eduzhai > Physical Sciences > Materials Sciences >

Membrane diffusion convection mass transfer catalyzed by dispersed nano catalysts

  • sky
  • (0) Download
  • 20211030
  • Save
https://www.eduzhai.net International Journal of Composite M aterials 2012, 2(5): 79-91 DOI: 10.5923/j.cmaterials.20120205.03 Diffusive Plus Convective Mass Transport Through Catalytic Membrane Layer with Dispersed Nanometer-Sized Catalyst Endre Nagy Research Institute of Chemical and Process Engineering, University of Pannonia, 8200, Veszprém, Egyetem u. 10, Hungary Abstract Mass transfer rates across catalytic memb rane interfaces accompanied by first-order, irreversible reactions have been investigated. The catalyst particles impregnated in the membrane mat rix are assumed to be very fine, nanometer-sized particles which are uniformly distributed in the structure of the membrane layer. Pseudo-homogeneous models have been developed to describe mass transport through this catalytic membrane layer. The models developed include the mass transport into and inside the catalytic part icles as well as through the memb rane layer taking into account convective and diffusive flows, so it is also valid in the limiting cases namely without convective flow (Pe=0), or with very large convective flow (Pe >> 1). The models describe two operating modes (with and without sweep phase on the permeate side of the catalytic memb rane layer) and apply two different boundary conditions for the feed boundary layer. One of the boundary conditions approaches the diffusive flow by the Fickian one assuming linear concentration distribution while the other one solves exactly the mass transport in the feed boundary layer. The different model results obtained are compared to each other proving the importance of the carefully decision of the operating modes and boundary conditions. The mathematical model has been verified by means of experimental data taken fro m the literature. Keywords Catalytic Membrane Layer, Dispersed Nanometer-Sized Catalytic Particles, First-Order Irreversible Reaction, Pseudo-Homogeneous Model 1. Introduction The catalytic memb rane reactor as a pro mising novel t ech no log y is widely reco mmend ed fo r carry ing out heterogeneous reactions. A number of reactions have been in v es t ig at ed b y mean s o f t h is p ro ces s , s u ch as dehydrogenation of alkanes to alkenes, partial o xidation reactions using inorganic or organic pero xides, as well as p art ial hy d rog en at io ns , h yd rat io n , et c. As cat alyt ic memb rane reactors for these reactions, intrinsically catalyt ic memb ranes can be used (e.g. zeolite or metallic memb ranes) or memb ranes that have been made catalytic by dispersion or imp regnation of catalytically act ive part icles such as metallic co mplexes, metallic clusters or activated carbon, zeo lite p art icles, et c. throughout dense po ly meric - or inorganic membrane layers[1]. In the majority of the above experiments, the reactants are separated from each other by the catalytic membrane layer. In this case the reactants are absorbed into the catalyt ic memb rane matrix and then t rans po rt ed by d iffus io n (and co nv ect io n ) fro m t he * Corresponding author: nagy@mukki.richem.hu (Endre Nagy) Published online at https://www.eduzhai.net Copyright © 2012 Scientific & Academic Publishing. All Rights Reserved memb rane interface into catalyst particles where they react. Mass transport limitation can be experienced with this method, which can also reduce selectivity. The application of a sweep gas on the permeate side dilutes the permeating component, thus increasing the chemical reaction gradient and the driving force for permeation (e.g. see Westermann and Melin[2]). At the present time, the use of a flow-through catalytic membrane layer is recommended more frequently for catalytic reactions[2]. If the reactant mixtu re is forced to flo w through the pores of a membrane which has been impregnated with catalyst, the intensive contact allows fo r h igh catalytic activity with negligib le diffusive mass transport resistance. By means of convective flow the desired concentration level of reactants can be maintained and side reactions can often be avoided (see review by Julbe et al.[3]). When describing catalytic processes in a memb rane reactor, therefore, the effect of convective flow should also be taken into account. Yamada et al.[4] reported isomerizat ion of 1-butene as the first application of a catalytic membrane as a flo w-through reactor. Th is method has been used for a nu mber of gas-phase and liquid-phase catalytic reactions such as VOC decomposition[5], photocatalytic oxidation[6], part ial oxidation[7], part ial hydrogenation[8-10] and hydrogenation of nitrate in water[11]. 