Separation and Purication Technology 41 (2005) 275287

Design and simulation of two phase hollow ber contactors for simultaneous membrane based solvent extraction and stripping of organic acids and bases
Rudolf Kert esz, Stefan Schlosser
Department of Chemical and Biochemical Engineering, Slovak University of Technology, Radlinsk eho 9, 81237 Bratislava, Slovakia Received 20 October 2003; received in revised form 12 August 2004; accepted 10 September 2004

Abstract A short-cut method for the design and simulation of two-phase hollow ber (HF) contactors with variable mass-transfer and distribution coefcients taking into account reaction kinetics is presented. A parametric study of membrane based solvent extraction (MBSE) and stripping (MBSS) of butyric acid (BA) based on experimental data is given. For the systems with concentration dependent mass-transfer and distribution coefcients a method dividing the length of bers into segments with the same inlet to outlet concentration difference in the feed phase is suggested. The mean values of mass-transfer and distribution coefcients are estimated for each segment independently. For the tested system with BA, convergence of the simulation results was achieved for more than 20 segments. Dependences of the total number of contactors (length of bers) on the shell Reynolds number and approach to the equilibrium at the rafnate end of the HF contactor exhibit a minimum at about 0.9 and 0.40, respectively. The overall number of contactors in MBSE and MBSS is very sensitive to the increase in BA yield in MBSE above 0.85. For following process parameters: volumetric ow rate 1 m3 h1 , feed concentration 350 mol m3 , yield of BA in MBSE 85%, Reshell = 0.9, approach to equilibrium at the rafnate end of the HF contactor 0.40, and concentration factor of 10, the number of contactors of Liqui-Cel type 10 in. 28 in. (Celgard) in series, without parallel lines of contactors, is ve in MBSE and eight in stripping. In case when two parallel lines of contactors are used the number of contactors in series is two in extraction and four in stripping. 2004 Elsevier B.V. All rights reserved.
Keywords: Design; Simulation; Hollow ber contactors; Membrane based solvent extraction; Membrane based solvent stripping; Butyric acid; Organic acids

1. Introduction Hollow ber (HF) modules are widely used nearly in all membrane processes, especially in ultraltration, microltration, gas permeation and pertraction through polymeric or liquid membranes. Recently, hollow ber contactors have begun

Abbreviations: BA, butyric acid; HF, hollow ber; MBSE, membrane based solvent extraction; MBSS, membrane based solvent stripping; P, extractant; TOA, trioctylamine Presented at the Membrane Science and Technology Conference PERMEA 2003, Tatransk e Matliare, Slovakia, September 711, 2003 (http://www.sschi.chtf.stuba.sk/permea/). Corresponding author. Tel.: +421 2 524 96743; fax: +421 2 524 96920. Schlosser). E-mail address: stefan.schlosser@stuba.sk (S. 1383-5866/$ see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2004.09.007

to be used in membrane based processes as membrane based absorption or desorption and membrane based solvent extraction (MBSE) or membrane based solvent stripping (MBSS) in two phase contactors [15]. Mass-transfer between two immiscible liquidssolvent extractionis in the classical arrangement realized in contactors with dispersed phase (spray or mixed columns, mixersettlers, etc.) or by combination of the dispersed phase and a liquid lm (packed columns). A new approach in contacting two uids is the formation of an immobilized L/L interface at the mouths of pores of the microporous wall, through which mass-transfer takes place, as it is schematically shown in Fig. 1. This approach is sometimes called nondispersive solvent extraction or sorption. Despite the fact that the microporous wall, membrane, does not play an active role

276

Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

Nomenclature ai effective interfacial area on the inner surface of the hydrophobic bers per unit length of the contactor, Eq. (26) (m) geometric surface area (m2 ) arithmetic mean value of the inner and outer geometric surface areas of bers (m2 ) molar concentration of the solute (undissociated acid) (mol m3 ) analytical (overall) molar concentration of the permeant (acid) (mol m3 ) diameter of the ber (m) hydraulic diameter of the shell (4 crosssectional area/wetted perimeter) (m) inner diameter of the shell (outer diameter of the central tube) (m) outer diameter of the shell (m) diffusion coefcient (m2 s1 ) distribution coefcient individual mass-transfer coefcient (m s1 ) overall mass-transfer coefcient in the extractor (m s1 ) overall mass-transfer coefcient in the stripper (m s1 ) length of the ber (m) effective length of bers in contactor (m) molar ux (mol s1 ) number of parallel contactors number of contactors in series total number of contactors (both in parallel and series) number of bers in contactor number of transfer units rate constant of the extraction reaction, Eq. (6) (m s1 ) rate constant of the stripping reaction, Eq. (7) (m s1 ) overall mass-transfer resistance (s m1 ) Reynolds number, Re = du Reynolds number for radial ow in the shell, Eq. (14) individual mass-transfer resistance based on reaction kinetics (s m1 ) Schmidt number, Sc = cross-section area of ber lumens (m2 ) Sherwood number, Eq. (12) linear velocity of the ow (m s1 ) mean linear velocity of radial ow in the shell, dened by Eq. (16) (m s1 ) volumetric ow rate (m3 s1 )

A Aw c ca d dh dki dko D k Ke Ks L Lc,eff n Ncp Ncs Nct Nf NTU re rs R Re Reshell Rk Sc Sf Sh u u S V

YMA/OA ratio of ux of mineral acid to organic acid, dened by Eq. (47) Z concentration factor of the solute in the concentrate (output from the stripper), dened by Eq. (37) Greek letters n+1 concentration ratio dened by Eq. (42) w wall thickness (m) porosity of the wall conv conversion of the reagent in the stripping solution e yield of the solute in MBSE dened by Eq. (30) kinematic viscosity (m2 s1 ) tortuosity of the pores RT residence time in the shell of contactor (s) Subscripts b boundary layer in the bulk phase e extractor (MBSE) F feed phase, feed boundary layer i inner surface of the ber wall loc local value ls logarithmic mean value MA mineral acid n number of the contactor segment o outside surface of the ber wall OA organic acid R stripping solution; stripping interface s stripper (MBSS) S solvent phase; boundary layer in the solvent TA total acid (organic plus mineral acid) w ber wall 0 initial value 1 feed or stripping solution inlet end of a HF contactor or a series of contactors 2 rafnate or stripping solution outlet end of a HF contactor or a series of contactors

in separation of the given mixture, such systems and related contactors are often termed with the adjective membrane. Processes with mass-transfer between two liquids will be termed as membrane based solvent extraction and membrane based solvent stripping, as suggested earlier [2,5,6]. The potential for application of MBSE is high and covers recovery or removal of organics and metals [2,3,5,6] or formation of integrated and hybrid systems with MBSE [5,7,8]. The technological scheme of MBSE combined with simultaneous regeneration of the solvent by MBSS, both in HF contactors, and the recirculation of solvent to extraction are shown in Fig. 2.

 Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

277

The aim of this paper is to present a short-cut method for design and simulation of two phase HF contactors with a concentration dependent mass-transfer and distribution coefcient in the simultaneous MBSE and MBSS with closed circuit of the solvent. A case study for the design of HF contactors in the recovery of butyric acid (BA) by simultaneous MBSE and MBSS will be presented, as well.

2. Theoretical 2.1. Mass-transfer in HF contactors The schemes of two-phase systems and related concentration proles of solute in a HF contactor in MBSE and MBSS with aqueous feed and stripping solutions owing in the hydrophobic ber lumen are schematically shown in Fig. 1. The overall reaction between molecules of organic acid (HA) and extractant (P) on the feed/solvent interface can be expressed as (HA)p P pHA + P (1)

Fig. 1. Scheme of the two-phase system in the membrane based solvent extraction (MBSE) and in the membrane based solvent stripping (MBSS) with immobilized L/L interface in hydrophobic hollow bers with concentration proles of the transported acid or base. Ffeed phase, Ssolvent, Rstripping solution.

The types of two phase HF contactors with a microporous wall and their advantages and shortcomings are discussed in references [2,3,5,6,9]. Mass-transfer characteristics of HF contactors are presented in paper [8,10]. Analysis of mass-transfer resistances in HF contactors, including the inuence of the reaction kinetics on them, is shown in papers [1012]. It has been found that mass-transfer resistance related to slower kinetics of the decomposition of the solute complex contributes substantially to the overall resistance in some systems [10,11,1315], e.g., in MBSS of butyric acid (BA) [10,13,16] or heterocyclic carboxylic acid [12,17,18]. In latter system also kinetics of the complex formation in MBSE contributes to the overall resistance. Simulation of integrated extraction and stripping using a mathematical model with constant distribution and overall mass-transfer coefcients for valeric acid is presented in paper [19], for phenol, lactic acid and phenylglycine in paper [20] and for butyric acid and heterocyclic carboxylic acid in paper [21].

and for the acid/extractant complex decomposition on the stripping interface as + pA + pH2 O (HA)p P + pOH P (2)

where species in the organic phase are marked with an overbar. A more detail discussion on extraction equilibrium of carboxylic acids with solvents containing trioctylamine as extractant and the related mechanism are given in papers [22,23]. The extraction mechanism consists of a physical extraction by the diluent (n-alkanes) in form of a monomer and a dimer of the solute (BA) in the organic phase and simultaneous reactive extraction of acid with the extractant (in this paper tertiary amine) when carboxylic acid is extracted in form of complexes with the extractant. In extraction of BA with trioctylamine (TOA) formation of complexes with 17 molecules of acid per one amine molecule is assumed [22]. The proposed model for BA ts experimental equilibrium data well [22]. Only undissociated acid molecules are extracted with tertiary amines. pH of the treated solution has to be adjusted well below the pKa value. Let us consider steady state or quasi-steady state conditions in MBSE. Thus, the solute ux from the aqueous phase to the bulk organic phase in the extraction, n e , can be expressed by equation n e = Ke Aei e (cF cFS,S ) = Ke Aei e cF cS DS (3)

Fig. 2. Flow sheet of MBSE with simultaneous regeneration of solvent by MBSS in a HF contactors and recirculation of solvent to extraction. In both contactors the solvent ows in the contactor shell.

where DS is the distribution coefcient related to the concentration of solute in the solvent, cS . Eq. (4) denes the overall mass-transfer coefcient in MBSE, Ke , for the effective interfacial area. cFS,S is the equilibrium concentration of the solute (undissociated acid) in the feed phase at the interface corresponding to its concentration in the bulk of the solvent,

278

Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S. Table 1 Lumped rate constant of the stripping reaction, Eq. (7), in MBSS and pertraction (PT) of butyric acid [8,25] Process Solvent 0.4 kmol m3 TOA in n-alkanes rs 106 m s1 4.84 5.58 8.35 Pure n-alkanes 29.2

for which holds cS = DS cFS,S (4)

The overall solute ux in stripping from the organic phase (solvent) to the aqueous stripping solution, n s , can be described by the following relation with simplication for cases, when concentration of the undissociated acid in the stripping solution is zero. This holds in cases with an excess of reagent, when pHR is well above the pKa value (when cR = 0), and the simplied form of Eq. (5) can be used n s = Ks Asi s (cS cR DR ) = Ks Asi s cS (5)

MBSS in HF contactor PT in HF contactor PT through layered BLM (from data in [13]) PT in HF contactor

n s = rs Asi s cSR

(7)

