier, J.P. (1993). Me& thodologie pour l'extrapolation des proce& de& s chimiques. Publications de l'Institut Franiais du PeH trole, ED ditions Technip. Fan, L., Yoshii, K., Yan, S., Zhou, J., & Fujimoto, K. (1997). Supercritical-phase process for selective synthesis of wax from syngas: Catalyst and process development. Catal. ¹oday 36, 295. Fan, L.-S. (1989). Gas}liquid}solid -uidization engineering. Butterworths Series in Chemical Engineering, Boston. Fortuny, A., Ferrer, C., Bengoa, C., Font, J., & Fabregat, A. (1995). Catalytic removal of phenol from aqueous phase using oxygen or air as oxidant. Catal. ¹oday, 24, 79. Frank, M.J.W. (1996). Mass and heat transfer phenomena in G-¸(-S) reactors relevant for reactive distillation. Ph.D. dissertation, Twente University, Enschede, The Netherlands. Funk, G.A., Harold, M.P., & Ng, K.M. (1990). Novel model for reaction in trickle beds with #ow maldistribution, 29, 738. Gabarain, L., Castellari, A.T., Cechini, J., Tobolski, A., & Haure, P. (1997a). Analysis of rate enhancement in a periodically operated trickle-bed reactor. A.I.Ch.E. J., 43, 166. Gabarain, L., Cechini, J., & Haure, P. (1997b). E!ect of variables on the periodic operation of a trickle bed reactor. Studies in Surface Science and Catalysis, 109 (Dynamics of Surfaces and Reaction Kinetics in Heterogeneous Catalysis), 459}467. Gianetto, A., & Specchia, V. (1992). Trickle-bed reactors: State of art and perspectives. Chem. Engng Sci., 47, 3197. Godia, F., & Sola, C. (1995). Fluidized-bed bioreactors. Biotechnol. Prog., 11, 479. Goto, S., Chatani, T., & Matouq, M.H. (1993). Hydration of 2-methyl2-butene in gas-liquid cocurrent up#ow and down#ow reactors. Can. J. Chem. Engng, 71, 821. Goto, S., & Mabuchi, K. (1984). Oxidation of ethanol in gas-liquid cocurrent up#ow and down#ow reactors. Can. J. Chem. Engng, 62, 865. GroK en, J.S., Mudde, R.F., & van den Akker, H.E.A. (1995). Timedependent behavior of #ow in a bubble column. ¹rans. Inst. Chem. Engrs., A9-A16. GroK en, J.S., Oldeman, G.C., Mudde, R.F., & van den Akker, H.E.A. (1996). Coherent structures and axial dispersion in bubble column reactors. Chem. Engng. Sci., 51, 2511}2520. Gupta, R. (1985). Pulsed #ow vapor-liquid reactors. U.S. Patent d 4526757. Han, J.H., Wild, G., & Kim, S.D. (1990). Phase holdup characteristics in three-phase #uidized beds. Chem. Engng J., 43, 67. Harold, M.P., & Watson, P.C. (1993). Bimolecular exothermic reaction with vaporization in the half-wetted slab catalyst. Chem. Engng Sci., 48, 981. Haure, P.M., Bogdashev, S.M., Bunimovich, M., Stegasov, A.N., Hudgins, R.R., & Silveston, P.L. (1990a). Thermal waves in periodic operation of trickle-bed reactor. Chem. Engng Sci., 45, 2255. Haure, P.M., Hudgins, R.R., & Silveston, P.L. (1990b). Steady-state models for SO oxidation in trickle-bed reactors. Chem Engng J., 43, 2 121. Hekmat, D., & Vortmeyer, D. (1994). Modeling of biodegradation processes in trickle bed bioreactors. Chem. Engng Sci. 49, 25. Herrmann, U., & Emig, G. (1998). Liquid-phase hydrogenation of maleic anhydride to 1,4-butanediol in a packed bubble column reactor. I&EC Res., 37, 759. Horowitz, G.I., Cukierman, A.L., & Cassanello, M.C. (1997). Flow regime transition in trickle beds packed with particles of di!erent wetting characteristics: Check-up on new tools. Chem. Engng Sci., 52, 3747. Huang, T.C., & Kang, B.C. (1995). Kinetic study of naphthalene hydrogenation over Pt/Al O catalyst. I&EC Res., 34, 1140. 2 3 Iliuta, I., & Iliuta, M.C. (1997). Comparison of two-phase up#ow and down#ow "xed bed reactors performance: Catalytic SO oxidation. 