80 Endre Nagy: Diffusive Plus Convective M ass Transport Through Catalytic M embrane Layer with Dispersed Nanometer-Sized Catalyst Fro m a chemical engineering point of view, it is important to predict the mass transfer rate of the reactant entering the memb rane layer fro m the upstream phase, and also to predict the downstream mass transfer rate on the permeate side of the catalytic memb rane as a function of the physico-chemical parameters. If this transfer (permeat ion) rate is known as a function of the reaction rate constant, it can be substituted into the boundary conditions of the differential mass balance equations for the upstream and/or the downstream phases. Basically, in order to describe the mass transfer rate, a heterogeneous model can be used for larger particles and/or a pseudo-homogeneous one for very fine catalyst particles[12]. Both approaches, namely the heterogeneous model fo r larger catalyst particles and the homogeneous one for submicron part icles, have been applied for mass transfer through a catalytic membrane layer. Nagy[12] has analysed diffusive mass transport through a membrane layer with dispersed catalyst particles. It was shown that both the heterogeneous and the pseudo-homogeneous models give practically the same results in the very fine, sub-micro meter particle size range. The convective velocity was not taken into account in Nagy’s model[12] cited. Recently Nagy analysis the effect of the convective velocity on the enzyme catalysed reaction[13] as well as summarizes the most important mass transport equations of a memb rane layer taking also into account the simultaneous effect of the convective and diffusive flows[14,15]. These papers extend previous investigations by including the effect of convective flow, applying two different operating modes, namely with and without sweep phase on the permeate side as well as two different models, namely an approaching and the exact models. Mathematical equations have been developed to describe the simultaneous effect of diffusive flow and convective flow and this paper analyses mass transport and concentration distribution by applying the model developed. The pseudo-homogeneous model will be presented in detail, assuming that the catalyst particles are in the nanometer-sized range, as this is the case in most catalytic memb rane reactors. The main purpose of this paper is to present the various steps of the mathemat ical solutions, as well as to study the effect of mass transport parameters of the catalyt ic memb rane layer on the mass transfer rates. At the end of this paper, the predicted data are co mpared with measured ones taken fro m the literature. The method presented in this paper can also be applied to higher order chemical reactio n s . 2. Theory In this section different mathemat ical models will be shown and solved in order to describe the mass transport through catalytic memb rane layer with forced flow through it. It is assumed that the catalyst particles are very fine particles with size less than 1 μm. Thus, the so called pseudo-homogeneous model[12,14,15] was applied for description of the mass transport through the catalytic memb rane layer. The catalyst particle can be porous one (e.g. zeo lite part icles), or dense one without diffusivity inside it (e.g. metal clusters). Accordingly, d ifferent mass transfer rate equations can be defined between the memb rane and particles as will be shown in this section. In order to increase the efficiency of the catalytic memb rane, the size of particles chosen should be as low as possible[12]. The use of nanometer-sized catalyst particles is thus recommended. The differential mass balance equation for diffusive and convective flows in the catalytic membrane layer, for unsteady-state, can be as: (1 − ε )D ∂2C ∂y2 − (1 − ε )υ ∂C ∂y − Q = (1 − ε) ∂C ∂t (1) with init ial and boundary conditions as: if t = 0, y > 0 then C=Co if t > 0, y = 0 then C=C* if t > 0, y = δ then C=Co The functions Q provides the specific source term induced by the mass transport into the catalytic nanoparticles distributed uniformly in the catalytic memb rane layer. The mass transfer rate into the catalytic particle can depend on the external mass transfer resistance around the particles as well as on the internal transport accompanied by chemical reaction. Note that in the case of nanometer-sized part icles the diffusion time, tD, (tD=R2/Dp) can be very small, thus, it can easily be much shorter than the reaction t ime, tr (tr=1/k1 for first-order reaction). Th is fact should be taken into account to define the Q value. E.g. for dp=100 n m and Dp=1 x10-12 m2/s, the diffusion time is equal to 0.01 s. In the slow reaction rate regime, if λ = tD / tr ≡ k1R2 / Dp < 0.3 , the saturated concentration, C*, can exist throughout the particle, i.e. if tr < 9 s according to the above example. In this case the so called effect iveness factor is considered to be unit and accordingly the internal mass transport can be regarded to be in s tan tan eou s . 2.1. Mass Trans port into the Dis persed Nanometer-Sized Catal yst Particles Two important cases will be discussed, namely instantaneous and finite internal mass transport rates. Both cases can be important when chemical react ion takes place inside of the particles or on the particle surface. As chemical reaction, the first-order one will be discussed. Several non-linear reactions can be approached by first-order one dividing the membrane layer into thin sub-layers as will be shown in the Appendix. Accordingly these reactions can be handled as first-order ones. 2.1.1. Mass Transport Rates with Instantaneous Internal Mass Transport The chemica l reaction takes place on the internal interface of the catalyst particles with the reaction rate: Q = ~k1ω~ C* ≡ k1C* where ~k1 is the react ion rate constant International Journal of Composite M aterials 2012, 2(5): 79-91 81 related to the catalyst interface, m3/(m2s), ω~ is the availab le catalytic surface area per unit volume of catalyst, m2/m3, and k1 also rate constant, 1/s ( k1 = ~k1ω). Thus, one can obtain applying the known Henry equation for the catalyst interface, namely