The mode of operation of the stripper with an aqueous stripping solution owing in the lumen of hydrophobic bers will be considered. The high stoichiometric excess of reactant (e.g. NaOH for acids) in the stripping solution will result in a zero concentration of undissociated acid in the stripping solution and consequently in a zero resistance in this phase (RR = 0). 2.1.1. Diffusion-reaction model Resistances in series approach, assuming only diffusion resistances in boundary layers and walls, are usually used [2,3]. The discrepancy between the experimental values of the overall mass-transfer coefcient in pertraction [11,14,15,24,25], and MBSS [11,14] of carboxylic acids, as well as in both MBSE and MBSS of butyric acid (BA) [10,13,16] or heterocyclic carboxylic acid [12,17,18] and the values calculated according to the diffusion model can be interpreted by a slower reaction of the complex decomposition on the stripping interface or complex formation on the extraction interface. For lactic acid the stripping is faster than extraction [26] and its kinetics will not inuence the masstransfer rate. However the kinetics of formation of the complex may play a role. The inuence of reaction kinetics was found also for metals [27,28]. The kinetics of competitive adsorption and desorption of complexes and free extractant molecules, together with the maximum value of the concentration of these molecules on the interface (independent of concentration in the bulk phase) can be important. Eqs. (1) and (2) can describe the overall reactions for the acid/carrier complex formation and decomposition on the stripping interface. For example in pertraction or MBSE of BA with a solvent containing trioctylamine (TOA) nine acid species are in the solvent including also the monomer and the dimer of BA. The mole fractions of individual complexes are concentration dependent. If two or more surface active substances are present in the solution, e.g., free carrier and acid-carrier complexes, the situation is more complex and could not be easily described, as discussed in papers [29,30]. As the rst approximation, the following simple rst order rate equations for the description of the kinetics of extraction and stripping reactions will be used [11,12,15,18] n e = re Aei e cFS (6)

where cFS and cSR are concentrations of acid and/or acid in form of complexes in the solvent close to the extraction or stripping interface, and re and rs are, de facto, lumped parameters reecting the kinetics of the complex formation or decomposition interfacial reactions, equilibrium and kinetics of desorption or adsorption of molecules of the complexes on the stripping interface, and desorption of free extractant molecules from the stripping interface, and cSR is the complex concentration in the solvent close to the stripping interface. The ratio of the molecules of complexes to free extractant molecules on the interface can possibly be concentration dependent what will be also reected in the stripping kinetics. For example in pertraction of butyric acid with TOA, the mean molar ratio of acid to extractant molecules in the complex is about 3, as shown in paper [22]. The values of the rate constant of the stripping reactions for butyric acid are given in Table 1 and for other organic acids in papers [8,10]. From this table and paper [25] it is evident that the value of the rate constant in stripping free butyric acid from pure n-alkanes is much higher than in case when the acid/TOA complexes are dominant in solvents containing TOA. Thus, the value of the rate constant will depend on the solvent composition. Let us consider a system with aqueous phase in the ber lumen and organic phase on the shell side, and a hydrophobic HF. For the steady state or quasi-steady state conditions, when uxes in individual liquid layers are equal, the following relations can be derived for the overall mass-transfer resistance in MBSE, taking into account diffusion resistances in boundary layers and walls and the resistance connected with kinetics of the reaction of formation of the permeantextractant complex 1 e Aei Aei e 1 = + + + Ke kF Aw,e kSw DS Aeo kSb DS re (8)

where the overall mass-transfer resistance is composed of four individual resistances Re = RF + RSw + RSb + Rk,e (9)

In case of the inuence of the stripping reaction kinetics on the overall resistance and for situations with a zero resistance in the stripping solution boundary layer following equations can be derived for stripping Rs = RSb + RSw + Rk,s (10)

 Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

279

1 Asi s Asi 1 = + + Ks Aso kSb Aw,s kSw rs

(11)

of 40% from the interval 2.3 to 2.8 depending on the system. In this work the value of tortuosity of 2.7 will be used. 2.2. Calculation of the length of bers in HF contactors In steady state operation of a HF contactor the length of bers can be calculated as a product of the number of transfer units (NTU ) and the length of the transfer unit (LTU ) based on the overall mass-transfer coefcient. L = NTU LTU (18)

Eqs. (8)(11) represent the diffusion-reaction model of MBSE and MBSS for the mode of operation when the solvent ows in the shell. The last terms in these equations represent resistances based on the kinetics of the complex formation and stripping reactions. By omitting the last term in Eq. (8) or (11) the classical diffusion model of MBSE and MBSS is obtained. 2.1.2. Estimation of individual mass-transfer coefcients Identication of individual mass-transfer coefcients depends on the construction of the module. In order to characterize the local mass-transfer coefcients several correlations have been proposed. For the laminar ow in tubes, correlation of Dahuron and Cussler [31] is often used: Sh = d2u kdi = 1.5 i L
1/3

Let us consider contactors with hydrophobic bers with a parallel counter-current ow of phases and an aqueous feed ow in the ber lumen, which is schematically drawn in Fig. 3 [2,5]. A constant volumetric ow rate of phases along the contactor will be assumed. For the differential material balance of the contactor holds F dcF = Ke,loc (di N) cF cS V DS dz (19)

(12)

For mass-transfer in the shell of cross ow modules, the following correlation suggested by Sch oner [32] was applied
0.33 Sh = 1.76Re0.82 , shell Sc

0.012 Re 2

(13)

where Reynolds number in the shell is dened by relation Reshell = S dh u (14)

and the hydraulic diameter and the mean velocity of liquid in the shell by equations dh =
2 nd 2 d 2 dki 4Vshell o = ko Ao ndo

where Ds is the distribution coefcient related to cs . Generally, the overall mass-transfer coefcient, the distribution coefcient and also the volumetric ow rate change along the contactor. To simplify this situation the contactor will be divided into n segments with the same inlet to outlet concentration difference in the feed phase with a constant value of the mentioned quantities in the segment dened for the mean concentration in the segment. By integrating Eq. (19) along the length of the nth segment of the contactor one obtains for boundary conditions z = 0, cF = cF,n , cS = cS,n , and for z = Ln , cF = cF,n , cS = cS,n+1
Ln 0

(15)

S S ln(dko /dki ) 2V 2V u S = = leff dshell,ls leff dko dki

Ke,loc dz = Ke L =

F V ai

cF,n+1 cF,n

dc F (cF (cS /DS ))

(20)

(16)

Other available equations for the estimation of individual mass-transfer coefcients are presented in papers [2,3,5]. The individual mass-transfer coefcient in the stagnant liquid lling the pores of the wall can be calculated by equation kw = 2 = w (do di ) (17)

From the overall material balance of solute in the contactor, from z = 0 up to the coordinate z = L with the respective

The diffusion coefcient was calculated by WilkeChang equation [33]. For example the diffusion coefcient of BA in water at 30 C is 11.01 1010 m2 s1 . The value of the diffusion coefcient of the BA complex with (BA)3 TOA structure in the solvent (0.4 kmol m3 TOA in n-alkanes) is 3.55 1010 m2 s1 . The values of tortuosity reported in papers dealing with solvent extraction with the polypropylene hollow bers Celgard X-10 with a porosity of 30% are ranging from 2 [34] to 3.5 [35]. Prasad and Sirkar [1] suggested the value of tortuosity of the polypropylene bers Celgard X-20 with a porosity

Fig. 3. Scheme of the two phase HF contactor and its nth segment with parallel ow of phases in MBSE.