2 Chem. Engng ¹echnol., 20, 455. Iliuta, I., & Thyrion, F.C. (1997a). Flow regimes, liquid holdups and two-phase pressure drop for two-phase cocurrent down#ow and up#ow through packed beds: Air/Newtonian and non-Newtonian liquid systems. Chem. Engng Sci., 52, 4045. Iliuta, I., & Thyrion, F.C. (1997b). Gas-liquid mass transfer in "xed beds with two-phase cocurrent down#ow: Gas/Newtonian and non-Newtonian liquid systems. Chem. Engng ¹echnol., 20, 538. Iliuta, I., Iliuta, M.C., & Thyrion, F.C. (1997c). Gas}liquid mass transfer in trickle-bed reactors: gas-side mass transfer. Chem. Engng ¹echnol., 20, 589. Iliuta, I. (1997). Performance of "xed bed reactors with two-phase up#ow and down#ow. J. Chem. ¹echnol. Biotechnol., 68, 47. Jakobsen, H.A. (1993). On the modeling and simulation of bubble column reactors using two--uid model. Dr. Ing. thesis, Universitet I Trondheim, Norway. Jakobsen, H.A., Sannaes, B.H., Grevskott, S., & Svendsen, H.F. (1997). Modeling of vertical bubble regime #ows. I&EC Res., 36, 4052. Jiang, P., Arters, D., & Fan, L.S. (1992). Pressure e!ects on the hydrodynamic behavior of gas-liquid-solid #uidized beds. I&EC Res., 31, 2322. Jiang, Y., Khadilkar, M.R., Al-Dahhan, M., Dudukovic, M.P., Chou, S.K., Ahmed, G., & Kahney, R. (1998). Investigation of a complex reaction network: II. Kinetics, mechanism and parameter estimation. A.I.Ch.E. J., 44, 921. Jiang, P., Luo, X., Lin, T.J., & Fan, L.S. (1997). High temperature and high pressure three-phase #uidization-Bed expansion phenomena. Powder ¹echnol., 90, 103. Jiriczny, V., & Stanek, V. (1996). An experimental setup to measure #ow transients in a countercurrent packed bed column. Chem. Biochem. Engng J., 10, 55. Kang, Y., Fan, L.T., Min, B.T., & Kim, S.D. (1991). Promotion of oxygen transfer in three-phase #uidized bioreactors by #oating bubble breakers. Biotechnol. Bioengng, 37, 580. Khadilkar, M. R., Wu, Y.X., Al-Dahhan, M.H., Dudukovic, M.P., & Colakyan, M. (1996). Comparison of trickle-bed and up#ow reactor performance at high pressure: Model predictions and experimental observations. Chem. Engng Sci., 51, 2139. Khadilkar, M.R., Al-Dahhan, M., & Dudukovic, M.P. (1998a). Parametric study of unsteady state #ow modulation in trickle bed reactors. Chem. Engng Sci., 53, ISCRE'15 Proceedings. Khadilkar, M.R., Dudukovic, M.P., & Mills, P.L. (1998b). Trickle-bed reactor models for systems with a volatile liquid phase. Chem. Engng Sci., 53, ISCRE'15 Proceedings. Khadilkar, M.R., Jiang, Y., Al-Dahhan, M., Dudukovic, M.P., Chou, S.K., Ahmed, G., & Kahney, R. (1998c). Investigation of a complex reaction network: I. Experiments in a high-pressure trickle-bed reactor. A.I.Ch.E. J., 44, 912. Kheshgi, H.S., Reyes, S.C., Hu, R., & Ho, T.C. (1992). Phase transition and steady state multiplicity in a trickle-bed reactor. Chem. Engng Sci., 47, 1771. Khinast, J., & Luss, D. (1997). Mapping regions with di!erent bifurcation diagrams of a reverse-#ow reactor. A.I.Ch.E. J., 43, 2034. Kiared, K., & Zoulalian, A. (1992). Study and modeling of catalytic sulfur dioxide oxidation in Verli"x three-phase reactor. Chem. Engng Sci., 47, 3705. Kim, J.O., & Kim, S.D. (1990). Bubble breakage phenomena, phase holdups and mass transfer in three-phase #uidized beds with #oating bubble breakers. Chem. Engng Proc., 28, 101. M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Kim, J.O., Park, D.H., & Kim, S.D. (1990). Heat transfer and wake characteristics in three-phase #uidized beds with #oating