HC= C*p as: Q = k1C*p = k1HC (2) In Eq. (2) the external mass transfer resistance is neglected, thus the interface concentration of the catalytic particles is equal to the “bulk” one in the me mbrane matrix. In the case of inorganic catalyst particles, as zeo lite, the Lang muir-Hinshelwood kinetics the most co mmon ly used kinetic exp ression to explain the heterogeneous catalytic process[16]. Assuming that the react ion of a co mponent occurs in a simple unimo lecular elementary reaction step and that the kinetics are first order with respect to the surface concentration of the adsorbed reactant, the reaction rate can be expressed as[16]: Q = kKC 1+ KC (3a ) This reaction is often approximated to first-order kinetics for condition KC << 1 or to zero-order kinetics for condition KC >> 1. Similar kinetic expression can be applied for biochemical reactions according to the Michaelis-Menten or Monod kinetics: Q = vmaxC Km +C (3b) Replacing Eqs. (3) into Eq. (1), the d ifferential equation obtained for steady-state case can only be solved by analytical approach (this will be shown in Appendix) or numerically. How the Eqs. (3) can be linearized is given by Eq. (A1) in the APPENDIX. 2.1.2. Mass Transport with Finite Internal Mass Transport It the reaction rate is fast then there can be strong concentration gradient inside of the particle, thus the internal transport should also be taken into account. The internal specific mass transfer rate in spherical particles, Jp, for steady-state conditions and when mass transport is accompanied by a first-order chemica l reaction, can be given according to Bird et al.[17] as follows: J p = βpC*p (4) where βp = Dp R   Ha p tanh Ha p −1 (5) and Ha p = k1R 2 Dp The external mass transfer resistance through the catalyst particle depends on the thickness of the diffusion boundary layer, δp. The value of δp can be estimated fro m the distance between particles[12]. As this value is limited by neighboring particles, the value o f βp will be so mewhat higher than that calculated from the well known equation, namely 2 = βpdp / Dp, where the value of δp is assumed to be infinite. Th is results in: β op = 2D dp + D δp (6) where δ p = h − dp 2 Fro m Eqs. (5) and (6), for the mass transfer rate with overall mass transfer resistance with Hp = Cp/C: Jp = βsumC = 1 β o p C +1 Hpβp (7) Thus, the value of Q will be as: Q = ωJ p ≡ ωβsumC (8) where ω is the catalyst surface per memb rane volume, m2/ m3 2.1.3. Reaction Occurs on the Outer Interface of the Catalyt ic Partic les It often might occur that the chemica l react ion takes place on the interface of the particles, e.g. in cases of metallic clusters, the diffusion inside the dense particles are negligibly. Assuming the Henry’s sorption isotherm of the reacting component onto the spherical catalytic surface (CHf = qf), applying Dd C/dr = kfHfC boundary condition at the catalyst’s interface, at r = R, the Φ reaction modulus can be given for first-order reaction, as follows[see Eq. (11) for Φ]: Φ = ωδ 2 D(1− ε ) β sum (9) where ( ) βsum = δp R + δp + 1 1 R + δp 2 (10) DR kf Hf R2 where kf is the interface reaction rate constant. The above model is obviously a simp lified one. 2.2. Mass Trans port in the Catalytic Membrane Layer Taking into account Eqs. (2) or (8) as source term, one can get simple first-order kinetics. The differential mass balance equation for the polymeric or macroporous ceramic catalyt ic memb rane layer, fo r steady-state, taking both diffusive and convective flo w into account, can be given, according to Eq. (1), as: d2C dY 2 − Pe dC dY − Φ2C = 0 (11) where Pe = υδ Φ = D; ωδ 2 D(1− ε ) βsum ; or 82 Endre Nagy: Diffusive Plus Convective M ass Transport Through Catalytic M embrane Layer with Dispersed Nanometer-Sized Catalyst Φ = ωδ 2 D(1− ε) k1 where υ denotes the convective velocity, D is the diffusion coefficient of the membrane, and δ is the membrane thickness. The membrane concentration, C is given here in a unit of measure of g mol/ m3. This can be easily obtained by means of the usually applied in the e.g. g/g unit of measure with the equation of C = wρ/M, where w concentration in kg/kg, ρ – membrane density, kg/m3, M-mo lar weight, kg/mo l. Introducing a new C~ = Ce− variable, PeY / C~ 2 [s ee Eq. (12) (12)] the following diffe rential equddaY2tiC~o2n−isΘo2bC~tai=ne0d from Eq. (11): (13) where Θ = Pe2 + Φ 2 4 The general solution of Eq. (13) is we ll known[14], so the concentration distribution in the catalytic memb rane layer can be given as follows: C = Teλ~Y + SeλY (14) with λ = Pe − Θ λ~ = Pe + Θ 2 2 The inlet and the outlet mass transfer rate can easily be expressed by means of Eq. (14). The overall in let mass transfer rate, namely the sum of the diffusive and convective mass transfer rates, is given by: ( ) J = υC Y=0 − D dC dX Y=0 = βo λT + λ~S (15) The outlet mass transfer rate is obtained in a similar way to Eq. (15) fo r Y=1: ( ) Jout = βo λTeλ~ + λ~Seλ (16) The value of parameters T and S can be determined fro m the boundary conditions which can vary according to the operating conditions used. Assuming that the external mass transfer resistance on the feed side, namely the diffusive resistance, is not negligible, the simultaneous effect of the flu id boundary layer should also be taken into account. Basically, the effect of the external, diffusive mass transfer resistance, in presence of convective flow can be described by two models: (i) the diffusive mass flow is regarded