280

Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

concentrations in aqueous and solvent phases cF and cs follows F (cF1 cF,n+1 ) = V S (cS1 cS,n+1 ) V (21)

By inserting cS from this equation into Eq. (20) and after integrating for a constant value of the overall mass-transfer coefcient and the distribution coefcient one obtains after rearrangement Le,n = F,n V 1 cF,n cS,n /DS ln F,n /DS V S,n K e ai 1 V cF,n+1 cS,n+1 /DS (22)

where i,n = Vi,n + Vi,n+1 V 2 (23)

Expressions for NTUs,n and LTUs,n for different systems and congurations are presented in papers [2,3,5]. The number of the stripping contactors in series can be calculated similarly as for extraction, see Eq. (27). More complex is the situation when the distribution coefcient and/or the volumetric ow rate of phases are concentration dependent. A more rigorous approach requires the formulation of appropriate differential equations for mass-transfer in individual layers, which have to be solved numerically. Relations for the distribution coefcient and/or volumetric ow rate as functions of concentration should be included. This approach was used, e.g., in papers [3640]. A model of MBSE for multicomponent systems is proposed in paper [41]. In separation of charged species, Fickian diffusion approach is not satisfactory and NernstPlanck equation has to be considered [42]. Modeling of HF contactors with cross ow of phases is presented in paper [32].

Considering Eq. (18) for the number of transfer units the following relation is obtained NTUe,n = 1 F,n /DS V S,n 1V ln cF,n cS,n /DS cF,n+1 cS,n+1 /DS 3. Short-cut method for the calculation of two phase HF contactors (24) In calculation of simultaneous MBSE and MBSS of solutes in HF contactors the typical input process data are: - ow rate and composition of the feed; - composition of the solvent with related L/L equilibrium data; - concentration factor of solute in the concentrate; - yield of solute in MBSE; - geometric characteristics of contactor(s) and their masstransfer data for the given system. Further quantities, e.g. those characterizing the hydrodynamic conditions in contactors (as linear velocity of the feed and stripping solution in the ber lumen and Reynolds number in the shell), approach to equilibrium at the rafnate end of the extractor, conversion of the reagent in the stripping solution, co-extraction of other solutes (e.g. mineral acids), etc. can be estimated only from experimental data for the given or similar system in a HF contactor. The basic scheme of the simultaneous MBSE and MBSS circuit with notation of the main process parameters is shown in Fig. 2. The owchart of a short-cut method for calculation of the length of bers and the number of parallel contactors and contactors in series in simultaneous MBSE and MBSS processes is presented in Fig. 4. The basic characteristics of two phase HF contactors produced commercially are presented in Table 2. 3.1. Material balance of MBSE and MBSS circuits Calculations are based on the methodology shown in Section 2.2. The yield of solute in extraction is dened by relation e = F1 cF1 V F,n+1 cF,n+1 V cF1 XcF,n+1 = cF1 VF1 cF1 (30)

and for the length of the transfer unit the following equation holds LTUe,n F,n V = Ke,n ai (25)

where Ke,n is the mean (integral) value of the overall masstransfer coefcient in the nth segment of the HF extractor and for ai ai = di Nf (26)

ai is an effective interfacial area on the inner surface of hydrophobic bers per unit length of the contactor. The number of extraction contactors in series can be calculated by relation Ncs,e,n = Le,n Lc,eff (27)

where Lc,eff is the effective length of bers in the selected type of the HF contactor. For membrane based stripping of the solvent (MBSS) an excess of reagent in the stripping solution is maintained resulting in a zero concentration of undissociated molecules of weak acids or bases in it or zero concentration of the complex on the strip interface. Under this assumption, equations for NTUs,n and LTUs,n can be derived for stripping of the solvent owing in the shell NTUs,n = ln LTUs,n = cS,n cS,n+1 (28)

S,n V Ks,n ai

(29)

 Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

281

Fig. 4. Flowchart of the short-cut calculation of the ber length and the number of HF contactors for simultaneous MBSE and MBSS.

For systems with a variable volume of phases, due to the mass-transfer of solute, the quantity X is dened by equation X= F,n+1 V k Mk cF1 = F1 k Mk cF,n+1 V (31)

F,n+1 = VF,n (OA cF,n MOA e ) V OA cF,n+1 = F,n cF,n (1 e ) V F,n+1 V

(34)

(35)

where k is the density of the pure acid (solute). Additivity of volumes is supposed. This was experimentally proved for BA, DMCCA and MPCA. For the ow rate of rafnate and the concentration of solute in it the following equations hold: F,n+1 = V F,n V OA,n V (32)

From the overall material balance the ow rate of the con R,n+1 can be estimated centrate V R,n+1 = VF,n cF,n VF,n+1 cF,n+1 V cR,n+1 = F,n+1 cF,n+1 F,n cF,n V V ZcF,n (36)

F,n cF,n V F,n+1 cF,n+1 ) OA,F,n MOA MOA (V OA,n = n V = OA OA (33)

282

Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

Table 2 Characteristics of HF contactors Liqui-Cel (Celgard) used in simulations Contactor Number of bers (cca.) Internal area of bers (m2 ) Effective length of bers (cm) Shell volume (cm3 ) Cross-section area of bers lumen (cm2 ) ai (m2 /m) Logarithmic mean shell diameter (cm) Hydraulic diameter in the shell (mm) Characteristics of bers External diameter (m) Internal diameter (m) Wall thickness (m) Porosity (%) Mean pore size (m) Tortuosity 2.5 in. 8 in. 10000 1.13 15.0 195 4.52 2.82 3.28 0.260 300 240 30 40 0.03 2.7 4 in. 28 in. 32000 15.32 63.5 4200 14.48 9.07 5.29 0.711 10 in. 28 in. 230000 108.38 62.5 26100 104.05 65.04 14.83 0.546