bubble breakers. Chem. Engng Proc., 28, 113. Kim, S.D., & Kang Y. (1997). Heat and mass transfer in three-phase #uidized-bed reactors } an overview. Chem. Engng Sci., 52, 3639. Koh, C.S., & Butt, J.B. (1995). Experimental and modeling study of kinetics and selectivity in the oxidation of a poly(a-ole"n) lubricant. I&EC Res., 34, 524. Korsten, H., & Ho!mann, U. (1996). Three-phase reactor model for hydrotreating in pilot trickle-bed reactors. A.ICh.E. J., 42, 1350. Krishna, R., Ellenberger, J., & Hennephof, D.E. (1993). Analogous description of the hydrodynamics of gas-solid #uidized beds and bubble columns. Chem. Engng J., 53, 89. Krishna, R., & Sie, S.T. (1994). Strategies for multiphase reactor selection. Chem. Engng Sci., 49, 4029. Krishna, R., & Ellenberger, J. (1996). Gas holdup in bubble column reactors operating in the churn-turbulent #ow regime. A.I.Ch.E. J., 42, 2627. Kuipers, J.A.M., & Van Swaaij, (1997). Application of computational #uid dynamics to chemical reaction engineering. Reviews Chem. Engng, 3, 1. Kumar, B.S. (1994). Computer tomographic measurements of void fraction and modeling of the -ow in bubble columns. Ph.D. thesis, Florida Atlantic University, Boca Raton, FL. Kumar, B.S., Devanathan, N., Moslemian, D., & Dudukovic, M.P. (1994). E!ect of scale on liquid recirculation in bubble columns. Chem. Engng Sci., 49, 5637. Kumar, B.S., Moslemian, D., & Dudukovic, M.P. (1995a). A gamma ray tomographic scanner for imaging void fraction distribution in bubble columns. Flow Meas. Instr., 6, 61. Kumar, S.B., Dudukovic, M.P., & Toseland, B.A. (1997). Measurement techniques for local and global #uid dynamic quantities in two and three-phase systems. In: J. Chaouki, F. Larachi, & M.P. Dudukovic' (Eds.), Non-invasive monitoring of multiphase -ows (pp. 1}46). Amsterdam: Elsevier. Kumar, S.B., Vanderheyden, W.B., Devanathan, N., & Padial, N.T. (1995b). Numerical simulation and experimental veri"cation of the gas-liquid #ow in bubble columns. A.I.Ch.E. Symp. Ser., 305, 11. Lababidi, H.M.S., Shaban, H.I., Al-Radwan, S., & Alper, E. (1998). Simulation of an atmospheric residue desulfurization unit by quasi-steady state modeling. Chem. Engng ¹echnol., 21, 193. Landau, M.V., Herskowitz, M., Givoni, D., Laichter, S., & Yitzhaki, D. (1998). Medium severity hydrotreating and hydrocracking of Israeli shale oil } II. Testing of novel catalyst systems in a trickle-bed reactor. Fuel, 77, 3. Lange, R., Hanika, J., Stradiotto, D., Hudgins, R.R., & Silveston, P.L. (1994). Investigations of periodically operated trickle-bed reactors. Chem. Engng Sci., 49, 5615. Lapin, A., & LuK bbert, A. (1994). Numerical simulation of the dynamics of two-phase gas-liquid #ows in bubble columns. Chem. Engng Sci., 49, 3661. Larachi, F., Bensetiti, Z., Grandjean, B.P.A., & Wild, G. (1998b). Twophase frictional pressure drop in #ooded-bed reactors-A state-ofthe-art correlation. Chem. Engng ¹echnol., to be published. Larachi, F., Cassanello, M., Chaouki, J., & Guy, C. (1996). Flow structure of the solids in a three-dimensional gas-liquid-solid #uidized bed. A.I.Ch.E. J., 42, 2439. Larachi, F., Chaouki, J., Kennedy, G., & Dudukovic, M.P. (1997b). Radioactive particle tracking in multiphase reactors: Principles and applications, In: J. Chaouki, F. Larachi, & M.P. Dudukovic, (Eds.), Non-invasive monitoring of multiphase -ows (pp. 335}406). Amsterdam: Elsevier. Larachi, F., Cassanello, M., & Laurent, A. (1998a). Gas-liquid interfacial mass transfer in trickle-bed reactors at elevated pressures. I&EC Res., 37, 718. 