to be constant through the boundary layer (this diffusive flow is called as Fickian one in this paper), accordingly the sum of the diffusive and convective flows varies in the boundary layer (Model A) or (ii) that it varies in the boundary layer, according to the change of the convective mass flow due to the curvature of the concentration in the boundary layer (Model B). This latter one should be regarded as exact solution, namely the su m of the d iffusive and convective flows is constant throughout the boundary layer. According to the mass transfer conditions of the permeate phase, namely there is a sweep phase or there is not sweep phase, two operating modes will be distinguished according to Figs. 1a and 1b. catalyst particles CLo1 CLo1 CLo2 CLo2 a) b) β1° β° β2° Figure 1. Concentration distribution in the membrane reactor with convective flow applying a sweep phase on the permeation side (Fig. 1a) and without sweep phase (Fig. 1b) The essential d ifference between the models is that there is a sweep phase that can remove the transported component fro m the downstream side providing the low concentration of the reacted component and its high d iffusive mass transfer rate (d C/d Y > 0; Models A1 and B1). There is no sweep phase, thus the outlet phase is moving by convective flow fro m the memb rane due to the lower pressure on the permeate side and there is no diffusive flow, on the outlet membrane interface (d C/d Y = 0; Models A2 and B2). 2.2.1. Mass Transport Models with Fickian Diffusive Flow in the Boundary Layer (Approaching Solution, Models A) The simultaneous effect of the memb rane and the boundary layers, on the mass transport, is taken into account. In presence of the convective flo w, the overall mass transfer rate will be the sum of the d iffusive and convective flows. Regarding the effect of the boundary layers on both sides of the membrane, the boundary conditions can be different on the feed side and permeate side, depending on the operating mode. Note that the Fickian diffusive flow along the diffusion path[as it is given in Eqs. (17) and (18)] assumed that the concentration distribution is linear, the concentration gradient is constant, in the boundary layer, independently of the presence of convective flow. Accordingly, the sum of the diffusive and the convective flow will change throughout the boundary layer due to the concentration change. In the reality, the concentration curve will be concave one due to the convective velocity thus, the sum of the diffusive and the convective flo w remains constant as a function of the local coordinate in the boundary layer. Model A1 (dC/dY>0 at Y=1). In this case, due to the effect of the sweeping phase, the external mass transfer resistance International Journal of Composite M aterials 2012, 2(5): 79-91 83 on both sides of the memb rane should be taken into account in the boundary conditions, though the role of βo2 is gradually dimin ished as the catalytic reaction rate increases. The concentration distribution in the catalytic memb rane, when applying a sweep phase on the two sides of the memb rane, is illustrated in Figure 1a. On the upper part of the catalytic membrane layer, in Fig. 1a, the fine catalyst particles are illustrated with black dots. It is assumed that these particles are homogeneously distributed in the memb rane matrix. Due to sweeping phase, the concentration of the bulk phase on the permeate side may be lower than that on the membrane interface. The value o f CoL here denotes the liquid or gas phase concentration on the bulk phases (see Fig. A1), on both sides of the catalytic membrane layer. The boundary conditions can be given for that case as: ( ) υC∗L2 + β1o C o L1 − C∗L2 =J at Y=0 (17) ( ) υC∗L2 + βo2 C∗L 2 − C o L 2 = Jout at Y=1 (18) Boundary conditions given by Eqs. (17) and (18) are only valid in two phase flows. Where C∗L1 and C∗L2 denotes the interface concentration on the both sides of memb rane layer, β1o and βo2 are mass transfer coefficients in the continuous phase, βo the membrane mass transfer coefficient (βo = D/δ), H denotes the distribution coefficient between the continuous phase and the membrane phase. The solution of the algebraic equations obtained, applying Eqs. (14) to (18) can be received by means of known mathematical man ipulations. Thus, the values of T and S obtained are as fo llo ws : T = − β1o ϕ 2C o L1 − βo2 ϕ 4C o L2 ϕ2ϕ3 − ϕ1ϕ4 1 βo (19) and S= β1oϕ1CoL1 + βo2ϕ3CoL2 ϕ2ϕ3 − ϕ1ϕ4 1 βo (20) where ϕ1 =  Pe H + β o 2 βoH − λ e λ~ ; ϕ2 =  Pe H + β o f2 βoH − λ~ eλ ; ϕ3 = Pe H − β1o βoH −λ ; ϕ4 = Pe H − β1o βoH − λ~ ; An important limiting case should also be mentioned, namely the case when the external d iffusive mass transfer resistances on both sides of membrane can be neglected, i.e. when β1o → ∞ and βo2 → ∞ . For that case the concentration distribution and the inlet mass transfer rate can be expressed by Eqs. (21) and (22), respectively[14]. { } [ ] C = ePe(Y−1)/ 2 sinh Θ H CoL1ePe / 2 sinh Θ(1 − Y) + C o L 2 sinh(ΘY) (21)   J = β   C o L1 − exp  Pe 2  Pe 2 Θ sinh Θ + Θ cosh Θ C o L2     (22) with H Pe tanh Θ + Θ β = βo  2  (23) tanh Θ Model A2 (dC/dY=0 at Y=1). For the convective flow catalytic membrane reactor operating in another mode, for instance in dead-end mode as in Figure 1b, the boundary condition on the permeate side of the memb rane should be changed. In this case the concentration of the permeate phase does not change during its transport from the outer memb rane interface, and consequently dC/dY = 0. If there is no sweeping phase on the downstream side then the correct boundary conditions will be as[the value of J is defined by Eq. (15)]: ( ) υC∗L1 + β1o CoL1 − C∗L1 = J at Y=0 (24) Tλ~eλ~Y + Sλeλ = 0 at Y=1 (25) After solution one can get as: T = 1 βo ϕ4 βλ~1oeCλ~ λeλ o L1 − ϕ3 (26) S= 1 βo ϕ3 β1o C o L1 λeλ λ~e λ~ − ϕ4 27) The values of φ3 and φ4 are the same as they are g iven after Eq. (20). Knowing the values of T and S parameters, the in let or the outlet mass transfer rates can be calculated by means of Eqs. (15) and (16), respectively. 