40

30

where the concentration cR,n+1 is dened, e.g., by requirement of the concentration factor of solute in the concentrate (output from the stripper), which is dened by relation Z= cR,n+1 cF1 (37)

For the value of the concentration of solute in the solvent leaving the extractor, cS,n , and from the material balance relation follows F,n cF,n V F,n+1 cF,n+1 + V S,n+1 cS,n+1 V cS,n = (44) VS,n The volumetric ow rate of the stripping solution can be estimated from relation R0 = V R2 V OA V H2 O,R V (45)

The volumetric ow rate of the solvent in the shell is dened by requirement of proper hydrodynamic conditions, which are characterized by the value of the Reynolds number, dened for example by Eq. (14). On other hand, also dimensions of the stripper depend on the value of Reshell (because the solvent ows in the shell), what follows from Eq. (29). S, The mean value of the volumetric ow rate in the shell, V is dened for the mean velocity of the radial ow in the shell, u s given by Eq. (16). Then, Reshell S dshell,ls,e leff,e Ncp,e S = V 2dh,e S dshell,ls = dko dki ln (dko /dki ) (38) (39)

where the ow rate of water formed in the neutralisation reaction of acids (both organic and mineral) can be expressed by equation TA,S,n MH2 O H2 O,R,n = n V H2 O = S,n cS,n V S,n+1 cS,n+1 )(1 + YMA/OA ) MH2 O (V H2 O (46)

S1 a V S,n+1 following equaFor the volumetric ow rates V tions can be derived S1 = 2V S + VOA V (40) 2 S,n+1 = V S,n V OA,n = 2V S VOA V (41) 2 The approach to equilibrium in MBSE is dened by equation cS,n+1 n+1 = (42) cS,n+1 This allows the calculation of the concentration of solute in the solvent entering the extractor from its concentration in the feed phase using equation cS,n+1 = cF,n+1 DF,n+1 n+1 (43)

The amount of the transported mineral acid has to be estimated from experimental data, e.g. from the ratio YMA/OA = n MA n OA (47)

In MBSE of BA from aqueous solutions with a low content of mineral salts this ratio can be equal to zero. For MBSE of MPCA from sulphate solutions at pH above 2 the value of YMA/OA is about 1 [11,18]. The initial concentration of reagent in the stripping solution can be estimated from equation creag,R0 = F,n+1 cF,n+1 F,n cF,n V F,n cF,n e V V = R0 conv VR0 conv V (48)

where conv is the conversion of reagent in the stripping solution, whose value has to be kept less than 1 (usually at about 0.7) to guarantee an excess of reagent at the stripping interface.

 Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

283

3.2. Design of HF contactors for MBSE and MBSS 3.2.1. MBSE The number of parallel contactors in extraction is calculated under the consideration of optimal velocity of the feed in the ber lumens. The number of parallel contactors calculated by equation Ncp,e = F1 V u F Sf (49)

4. MBSE and MBSS of butyric acid (BA)a case study 4.1. Mass-transfer data MBSE and MBSS of butyric acid (BA) with TOA as an extractant were studied in papers [10,14,16]. This system and related data will be used in the presented case study. The system studied in these papers was follows: Feed:Aqueous solutions of butyric acid, cF0 = 0.48 1.2 kmol m3 , pHF0 = 2.42.6 (without adjustment of pH with mineral acid). Solvent: 0.4 kmol m3 of trioctylamine (TOA) in nalkanes (dodecane fraction). The density and kinematic viscosity of the solvent at 30 C were 751.9 kg m3 and 2.026 106 m2 s1 , respectively. Stripping solution: Aqueous solution of NaOH with a concentration of 2.53.0 kmol m3 . Temperature: 30 C. For simultaneous extraction and stripping, two hollow ber contactors Liqui-Cel Extra-Flow 2.5 in. 8 in. (Hoechst-Celanese, USA) with a cross-ow of phases were used. Contactors were connected to the reservoirs of three phases in the recirculation mode, as described in papers [12,16,18] together with operation modes and analytical methods used. Celgard X-30 microporous polypropylene hollow-bers were assembled in these contactors. The characteristics of the contactor and bers are presented in Table 2. The stripping solution owed through the bers lumen and the solvent in the shell. The overall mass-transfer coefcients in MBSE and MBSS were evaluated from the concentration of acid in the feed, solvent and strip reservoirs and in the outputs from contactors, considering variable distribution coefcients and experimentally determined driving forces. The value of the distribution coefcient of BA in the solvent with TOA is concentration dependent, as can be seen in Fig. 5. This has to be taken into account in modeling and calculations. The proposed model related to the mechanism of extraction suggested in paper [22] ts experimental data well. For purposes of the short-cut design following empirical

should be rounded off to an integer. Thus, the number of parallel contactors has to be an integer and the feed velocity has to be from an optimal interval, which should be estimated experimentally. The length of bers is estimated from Eq. (22), which was derived for constant mass-transfer and distribution coefcients in the contactor. When this is not the case, the contactor is divided into segments with the same concentration difference in the feed phase, where constant values of the mentioned quantities can be supposed. The lengths of such segments are generally unequal. The value of the overall mass-transfer coefcient is estimated from Eq. (8), or another equation relevant for the given system. The mean concentration in this segment, usually a logarithmic mean value, is for the nth segment with inlet and outlet concentrations in this segment cF,n and cF,n+1 dened by relation cFls,n = cF,n cF,n+1 ln(cF,n /cF,n+1 ) (50)