1993 Larachi, F., Cassanello, M, Laurent, A., Midoux, N., & Wild, G. (1997a). Gas-liquid interfacial areas in three-phase "xed-bed reactors. Chem. Engng Process., 36, 497. Larachi, F., Cassanello, M., Marie, M.-N., Chaouki, J., & Guy, C. (1995b). Solids circulation patterns in three-phase #uidized beds containing binary mixtures of particles as inferred from RPT. ¹rans IChemE part A, 73, 263. Larachi, F., Kennedy, G., & Chaouki, J. (1995a). 3-D mapping of solids #ow "elds in multiphase reactors with RPT. AIChE J., 41, 439. Larachi, F., Laurent, A., Wild, G., & Midoux, N. (1993). E!et de la pression sur la transition ruisselant-pulseH dans les reH acteurs catalytiques à lit "xe arroseH . Can. J. Chem. Engng, 71, 319. Lara-MaH rquez, A., Larachi, F., Wild, G., & Laurent, A. (1992). Mass tranfer characteristics of "xed beds with cocurrent up#ow and down#ow. A special reference to the e!ect of pressure. Chem. Engng Sci., 47, 3485. Lati", M.A., Rode, S., Midoux, N., & Storck, A. (1992a). Hydrodynamic study of a trickle-bed reactor by means of microelectrodes: Analysis of the probability densities. Chem. Engng Sci., 47, 2331. Lati", M.A., Rode, S., Midoux, N., & Storck, A. (1992b). The use of microelectrodes for the determination of #ow regimes in a trickle-bed reactor. Chem. Engng Sci., 47, 1955. Laurenzis, A., Heits, H., WuK bker, S.-M., Heinze, U., Friedrich, C., & Werner, U. (1998). Continuous biological waste gas treatment in stirred trickle-bed reactor with discontinuous removal of biomass. Biotech. Bioengng, 57, 497. LaVopa, V., & Satter"eld, C.N. (1988). Some e!ects of vapor-liquid equilibria on performance of a trickle-bed reactor. Chem. Engng Sci., 43, 2175. Lee, J.K., Hudgins, R.R., & Silveston, P.L. (1995). Cycled trickle-bed reactor for SO Oxidation. Chem. Engng Sci., 50, 2523. 2 Lemco!, N.O., Cukierman, A.L., & Martinez, O.M. (1988). E!ectiveness factor of partially wetted catalyst particles: Evaluation and application to the modeling of trickle bed reactors. Catal. Rev. Sci. Engng, 30, 393. Lerou, J.J., & Ng, K.M. (1996). Chemical reaction engineering: A multiscale approach to a multiobjective task. Chem. Engng Sci., 51, 1595. Limtrakul, S. (1996). Hydrodynamics of liquid -uidized beds & gas-liquid -uidized beds., D.Sc. thesis, Washington University, St. Louis. Lin, T.-J., Reese, J., Hong, T., & Fan, L.-S. (1996). Quantitative analysis and computation of two-dimensional bubble columns. A.I.Ch.E. J., 42, 301. Llano, J.J., Rosal, R., Sastre, H., & Diez, F. (1997). Determination of wetting e$ciency in trickle-bed reactors by a reaction method. I&EC Res., 36, 2616. Luo, X., Jiang, P., & Fan, L.S. (1997). High-pressure three-phase #uidization: Hydrodynamics and heat transfer. A.I.Ch.E. J., 43, 2432. Luss, D. (997). Temperature fronts and patterns in catalytic systems. I&EC Res., 36, 2931. Martinez, O.M., Cassanello, M.C., & Cukierman, A.L. (1994). Threephase "xed bed catalytic reactors: Application to hydrotreatment processes. ¹rends Chem. Engng, 2, 393. Mazzarino, I., Occhetti, M., Baldi, G., & Sicardi, S. (1989). Performance of a packed bed reactor with a two-phase upward #ow. Chem. Engng Commun., 75, 225. Mazzarino, I., Santos, A., & Baldi, G. (1994). Catalytic oxidation of n-propanol in a multiphase up#ow reactor: Surface tension e!ects. Chem. Engng Sci., 49, 5699. McManus, R.L., Funk, G.A., Harold, M.P., & Ng, K.M. (1993). Experimental study of reaction in trickle-bed reactors with liquid maldistribution. I and EC Res., 32, 570. Menzel, T., in der