2.2.2. Models with constant diffusive plus convective flows in the boundary layer (exact solution, Models B) This, so called exact, model takes into account that the concentration change is not linear due to the presence of the convective flow but the sum of the convective and the diffusive flows will be constant throughout the boundary layer. The d ifferential mass balance equation, given for the boundary layer, should be solved. That is, with constant transport parameters and in dimensionless space coordinate as (Y=y/ δ1, δ1 is the thickness of the feed boundary layer, m): d2CL dY 2 − Pe L1 dCL dY =0 (28) where (PeL1 denotes the Peclet number in the boundary layer) 84 Endre Nagy: Diffusive Plus Convective M ass Transport Through Catalytic M embrane Layer with Dispersed Nanometer-Sized Catalyst Pe L1 = υδ1 D1 ≡ υ β1o The solution of differential equation, Eq. (28) is as fo llo ws [1 4] : CL = TePeL1Y + S (29) Values of T and S parameters can be determined, without external d iffusive mass transfer resistances, by the boundary conditions as: At Y=0 CL = CoL1 (30) At Y=1 CL = C*L1 (31) The concentration distribution, applying Eqs. (30) and (31) as boundary conditions, is as follows: ( ) CL1 = ePeL1Y /2 sinh PeL1 / 2  sinh   PeL1 2 [1 − Y ]   CLo1     (32) + e−PeL1/2 sinh   PeL1Y 2   CL*1   Thus, the physical mass transfer rate, through the boundary layer, can be obtained, by means of Eq. (33), taking into account both the diffusive and the convective ( ) flows[see Eq. (15)] as follows: Jo = β o ◊ CoL1 − e-PeL1 C*L1 (33) where β◊o = D1 δ1 PeL1 ePeL1 ePeL1 −1 ≡ β1o PeL1 2 e PeL1 / 2 sinh ( PeL1 / 2) (34) Note that the physical mass transfer coeffic ient in the case of the simu ltaneous effect of the diffusion and convection, βo◊ , ( βo◊ denotes the overall mass transfer coefficient in presence of convective plus diffusive flows) can be much higher than that of the diffusive mass transfer coefficient, β1o = D1 / δ1 . The βo◊ value increases linearly if PeL1 > about 3. It is also to be noted that the driving force depends on the Peclet number as it is given by Eq. (33). W ith the increase of the PeL1-number, the driving force also increases. In limiting case, namely if PeL1→∞, the value of e−PeL1C∗L1 = 0 . Note that if the diffusive flo w is equal to zero on the down stream s ide of the boundary layer then the mass transfer rate on the down stream side is as: J = υ CoL2 , thus this fact does not automatically mean that the outlet concentration gradient is equal to zero. Knowing the mass transfer rate into the boundary layer (Eq. 33) and the memb rane layer[Eq. (22) for Model A1 and Eq. (39) for Model B1], applying the well-known resistance-in-series model, the overall mass transfer rate can be given for the above case, as well. Model B1 (dC/dY>0 at Y=1): The overall in let mass transfer rate, applying Eqs. (22) and (33) is, for first-order chemical reaction in the memb rane layer, as follows: ( ) = J◊ov βov CLo1 − Ke− C PeL1 o L2 (35) where βov = 1 β1o◊ 1 + e− PeL1 Hβ (36) and K = Θ exp  Pe 2  Pe 2   sinh Θ + Θ cosh Θ (37) The value of β is defined by Eq. (23). Note that the mass transfer resistance in the boundary layer of the permeate side is not involved in Eq. (35). It can be g iven, but that case is not discussed here because its complexity and of its lesser importance due to the chemical reaction. Model B2 (dC/dY=0 at Y=1). Look at first the concentration distribution and the mass transfer rate in the memb rane layer fo r that case, namely when dC/d Y = 0 at Y=1. Assuming that there is no mass transfer resistance in the feed phase, thus, at Y=0 then C = C∗ ≡ HC o L1 , Applying to it Eq. (25) as boundary condition for the permeate side, and using Eq. (14), one can get as[13]: C = exp  PeY 2   sinh[Θ(1 − Pe Y )] Pe 2 sinh Θ + + Θ Θ cosh[Θ(1 cosh Θ − Y )] HC o L1 (38) 2 with Φ and Θ defined after Eqs. (11) and (13). Note that Eq. (38) does not involve the external mass transfer resistances. The inlet mass transfer rate, namely the sum of the diffusive and the convective flows, can be given as: J◊ = βo  Pe 2 4 + Θ2  tanh Θ + PeΘ  Pe tanh Θ + Θ HC1o ≡ βC1o (39) 2  The overall mass transfer rate, applying Eqs. (33) and (39), will be as: J ◊ov = βovC1o (40) with βov = 1 βo 1◊ 1 + e−PeL1 Hβ (41) where β1o◊ and β is defined by Eqs. (34) and (39). 