For the estimation of the individual mass-transfer coefcient in a ber lumen by Eq. (12) a guess of the ber length is used. The calculated length is applied in the next iteration till a satisfactory agreement is reached. In systems with the concentration dependent distribution coefcient equation(s) describing this dependence should be included into the calculation procedure for estimation of the overall mass-transfer coefcient. For a case study with MBSE of BA an appropriate model of L/L equilibrium with a good t to experimental data, as shown in Fig. 6, is published in paper [22]. Another possibility, which usually simplies calculations, is to correlate experimental or simulated equilibrium data with an empirical interpolation polynomial. 3.2.2. MBSS The number of parallel contactors in stripping is estimated S from the mean ow rate of the solvent in the shell V Ncp,s = S dh,s S 2V Reshell S dshel,ls,s leff,s (51)

The length of bers in the stripper is calculated from Eqs. (28) and (29). The total number of contactors for extraction and stripping is given by Nct = Ncs,e Ncp,e + Ncs,s Ncp,s (52)

Fig. 5. Distribution coefcient of BA vs. concentration of BA in the aqueous phase for the solvent with 0.4 kmol m3 of TOA in n-alkanes at 30 C. Experimental equilibrium data from paper [22] are tted according to Eq. (53) with coefcients presented in Table 3 as shown by lines.

284

Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

Table 3 The values of coefcients of the empirical Eq. (53) for correlation of the distribution coefcient of BA based on aqueous (k = F) or solvent phase (k = S) concentrations for the solvent with 0.40 kmol m3 TOA in n-alkanes k F S A 0.1578 0.1578 B 17.69 21.25 C 279.4 74.63 D 2802 169.6 E 10434 227.5 F 20287 176.1 G 21877 77.63 H 12406 18.14 I 2889 1.745

equation was used

2 3 4 5 6 + Dck + Eck + Fck + Gck D = A + Bck + Cck 7 8 +Hck + Ick

(53)

which is more simpler than a mentioned physical model. This empirical equation ts experimental data well, as shown in Fig. 5 by lines. The values of coefcients in Eq. (53) for estimation of the distribution coefcients from the aqueous phase concentration (k = F) or the solvent phase concentrations (k = S) are presented in Table 3. Extraction and stripping were studied at different hydrodynamic conditions. The experimental data from MBSE of BA in a HF contactor show that the value of the overall masstransfer coefcient in MBSE, Ke , depends on concentration (Fig. 6a), what is mostly a result of the concentration dependency of the distribution coefcient of BA. The value of the overall mass-transfer coefcient in MBSS, Ks , practically does not depend on the BA concentration in the solvent (Fig. 6b). With increasing velocity of the feed in the bers lumen the value of Ke , increases and approaches a steady value at a velocity of about 2.5 cm s1 , as shown in papers [10,14,16]. The main diffusion resistance in the solvent is in the wall pores. With increasing volumetric ow rate of the solvent in the extractor shell the value of Ke , increased only slightly and a steady value was approached soon. In case, when the organic phase circulated in the shell of the stripper, the value of Ks was slightly greater than in the mode where the solvent circulated in the stripper through the bers lumen and is practically independent on the solvent ow rate (Fig. 7).

The experimental values of Ke are in a good agreement with the values estimated from the diffusion model (Fig. 6a). This means that kinetics of the formation of BATOA complexes does not contribute to the overall resistance in MBSE. The value of Ks is more or less independent of the BA concentration (Fig. 6b). The values of Ks calculated from the diffusion model are about two times higher than experimental values (Fig. 6b). This difference is connected with the resistance associated with the kinetics of the stripping process, i.e., the chemical reaction of the complex decomposition and other interfacial phenomena. The estimated values of the stripping rate constant are presented in Table 1. The kinetic mass-transfer resistance plays a decisive role in the overall resistance in MBSS of BA with its participation of about 60% in the overall resistance. 4.2. Calculation of HF contactors and process simulations HF contactors 10 in. 28 in. are supposed in simulations with characteristics shown in Table 3. The values of the F1 = 1 m3 h1 , process parameters were: cF1 = 450 mol m3 , V Reshell = 0.9, e = 0.90, n+1 = 0.40, Z = 10, conv = 0.70 if not otherwise stated. In all simulations no parallel modules are required for the assumed feed ow rate. Thus, the total number of contactors is equal to the number of contactors in series. In extraction a counter-current ow of aqueous feed in the hollow bers lumen and of the solvent in shell is supposed. The stripping modules are connected in series on the shell side where the solvent ows co-currently with the stripping solution in the ber lumen.

Fig. 6. Inuence of the concentration of BA in the feed on the overall mass-transfer coefcient in MBSE, (a) and for the concentration of BA in the solvent on the overall mass-transfer coefcient in the membrane based stripping (b). Mode of operation: solvent in the shell in both contactors. cF0 = 0.453 kmol m3 , uF = 1.78 cm s1 , uR = 0.45 cm s1 , S = 391 cm3 min1 . V

Fig. 7. Inuence of the volumetric ow rate of solvent and of the mode of operation of the HF contactor on the overall mass-transfer coefcient of BA. uF = 1.8 cm s1 , cF0 = 0.453 kmol m3 , uR = 0.45 cm s1 .

 Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

285

Fig. 8. Total number of the HF contactors 10 in. 28 in. in MBSE and MBSS of BA, as well as the overall number of contactors vs. the shell side Reynolds number for a different number of contactor segments with the same concentration difference in the feed phase. Values of the process F1 = 1 m3 h1 , e = 0.90, n+1 = 0.40, parameters were: cF1 = 450 mol m3 , V Z = 10, conv = 0.7.

Fig. 10. The value of the overall mass-transfer coefcient in individual segments along MBSE contactors. The values of process parameters F1 = 1 m3 h1 , Reshell = 0.9, Ncp,e = 1, e = 0.90, were: cF1 = 450 mol m3 , V n+1 = 0.40, Z = 10, conv = 0.7.