Weide, T., Staudacher, O., Wein, O., & Onken, U. (1990). Reynolds shear stress for modeling of bubble column reactors. I&EC Res., 29, 988. Meyers, R. A. (1996). Handbook of petroleum re,ning processes, (2nd ed.). New York: McGraw-Hill. 1994 M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Mills, P.L., Beaudry, E.G., & Dudukovic, M.P. (1984). Comparison and prediction of reactor performance for packed beds with two phase #ow: Down#ow, up#ow and countercurrent #ow. Instn. Chem. Engrs. Symp. Ser., 87, 527}534. Mills, P.L., & Chaudhari, R.V. (1997). Multiphase catalytic reactor engineering for pharmaceuticals. Catal. ¹oday, 37, 367. Mills, P.L., Chaudhari, R.V., & Ramachandran, P.A. (1992). Multiphase reaction engineering for "ne chemicals and pharmaceuticals. Rev. Chem. Engng, 8 (1/2), 1. Mills, P.L., & Dudukovic, M.P. (1984). A comparison of current models for isothermal trickle-bed reactors. Application of a model reaction system. ACS Symp. Ser., 234, 37. Molga, E.J., & Westerterp, K.R. (1997a). Gas-liquid interfacial area and holdup in a cocurrent up#ow packed bed bubble column reactor at elevated pressures. I&EC Res., 36, 622. Molga, E.J., & Westerterp, K.R. (1997b). Experimental study of a cocurrent up#ow packed bed bubble column reactor: Pressure drop, holdup and interfacial area. Chem. Engng Proc., 36, 489. Moslemian, D., Devanathan, N., & Dudukovic, M.P. (1992). Radioactive particle tracking technique for investigation of phase recirculation and turbulence in multiphase systems. Rev. Sci. Instrum., 63, 4361. Mudde, R.F., Lee, D.J., Reese, J., & Fan, L.-S. (1997). Role of coherent structures on Reynolds stresses in a 2-D bubble column. A.I.Ch.E. J., 43, 913. Nacef, S. (1991). Hydrodynamique des lits -uidise& s gaz-liquide-solide. E+ets du distributeur et de la nature du liquide., Ph.D. thesis, Institut National Polytechnique de Lorraine, Nancy, France. Nore, O. (1992). ¸es lits -uidise& s triphasiques destine& s a% la biotechnologie, utilisant des particules de faible mass volumique: Hydrodynamique, transfert de chaleur et de matie% re., Ph.D. thesis, Institut National Polytechnique de Lorraine, Nancy, France. Nore, O., Briens, C., Margaritis, A., & Wild, G. (1992). Hydrodynamics, gas}liquid mass transfer and particle-liquid heat and mass transfer in a three-phase #uidized bed for biochemical process applications. Chem. Engng Sci., 47, 3573. Oyevaar, M.H., Morssinkho!, R.W.J., & Westerterp, K.R. (1990). The kinetics of the reaction between CO and diethanolamine in aqueous 2 ethylene-glycol at 298 K: A viscous gas-liquid reaction system for the determination of interfacial areas in gas-liquids contactors. Chem. Engng Sci., 45, 3283. Pintar, A., Bercic, G., & Levec, J. (1997). Catalytic liquid-phase oxidation of aqueous phenol solutions in a trickle-bed reactor. Chem. Engng Sci., 52, 4143. Rajashekharam, M.V., Jaganathan, R., & Chaudhari, R.V. (1998). A trickle-bed reactor model for hydrogenation of 2,4-dinitrotoluene: Experimental veri"cation. Chem. Engng Sci., 53, 787. Ravindra, P.V., Rao, D.P., & Rao, M.S. (1997). A model for the oxidation of sulfur dioxide in a trickle-bed reactor. I&EC Res., 36, 5125. Reese, J., Chen, R.C., Tzeng, J.-W., & Fan, L.-S. (1993). Characterization of the macroscopic #ow structure in gas-liquid and gas-liquidsolid #uidization systems using particle image velocimetry. Int. <ideo J. Engng Res., 3, 17. Reese, J., Chen, R.C. & Fan, L.-S. (1995). Three-dimensional particle image velocimetry for use in three-phase #uidization systems. Exp. Fluids, 19, 367. Reese, J., & Fan, L.