2.3. Intrinsically Catalytic Membrane Reactor International Journal of Composite M aterials 2012, 2(5): 79-91 85 In this case, only the value of Ф differs fro m that of the catalytic memb rane layer with dispersed catalyst particles. In the case of a first-order reaction, the value of Ф can be expressed for an intrinsic memb rane layer with the following simp le equation: Φ = k1δ2 (42) D The mass transfer rate o r the concentration profile can be estimated in a similar way to that for a memb rane layer with dispersed catalyst particles. This case is not discussed in detail in this paper. The effect on the mass transfer rate and concentration distribution in the membrane reactor is basically the same as that obtained by the model developed for dispersed catalyst particles. Note that the application of the Fickian diffusion flow for the inlet and the outlet boundary layers[Eqs. (17) and (18)] is an approximation. The question is that its application can be acceptable and under what conditions. Typical concentration distribution curves of a catalyt ic membrane layer are shown for Model A1 (Fig. 2) and for Model B1 (Fig. 3). The concentration was predicted by Eqs. (19) and (20) as well as by Eq. (21). 3. Results and Discussion Two important transport models for convection flow catalytic membrane layer are presented in this paper. The difference between them is determined by the flow conditions on the permeate side, namely the permeated component is transported by a sweep phase fro m the memb rane interface (d C/d Y>0 at Y=1, Models A1 and B1) or there is no sweep phase (dC/dY=0 at Y=1, Models A2 and B2) on the permeate side. Obviously, these models can give essentially different in let mass transfer rate. On the other hand, two important cases are also discussed regarding the external mass transfer resistance, namely modeled it by the so called Fickian diffusion flow (Models A1 and A2) and by the so called exact model, where the diffusion flow permanently increases due to the decreasing convective flow (their sum should be cons tant throughout the boundary layer) on the diffusion path in the boundary layer (Models B1 and B2). 3.1. Models for Membrane Reactor wi th S weep Phase on the Permeate Side (dC/ dY>0 at Y=1) Figure 2. Concentration distribution in the membrane layer at different values of the reaction modulus, Φ, applying the Fickian diffusion flow in the boundary layers (Model A1 dC/dY>0 at Y=1); ( β1o = βo = 1x10−4 m/s; βo2 → ∞ ; Pe=1; PeL1=1;[ Pe L1 = Peβo / β1o ]; H=1; C o L 2 =0) Figure 3. Concentration distribution in the membrane layer at different values of the reaction modulus, Φ, applying the exact diffusion flow in the inlet boundary layers (Model B1 dC/dY>0 at Y=1); (β1o = βo = 1x10−4 m/s; β o 2 → ∞ ; Pe=1; PeL1=1;[ PeL1 = Peβo / β1o ]; H=1; C oL 2 =0) The value of C*L1, (see Fig. A 1) in Eq. (21), was predicted by means of Eq. (33) knowing the J value fro m Eq. (22). As you know there is a sweep phase on the permeate site, consequently there is a diffusive flow, as well on the outlet memb rane interface, in both cases. The transport parameters were chosen to be the same for both models ( β1o = βo = 1x10−4 m/s; βo2 → ∞ ; Pe=1; PeL1=1; PeL1 = Peβo / β1o , thus, the ν convective velocity was the same in the boundary and the memb rane layers; H=1; C o L2 = 0 ). The two models g ive significantly different concentration distribution. Accordingly the concentration gradient, and thus, the overall mass transfer rates will be d ifferent. The difference between Models A1 and B1 at e.g. Φ =0.01 is caused by the curvature of the of the concentration distribution in the boundary layer (not shown here) due to its convective velocity, namely PeL1=1 for Model B1. It is interesting to note that the inlet concentration increases with the increase of Φ value for the exact solution, i.e. for Model B1. Let us look at the inlet mass transfer rates of the two models as a function of the reaction rates. Fig. 4 shows it when the Fickian diffusion flow is applied for the boundary layers (Model A1). Jo represents the physical mass transfer rate into the catalytic membrane, this being the sum of the diffusive and convective flows at Ф=0. As can be seen, the tendency of the curves is different in the reaction ranges Pe < 1 and Pe > 1. In the first case, the value of J/Jo increases with increasing value of Ф. In the range o f, Pe > 1, however, the mass transfer rate decreases as a function of Ф. Perfectly 86 Endre Nagy: Diffusive Plus Convective M ass Transport Through Catalytic M embrane Layer with Dispersed Nanometer-Sized Catalyst other trend is shown by the exact model in presence of sweep phase on the permeate side (Model B1, Fig. 5). As it is expected, the mass transfer rate g radually increases with the increase of the reaction rate. On the other hand, the effect of the reaction rate decreases gradually with the increase of the Peclet nu mber (note here also PeL1=Pe because β1o = βo ). Obviously, the two models gives the same mass transfer rate when there is no convective flow in the boundary layer (at Pe=0.01 it is practically true), but the difference strongly increases with the increase of the Pe-nu mber. Figure 6. The relative values of masstransfer rates of different models as a function of the Peclet number (JModel B1/JModel A1) (parameters as in Fig 2 excluding β1o and consequently Pe1) 3.2. Models for Membrane Reactor Without S weep Phase on the Permeate Side (dC/ dY=0 at Y=1 ) Typical concentration distributions are shown in Figs. 7 and 8, for Fickian diffusion rate in the boundary layer (Model A2) and for the e xact solution (Model B2). Figure 4. Enhancement as a function of the reaction modulus, Φ, for the case of Fickian diffusion flow in the boundary layer with sweep phase on the permeate side (Model A1; parameters as in Fig.2 excluding Pe) Figure 5. Enhancement as a function of the reaction modulus, Φ, for the case of exact model with sweep phase on the permeate side (Model B1; parameters as in Fig.2 excluding Pe) The ratio