The assumption of a parallel ow of phases in the extractor, as used in section 0, simplies the mass-transfer analysis. In reality, in the used modules the solvent ows in every half of the contactor shell in a radial direction and at the shell wall and between halves of the contactor it ows in an axial direction and counter-currently to the aqueous phase owing in the ber lumens, as it is between contactors. Assumption of a counter-current parallel ow of phases could be a rst approximation of the real situation in the described contactors for which experimental data and correlations are available. Calculations and simulations were done as described in Sections 3.1 and 3.2 and shown in Fig. 4. 4.3. Discussion of simulation results The results of simulations are presented in Figs. 816. With increasing value of the Reynolds number in the shell the number of contactors in series needed in extraction decreases. However, the number of contactors, required in the

Fig. 11. Inuence of the number of segments, to which was divided the interval from inlet to outlet concentrations of BA in the feed phase (cF1 to cF2 ) in calculations of the total number of HF contactors 10 in. 28 in. in MBSE of BA. The values of process parameters are shown in Fig. 10.

stripping, increases. Thus, the overall number of contactors in MBSE and MBSS exhibits a minimum at about Reshell = 0.9, as shown in Fig. 8. The lengths of the contactor calculated for each segment were more or less similar (slight increase with decreasing the mean concentration in the segment). However, in the last segment they jumped to a much higher value due to a great decrease in the driving force on the rafnate end of

Fig. 9. Inuence of the shell side Reynolds number on the number of contactors 10 in. 28 in. in MBSE estimated for a selected contactor segments (a) and on the value of the overall mass-transfer coefcient in individual segments of MBSE contactors (b). The overall number of contactor segments in the feed phase used in calculations was 40. The values of process parameters are shown in Fig. 8.

Fig. 12. Total number of contactors 10 in. 28 in. in MBSE and MBSS of BA, as well as the overall number of contactors vs. approach to equilibrium on the outlet end of the contactor. The values of process parameters are shown in Fig. 10. The overall number of segments in the feed phase used in calculations was 40.

286

Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

Fig. 13. Total number of contactors 10 in. 28 in. in MBSE and MBSS of BA, as well as the total number of contactors vs. yield of BA in MBSE. The overall number of segments in the feed phase used in calculations was 40. The values of process parameters are shown in Fig. 10.

Fig. 16. Total number of HF contactors 10 in. 28 in. in MBSE and MBSS of BA, as well as the overall number of contactors vs. inlet feed concentration of BA to MBSE. The overall number of the segments in the feed phase used in calculations was 40. The values of process parameters are shown in Fig. 10.

Fig. 14. Inuence of the yield of BA on the total number of contactors 10 in. 28 in. in MBSE estimated in selected individual segments in the feed phase (a) and on the value of the overall mass-transfer coefcient in selected individual segments (b) for the values of process parameters shown in Fig. 10.

the contactor, as shown in Fig. 9a, as well as due to a decrease in the value of the overall mass-transfer coefcient, as shown in Fig. 9b. The estimated length of bers or the total number of HF contactors, Nct is a function of the number of the segments to which the contactor was divided in the calculation, as shown in Fig. 11. The convergence is good for more than 20 segments in the studied system. The dependences of the number

of contactors in MBSE and MBSS are sensitive functions of the approach to equilibrium on the rafnate end of the extractor (dened by ratio in Eq. (42)), as shown in Fig. 12, with the optimal value of n+1 of about 0.4. From simulation presented in Fig. 13 it is evident that the number of contactors in MBSE and MBSS is very sensitive to the increase in BA yield in MBSE above 0.85. A more detailed analysis showed that the largest part of ber length required belongs to the last segment with the lowest mean concentration, what is connected with a decrease of the distribution coefcient of BA and consequently of Ke and with a decrease of the driving force in MBSE at concentrations below 0.18 kmol m3 , as shown in Figs. 14 and 15. The total number of HF contactors 10 in. 28 in. in MBSE and MBSS of BA, as well as the overall number of contactors, monotonously decrease with the increasing inlet feed concentration, as shown in Fig. 16. This is connected with an increase in the value of the distribution coefcient, as well as of the overall mass-transfer coefcient in the extractor. For 85% yield of BA in MBSE, which is from the application point of view reasonable, and for following process F1 = 1 m3 h1 , cF1 = 350 mol m3 , Reshell = 0.9, parameters: V n+1 = 0.40, Z = 10, YMA/OA = 0, and conv = 0.70 the number of contactors of Liqui-Cel type 10 in. 28 in. (Celgard) in series, without parallel lines of contactors, is ve in MBSE and eight in stripping. In case when two parallel lines of contactors are used the number of contactors in series is 4 (3.91) in extraction and 2 in stripping.

5. Conclusions The presented short-cut method for simulation of HF contactors in the circuit with simultaneous MBSE and MBSS is a useful tool for the design of simultaneous MBSE and MBSS systems. In the systems with the concentration dependent mass-transfer and distribution coefcients contactor has to be divided to segments preferably with the same in-

Fig. 15. The value of the inlet/outlet driving force ratio in individual segments along MBSE contactors for selected yields of BA in extraction (a) and dependence of this ratio on the yield of BA in MBSE in selected segments (b).

 Schlosser / Separation and Purication Technology 41 (2005) 275287 R. Kert esz, S.

287

let to outlet concentration difference. For the tested system with butyric acid, convergence of the simulation results was achieved for more than 20 segments. Dependences of the total number of contactors (length of bers) on the shell Reynolds number and approach to the equilibrium at the rafnate end of the MBSE contactor exhibit a minimum at about 0.9 and 0.40, respectively. The number of contactors in MBSE and MBSS is very sensitive to the increase in BA yield in MBSE above 85%.

[15] [16]

[17]

[18]

Acknowledgements Support of the grant from the Slovak Grant Agency No. VEGA 1/9136/02 is acknowledged.