-S. (1994). Transient #ow structure in the entrance region of a bubble column using particle image velocimetry. Chem. Engng Sci., 49, 5623. Reese, J., & Fan, L.S. (1997a). Hydrodynamics of a bubble column with a conical-shaped entrance region. Chem. Engng Sci., 52, 1553. Reese, J., & Fan, L.-S. (1997b). Particle image velocimetry: Application for the characterization of the #ow structure in three-phase #uidized beds. In: J. Chaouki, F. Larachi, & M.P. Dudukovic, (Eds.), Non-invasive monitoring of multiphase -ows (p. 495). Amsterdam: Elsevier. Reese, J., Jiang, P., & Fan, L.-S. (1996). Bubble characteristics in three-phase systems used for pulp and paper processing. Chem. Engng Sci., 51, 2501. Reinecke, N., & Mewes, D. (1997). Investigation of the two-phase #ow in trickle-bed reactors using capacitance tomography. Chem. Engng Sci., 52, 2111. Reinecke, N., Petritsch, G., Schmitz, D., & Mewes, D. (1998). Tomographic measurement techniques-visualization of multiphase #ows. Chem. Engng ¹echnol., 21, 7. Rihn, M.J., Zhu, X.Q., Suidan, M.T., Kim, B.J., & Kim, B.R. (1997). The e!ect of nitration VOC removal in trickle-bed bio"lters. =ater Res., 31, 2997. Ruzicka, J., & Hanika, J. (1994). Partial wetting and forced reaction mixture transition in a model trickle bed reactor. Catal. ¹oday. Proc. (Int. Symp. Dynamics of Catalytic Systems) 20, 467. Saroha, A.K., & Nigam, K.D.P. (1996). Trickle bed reactors. Rev. Chem. Engng, 12, 207. Schmitz, D., Reinecke, N., Petritsch, G., & Mewes, D. (1997). High resolution X-ray tomography for stationary multiphase #ows. OECD/CSNI Specialist Meeting on Advanced Instrumentation and Measurement Techniques, Sante Barbara, CA, 17}20, March. SchuK gerl, K. (1997). Three-phase bio#uidization } Application of threephase #uidization in the biotechnology } a review. Chem. Engng Sci., 52, 3661. Silveston, P. (1990). ;nsteady State Processes in Catalysis, (pp. 217}232). Sims, B.W., Gaskey, S.W., & Luss, D. (1994). E!ect of #ow regime and liquid velocity on conversion in a trickle-bed reactor. I&EC Res., 30, 2530. Sokolichin, A., & Eigenberger, G. (1994). Gas liquid #ow in bubble columns and loop reactors, Part I detailed modeling and numerical simulation. Chem. Engng Sci., 49, 5735. Stanek, V., & Jiriczny, V. (1997). Experimental observation of pressure drop overshoot following an onset of gas #ow in countercurrent beds. Chem. Engng J., 68, 207. Stegasov, A.N., Kirillov, V.A., & Silveston, P.L. (1994). Modeling of catalytic SO oxidation for continuous and periodic liquid #ow 2 through a trickle bed. Chem. Engng Sci., 49, 3699. Sto!els, M., Amann, R., Ludwig, W., Hekmat, D., & Schleifer, K.-H. (1998). Bacterial community dynamics during start-up of a tricklebed bioreactor degrading aromatic compounds. Appl. Environ. Microbiol., 64, 930. StuK ber, F., Wilhelm, A.M., & Delmas, H. (1996). Modeling of threephase catalytic up#ow reactor: A signi"cant chemical determination of liquid-solid and gas}liquid mass transfer coe$cients. Chem. Engng Sci., 51, 2161. Tahraoui, K., Ronze, D., & Zoulalian, A. (1992). Hydrodynamique et transfert gaz-liquide dans un reH acteur Verli"x rempli de di!eH rents garnissages. Can. J. Chem. Engng, 70, 636. Tang, W.T., & Fan, L.S. (1990). Gas-liquid mass transfer in a threephase #uidized bed containing low density particles. I&EC Res., 29, 128. Thulasidas, T.C., Cerro, R.L., & Abraham, M.A. (1995a). The monolith froth reactor: Residence time modelling and analysis. ¹rans IChemE, 73, 314. Tomiyama, A., Sou, A., Minagawa, H., & Sakaguchi, T. (1993). Numerical