of the mass transfer rates is plotted in Fig. 6, name ly the approaching solution (Model A1) is related to the e xact one, (Model B1, i.e. JModel B1/JModel A1 is plotted, noted by B1/A1 in the axis of ord inate) as a function of the memb rane Peclet nu mber, at different values of the in let β1o mass transfer coefficient. As can be seen there exists an essential diffe rence between the two models. This difference can be very large with increasing value of the Peclet number (note here PeL1=Pe). The difference between the two models can only be neglected at low values of Pe-nu mber. That means that the Model A1 can be applied in very limited cases, only. Figure 7. Concentration distribution in the membrane layer at different values of the reaction modulus, Φ, applying the Fickian diffusion flow in the boundary layers (model A2, dC/dY=0 at Y=1); paramet ers as in Fig. 2 The concentration was predicted by Eqs. (26) and (27) for Model A and Eq. (38) for Model B. The C*L1 value was calculated by Eq. (33) for Eq. (38) replacing CoL1 by it. Here the diffusive flow in the catalytic memb rane layer is equal to zero, thus is can not have any influence on the concentration distribution in the catalytic memb rane layer. Against that there is difference between the models. The Model B gives somewhat higher memb rane concentration. With increasing value of the Φ reaction modulus, the concentration decreases, and due to it, the difference between the models also decreases. It can be stated the difference between the models is much less than that in the case when dC/dY>0 at Y=1. On the other hand, the average value of the membrane concentration can be higher here comparing that to the case of the models with dC/d Y>0 at Y=1. Accordingly the reaction rate can also be higher in this case. This can be important especially when the reaction rate constant is low, e.g. in the cases of bioreactions. International Journal of Composite M aterials 2012, 2(5): 79-91 87 The overall mass transfer rates can also differ fro m each other (not shown here). velocity. Accordingly, it can be predicted by the models presented which one is more advantageous to application for a given reaction system. Figure 8. Concentration distribution in the membrane layer at different values of the reaction modulus, Φ, applying the exact diffusion flow in the boundary layers (Model B2, dC/dY>0 at Y=1); paramet ers as in Fig. 2 3.3. Comparison of Operating Modes by the Exact Model Figure 9. The ratio of the mass transfer rates with different operation modes applying the exact solution (JModel B1/JModel B2) as a function of the reaction modulus at different values of Peclet number ( β1o = βo = 1x10−4 m/s; βo2 → ∞ ; Pe=1; PeL1=1; H=1; CoL2 = 0 ) The presence of a sweep phase can strongly affect the outlet concentration and thus, the diffusive outlet flow on the permeate side of the catalytic membrane layer. Accordingly, the sum of the diffusive and the convective flows should depend on the operating conditions, namely there is a sweep phase (dC/d Y>0 at Y=1) or there is not a sweep phase (dC/d Y=0 at Y=1). Thus, it seems to be interesting to investigate how the ratio of the mass transfer rates can depend on the values of Φ and Pe (Fig. 9) applying the exact model, na mely Model B. As can be seen the ratio of the mass transfer rates, namely JModel B1/JModel B2 can essentially differ fro m unity. Th is ratio gradually tends to unit with increasing reaction rate due to the lo w concentration values of reactant in the membrane layer. Otherwise, it strongly depends on Peclet number at lower values of reaction modulus. This dependence is decreasing with increasing convective 3.4. Case Study The catalytic membrane reactor in a flo w-through mode also appears to be a promising process for industrial application[2]. A special case will be shown to demonstrate the role of the convective velocity in the membrane reactors. In this examp le, the memb rane operates in dead-end mode and no separation procedure is performed. The task of the memb rane is to provide for intensive contact between reactant and catalyst, combined with a short contact time and a narrow residence time distribution. Ilinitch et al.[11] have measured the reduction of aqueous nitrates using mono- and bimetallic pallad iu m-copper catalysts impregnated in γ-Al2O3 support layers. The metal content was kept between 1.7 and around 7 w% with a particle size belo w 3-5 n m. The memb rane layer is p laced in a tank with stirring to circulate all feed solution through this membrane. The concentration of the reactant passing the membrane can be much lower than that in the feed phase. This value depends on the convective stream and the chemical reaction rate. The authors measured the nitrate conversion in three different modes, i.e. without convective flo w and for two different values of convective flow with a surface velocity of 7 x 10-6 and 16.5 x 10-6 m/s (see Figure 12 in Ilinitch et al.