[19] [20] [21] [22]

References
[1] R. Prasad, K.K. Sirkar, AICHE J. 34 (1988) 177. [2] R. Prasad, K.K. Sirkar, Membrane-based solvent extraction, in: W.S.W. Ho, K.K. Sirkar (Eds.), Membrane Handbook, vol. 727, Van Nostrand Reinhold, New York, 1992. [3] B.W. Reed, M.J. Semmens, E.L. Cussler, Membrane contactors, in: R.D. Noble, S.A. Stern (Eds.), Membrane Separation Technology: Principles and Applications, Elsevier Science, Amsterdam, 1995, p. 467. [4] A. Gabelman, S.T. Hwang, J. Membr. Sci. 159 (1999) 61. Schlosser, Membrane based processes with immobilized interface, [5] S. in: K. Bako, L. Gubicza, M. Mulder (Eds.), Integration of Membrane Processes into Bioconversions, vol. 55, Kluwer Academic, New York, 2000. Schlosser, Hollow ber contactors in environmental protection [6] S. and in biotechnology, in: I.K.M. Bodzek (Ed.), Proc. Membrany i procesy membranowe w ochrorone srodowiska, Ustron-Jaszowiec, PL, 2325 October 1997, p. 79. [7] J.L. L opez, S.L. Matson, J. Membr. Sci. 125 (1997) 189. Schlosser, R. Kert [8] S. esz, J. Mart ak, Separat. Purif. Technol., this issue. [9] K.K. Sirkar, Chem. Eng. Commun. 157 (1997) 145. Schlosser, E. Sabolov [10] S. a, Mass-transfer characteristics of hollow ber contactors for pertraction and membrane based extraction of organic acids, in: S. Luque, M.S. Alvarez (Eds.), Proceedings of the Engineering with Membranes, Granada, 46 June 2001, p. 313. Schlosser, E. Sabolov [11] S. a, R. Kert esz, L . Kubi sov a, J. Sep. Sci. 24 (2001) 509. Schlosser, J. Mart [12] L. Kubi sov a, E. Sabolov a, S. ak, R. Kert esz, Desalination 163 (2004) 27. Schlosser, E. Sabolov [13] S. a, Chem. Pap. 53 (1999) 403. Schlosser, Analysis of mass-transfer resistances in [14] E. Sabolov a, S. extraction and pertraction of butyric acid in HF contactors consid ering kinetics of complex decomposition, in: J. Marko s, V. Stefuca (Eds.), Proceedings of the 28th International Conference on SSCHI, [23] [24]

[25] [26]

[27] [28] [29] [30] [31] [32] [33] [34] [35] [36]

[37] [38] [39] [40] [41] [42]

Full texts on CD ROM, Tatransk e Matliare, Slovakia, 2226 May 2001. Schlosser, E. Sabolov S. a, J. Membr. Sci. 210 (2002) 331. Schlosser, Pertraction and membrane based solvent E. Sabolov a, S. extraction of butyric acid in hollow ber contactors, in: Proceedings of the 14th International Congress CHISA 2000, Full text on CD ROM, Praha, CZ, 2831 August 2000, p. 13. Schlosser, L. Kubi E. Sabolov a, S. sov a, Membrane based solvent extraction of heterocyclic carboxylic acid in hollow ber contactors, in: J. Marko s, V. Stefuca (Eds.), Proceedings of the 28th International Conference on SSCHI, Full texts on CD ROM, Tatransk e Matliare, Slovakia, 2226 May 2001. Schlosser, J. Mart L. Kubi sov a, E. Sabolov a, S. ak, R. Kert esz, Desalination 148 (2002) 205. M. Rodriguez, R.M.C. Viegas, S. Luque, I.M. Coelhoso, J.P.S.G. Crespo, J.R. Alvarez, J. Membr. Sci. 137 (1997) 45. M.J. Gonz alez-Munoz, S. Luque, J.R. Alvarez, J. Coca, Desalination 163 (2004) 1. Schlosser, E. Mikulov M. Vajda, E. Sabolov a, S. a, Chem. Pap. 57 (2003) 3. Schlosser, J. Mart E. Sabolov a, S. ak, J. Chem. Eng. Data 46 (2001) 735. Schlosser, Chem. Pap. 54 (2000) 413. J. Mart ak, S. Schlosser, Pertraction through liquid and polymeric membranes, S. in: K. Bako, L. Gubicza, M. Mulder (Eds.), Integration of Membrane Processes into Bioconversions, vol. 73, Kluwer Academic, New York, 2000. Schlosser, Separat. Sci. Technol., submitted for pubE. Sabolov a, S. lication. Schlosser, Pertraction and extraction of lactic acid L . Kubi sov a, S. in hollow-bre contactors, in: Proceedings of the 12th International Conference in CHISA 96, Full paper No. P3.83, Praha, Czech Republic, 2530 August 1996, p. 1. R.S. Juang, J.D. Chen, H.C. Huan, J. Membr. Sci. 165 (2000) 59. C.F. Yang, E.L. Cussler, J. Membr. Sci. 166 (2000) 229. T. Kanki, H. Kim, A. Tomita, T. Asano, N. Sano, Sep. Purif. Methods 19 (2000) 93. J. Szymanowski, I. Miesiac, K. Schugerl, A. Sobczynska, Solvent Extr. Ion Exchange 10 (1992) 509. L. Dahuron, E.L. Cussler, AICHE J. 34 (1988) 130. P. Sch oner, P. Plucinski, W. Nitsch, U. Daiminger, Chem. Eng. Sci. 53 (1998) 2319. R.C. Reid, J.M. Prausnitz, T.K. Sherwood, The Properties of Gases and Liquids, McGraw-Hill, New York, 1977. Schlosser, I. Rothova, H. Frianova, J. Membr. Sci. 80 (1993) 99. S. A. Sengupta, R. Basu, K.K. Sirkar, AICHE J. 34 (1988) 1698. A.I. Alonso, M.I. Ortiz, B. Galan, A. Irabien, Simulation of hollow ber non-dispersive extraction with non-linear equilibrium conditions, in: Proceedings of ISEC 93, 1993, p. 801. F. Kubota, M. Goto, F. Nakashio, T. Hano, Sep. Sci. Technol. 30 (1995) 777. Y.J. Qin, J.M.S. Cabral, AICHE J. 43 (1997) 1975. Y.J. Qin, J.M.S. Cabral, AICHE J. 44 (1998) 836. A.I. Alonso, C.C. Pantelides, J. Membr. Sci. 110 (1996) 151. A. Kubaczka, A. Burghardt, T. Mokrosz, Chem. Eng. Sci. 53 (1998) 899. Y.J. Qin, J.M.S. Cabral, AICHE J. 44 (1998) 1529.