analysis of a single bubble by VOF Method. JSME Int. J. Series B: Fluids and ¹hermal Engng, 36, 51. Torvik, R., & Svendsen, H.F. (1990). Modeling of slurry reactors: A fundamental approach. Chem. Engng Sci., 45, 2325. Toye, D., Marchot, P., Crine, M., & Homer G.L. (1994). The use of large scale computer assisted tomography for the study of hydrodynamics in trickling "lters. Chem. Engng Sci., 49, 5271. Toye, D., Marchot, P., Crine, M., & L'Homme, G. (1996). Modeling of multiphase #ow in packed beds by computer-assisted X-ray tomography. Meas. Sci. ¹echnol., 7, 436. M.P. Dudukovic et al. /Chemical Engineering Science 54 (1999) 1975}1995 Toye, D., Marchot, P., Crine, M., & L'Homme, G. (1997). Computerassisted tomography for liquid imaging in trickle #ow columns, In: J. Chaouki, F. Larachi, & M.P. Dudukovic (Eds.), Non-invasive monitoring of multiphase -ows (p. 105). Amsterdam: Elsevier. Trambouze, P. (1990). Countercurrent two-phase #ow "xed bed catalytic reactors. Chem. Engng Sci., 45, 2269. Trambouze, P. (1993). Engineering of hydrotreating processes. In: H.I. De Lasa, G. Dogu, & A. Raella (Eds.), Chemical reactor technology for environmentally safe reactors and products, NATO Adv. Study Inst. Series E: Applied Sciences (p. 425). Tsamatsoulis, D., & Papayannakos, N. (1995). Simulation of non-ideal #ow in a trickle bed reactor by a cross-#ow model. Chem. Engng Sci., 50, 3685. Tsutsumi, A., Dastidar, A.G., & Fan, L.S. (1991). Characteristics of gas}liquid}solid #uidization with non-wettable particles. A.I.Ch.E. J., 37, 951. Tukac, V. (1997). Glucose hydrogenation in a trickle-bed reactor. Collect. Czech. Chem. Commun., 62, 1423. Tukac, V., & Hanika, J. (1997). Catalytic oxidation of substituted phenols in a trickle-bed reactor. Collect. Czech. Chem. Commun., 62, 866. Tukac, V., & Hanika, J. (1998). Catalytic wet oxidation of substituted phenols in the trickle-bed reactor. Chem. Engng ¹echnol., 71, 262. Tzeng, T.-W., Chen, R.C., & Fan, L.-S. (1993). Visualization of #ow characteristics in a two-dimensional bubble column and three-phase #uidized bed. A.I.Ch.E. J., 39, 733. Valerius, G., Zhu, X., Hofmann, H., Arntz, D., & Haas, T. (1996). Modelling of a trickle-bed reactor. II. The hydrogenation of 3-hydroxypropanal to 1,3-propanediol. Chem. Engng Proc., 35, 11. Van Hasselt, B.W., Lindenbergh, D.J., Calis, H.P., Sie, S.T., & Van Den Bleek, C.M. (1997). The three-levels-of-porosity reactor. A novel reactor for counter-current trickle-#ow processes. Chem. Engng Sci., 52, 3901. Vergel, C., Euzen, J.P., Trambouze, P., & Wauquier, J.P. (1995). Twophase #ow catalytic reactor, in#uence of hydrodynamics on selectivity. Chem. Engng Sci., 50, 3303. Villermaux, J. (1993). Future challenges for basic research in chemical engineering. Chem. Engng Sci., 48, 2525. Wang, R., Mao, Z.-S., Chen, L., & Chen, J.Y. (1997). Experimental evidence of hysteresis of pressure drop for countercurrent gas-liquid #ow in a "xed bed, Chem. Engng Sci., 53, 367. Warna, J., & Salmi, T. (1996). Dynamic modeling of three phase reactors. Comput. Chem. Engng, 20, 39. Weissermel, K., & Arpe, H.-J. (1993). Industrial organic chemistry, (2nd ed.). Weinheim: VCH Press. Westerterp, K.R., Molga, E.J., & van Gelder, K.B. (1997). Catalytic hydrogenation reactors for the "ne chemicals industries. Their design and operation. Chem. Engng Proc., 36, 17. Westerterp, K.R., & Wammes, W.J.A. (1992). Three-phase trickle-bed reactors. In Principles of chemical reaction engineering and plant design, Ullman's Encyclopedia of Industrial Chemistry, (5th Ed., Vol. B4, p. 309). Wild, G., Larachi, F., & Charpentier, J.C. (1992). Heat and mass transfer in gas-liquid-solid "xed bed reactors. In: M. Quintard, M. Todorovic (Eds.), Heat and mass transfer in porous media (p. 616). Amsterdam: Elsevier. 