[11]). For evaluation of the experimental results, the model A has been used with βo2 = 0 . It is easy to see that the outlet concentration is equal to the “bulk” concentration behind the memb rane layer, as illustrated in Fig. 1b. The concentration change in the circulated reaction solution can be given as follows[CL1 represents the bulk concentration denoted by C1o in Eqs. (19) to (20)]: −V dCL1 dt = R 2π (J − J out ) (43) ( ) ( ) ≡ R2πβ o T λ 1− eλ + Sλ 1− eλ The value of CL1 denotes as the bulk concentration in the stirred tank. Its value changes as a function of time. The values of T and S parameters are given in Eqs. (19) and (20). Indicating the initial concentration of the reaction solution as CLo and solving Eq. (43), the following time function of concentration is obtained: CL = CLoe−θt (44) with ( ) ( ) θ = R~ 2πβo Tλ 1− eλ~ + Sλ~ 1− eλ V The data used for calculation are listed in Table 1. Taking the diffusion stream through the membrane into account, the value of tD represents the residence time of the reaction solution as given by the following equation: tD = Vδ R 2πD (45) 88 Endre Nagy: Diffusive Plus Convective M ass Transport Through Catalytic M embrane Layer with Dispersed Nanometer-Sized Catalyst Performing the calculation using data from Table 1, the value of tD obtained was 400 min. Table 1. Values of parameters used for calculation of the concentration change in the membrane layer[11,15] V= 85 x 10-6 m3 δ=4,6 x 10-3m D= 1 x 10-8 m2/s R~ =2.25 x 10-2 m βo = 2 x10−6 m / s υ1=7 x 10-6 m/s υ2=16.5 x 10-6 m/s R2π βo =3.2 x 10-9 m3/s P e1 =3 .5 P e2 =8 .3 A mathematical model has been developed in order to predict mass transport through a catalytic membrane reactor containing dispersed nanometer-sized catalyst particles using forced convective flow through the membrane layer, as well as for the case where the nanometer-sized catalyt ic particles are regularly dispersed in the membrane layer or the memb rane intrinsically catalytic. Transport models developed include the mass transport into and inside the catalytic partic les as well as through the catalytic me mbrane layer, taking into account both convective and diffusive flows. It has been shown that the two operating modes, namely with or without sweep phase on the permeate side can give essentially different in let mass transfer rates. On the other hand, the application of the Fickian d iffusive flo w in the feed boundary condition in presence of convective flow can lead significant error in the mass transfer rate predicted. This error quickly increases with increasing boundary layer’s Peclet nu mber due to the increasing curvature of the concentration distribution. Accordingly the application of the mass transfer coefficient predicted by the dimensionless number fo r the feed boundary layer should be avoided in the presence of forced convective flow. The models presented describe mass transport in good agreement with the measured data, proving that it can be used to estimate the mass transport process for a catalytic memb rane reactor. Figure 10. Performance of catalytic membrane reactor situated in a perfectly mixed tank at different values of membrane Peclet number (points are measured data[11], lines are the predicted one) The concentration of the reactant solution obtained was about 0.4 for the case of the diffusion driven mode (Pe=0) with Ф=1, whereas the value is around 0.08 for the case of the convective flow mode with Pe=10 at t / tD = 3. These data are in line with the measured values, as can be seen in Fig. 10. The points represent the measured data whilst the continuous lines indicate the calculated values. It should be noted that the overall first-order kinetics was assumed for the nitrate-ion in the calculat ion. The H2 concentration was kept constant during the reaction. The Ф values should be estimated for calcu lation of the conversion vs. time function. The value was obtained by fitting the measured conversion data fro m diffusion-driven flo w (Pe=0). The continuous line for Pe=0 in Figure 10 was obtained using an estimated value of Ф=1.8. This value was then used for calculation of the curves for Pe=3.5 and 8. The continuous lines obtained by simu lation are plotted together with the measured points. The calculated data for Pe=3.5 are slightly lo wer than the measured values, whereas the data obtained for Pe=8 are in surprisingly good agreement with the measured points. The good agreement between the measured and the calculated data proves that the model developed is suitable for estimating mass transport and conversion in the presence of both convective and diffusive flows. 4. Conclusions ACKNOWLEDGEMENTS This work was supported by the National Develop ment Agency grant (TÁMOP – 4.2.2/B-10/1-2010-0025) Nome nclat ure C = concentration in the membrane, mol/m3 Ci= concentration in the boundary layer, mol/m3 (i=1,2) Co= bulk concentration at t=0, mol/m3 C∗p = concentration on the catalyst interface, mol/m3 dp= particle size, m D= diffusion coefficient in the membrane matrix, m2/s h= distance between particles, m H = solubility coefficient of reactant between polymer matrix and the continuous phase, Hd = solubility coefficient between catalytic particles and the membrane phase, Hf = adsorption coefficient on the catalyst surface (=q/C), (mol/m2) /(mol /m3) Jo = mass transfer rate without chemical reaction, mol/(m2s) J = mass transfer rate into the catalytic membrane layer in presence of chemical reaction with constant diffusive flow in the boundary layers, mol/(m2s) Jout = outlet mass transfer rate, mol/(m2s) J◊ = mass transfer rate obtained by variable diffusive flow, mol/(m2s) j = mass transfer rate into catalytic particles, mol/(m2s) k = reaction rate constant, 1/s Pe = Peclet number or membrane Peclet number[Eq. (11)], q = molar loading on the catalyst surface, mol/m2

... pages left unread,continue reading

Document pages: 13 pages

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×