1995 Wild, G., Larachi, F., & Laurent, A. (1991). The hydrodynamic characteristics of cocurrent down#ow and cocurrent up#ow gas-liquid-solid catalytic "xed bed reactors: The e!ect of pressure. Rev. Inst. Franc7 . Pe& tr., 46, 467. Wild, G., & Poncin, S. (1996). Hydrodynamics. In K.D.P. Nigam, A. Schumpe (Eds.), three-phase sparged reactors. Topics in Chem. Engng., (Vol. 8, pp. 11}12). Amsterdam: Gorden & Breach, Overseas Publishers Association. Wright, P.C., & Rapper, J.A. (1996). A review of some parameters involved in #uidized bed bioreactors. Chem. Engng ¹echnol,. 19, 50. Wu, R., McReady, M.J., & Varma, A. (1995). In#uence of mass transfer coe$cient #uctuation frequency on performance of three-phase packed bed reactors. Chem. Engng Sci., 50, 3333. Wu, R. (1997). Enhancing performance of three-phase packed-bed catalytic reactors by pulsing -ow regime: modeling and experimental study. Ph.D. thesis. Univ. of Notre Dame, Notre Dame, USA. Wu, Y., Al-Dahhan, M.A., Khadilkar, M.R., & Dudukovic, M.P. (1996a). Evaluation of trickle-bed reactor models for a liquid -limited reaction. Chem. Engng Sci., 51, 2721. Wu, Y., Khadilkar, M.R., Al-Dahhan, M.H., & Dudukovic, M.P. (1996b). Comparison of up#ow and down#ow two-phase #ow packed-bed reactors with and without "nes: Experimental observations. I&EC Res., 35, 397. WuK bker, S.-M., Laurenzis, A., Werner, U., & Friedrich, C. (1997). Controlled biomass formation and kinetics of toluene degradation in a bioscrubber and in a reactor with a periodically moved trickle bed. Biotechnol. BioEngng, 55, 686. Yamada, H., & Goto, S. (1997). Periodic operation of trickle bed reactor for hydrogenolysis in gas-liquid-liquid-solid four phases. J. Chem. Engng Japan., 30, 478. Yan, T.Y. (1980). Dynamics of a trickle bed hydrocracker with a quenching system. Can. J. Chem. Engng, 58, 259. Yang, J., & Li, J. (1992). Simulation of hydrodewaxing reactor by stochastic method. J. Chem. Engng Chin. ;niv., 6, 264. Yang, X.L., Wild, G., & Euzen, J.P. (1992a). A comparison of the hydrodynamics of packed-bed reactors with cocurrent up#ow and down#ow of gas and liquid. Chem. Engng Sci., 47, 1323. Yang, Y.B., Devanathan, N., & Dudukovic, M.P. (1992b). Liquid backmixing in bubble columns via computer automated particle tracking (CARPT). Chem. Engng Sci., 47, 2859. Yang, Y.B., Devanathan, N., & Dudukovic, M.P. (1993). Liquid backmixing in bubble columns via computer automated particle tracking (CARPT). Exp. Fluids, 16, 1. Yao, B.P., Zheng, C., Gasche, H.E., & Hofmann, H. (1991)). Bubble behavior and #ow structure in bubble columns. Chem. Engng. Process., 29, 25}75. Zhang, J.P., Epstein, N., Grace, J.R., & Zhu, J. (1995). Minimum liquid #uidization velocity of gas-liquid #uidized beds. ¹rans IChemE part A, 73, 347. Zhang, J.P., Grace, J.R., Epstein, N., & Lim, K.S. (1997). Flow regime identi"cation in gas-liquid #ow and three-phase #uidized beds. Chem. Engng Sci., 52, 3979. Zhu, X.D., & Hofmann, H. (1997). Hydrogenation of 3-hydroxypropanal in trickle beds } Experimental and modeling. Chem. Engng ¹echnol., 20, 131. Zhukova, T., Pisarenko, B.V.N., & Kafarov, V.V. (1990). Modeling and design of industrial reactors with a stationary bed of catalyst and two-phase gas-liquid #ow } a review. Int. Chem. Engng, 30, 57.