Modeling of counter-current adsorption in continuous liquid–solid circulating fluidized bed adsorption chromatography

Chemical Engineering Science 63 (2008) 1062 – 1071 www.elsevier.com/locate/ces Modeling of counter-current adsorption in continuous liquid–solid circulating fluidized bed adsorption chromatography Anil Gaikwad a,b,∗ , Sandeep Kale b , Arvind Lali b a Van’t Hoff Institute for Molecular Sciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands b Department of Chemical Engineering, Mumbai University Institute of Chemical Technology, Nathalal Parikh Marg, Matunga (E), Mumbai 400 019, India Received 1 March 2007; received in revised form 2 July 2007; accepted 26 October 2007 Available online 19 November 2007 Abstract Continuous chromatography using a liquid–solid circulating fluidized bed (LSCFB) is an efficient and eco-friendly system used for the separation and purification of biomolecules and radioactive waste. It combines continuous counter-current adsorption and simultaneous regeneration of solids. Here, we developed a mathematical model to study the performance of these counter-current adsorbers using MATLABTM which is a fundamental tool for scale up. This model predicts the adsorption zone length and height equivalent to theoretical plates (HETPs). In addition, this work also illustrates the comparative performance studies of adsorption in a conventional packed bed, expanded bed and LSCFB in terms of HETP values. At lower velocities such as 0.35 cm/s, HETPs of the LSCFB, packed bed, expanded bed are 5, 12, 18 cm, respectively. 䉷 2007 Elsevier Ltd. All rights reserved. Keywords: Counter-current adsorption; HETP; Chromatography; LSCFB 1. Introduction There are many reports on the gas–solid circulating fluidized bed systems, liquid–solid and gas–liquid–solid circulating fluidized bed systems have been scantily studied. Compared with conventional fluidized beds, circulating fluidized beds have many advantages including better interfacial contact, reduced back-mixing and continuous solid–liquid handling for higher throughputs (Lim et al., 1995). The continuous ion exchange processes using different fluidized bed systems were extensively investigated (Byers et al., 1997; Gordon et al., 1990; Higgins, 1969; Himsley, 1981; Porter, 1975; Slater and Prud’homme, 1972; Turner and Church, 1963; Van der Meer, 1985; Rodrigues et al., 1992). Most of the industries are giving a lot of attention for the handling of particles in circulating fluidized beds. Handling large amount of particles, transporting them in and out of the vessel continuously becomes necessary. These milestones can be easily achieved by using liquid–solid circulating fluidized bed (LSCFB) with higher mass transfer ∗ Corresponding author. Van’t Hoff Institute for Molecular Sciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands. Tel.: +31 20 5256477; fax: +31 20 5255604. E-mail address: gaikwad@science.uva.nl (A. Gaikwad). 0009-2509/$ - see front matter 䉷 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2007.10.037 capacities than conventional fluidized beds. Thereafter, detailed hydrodynamic studies of the LSCFB became essential (Liang et al., 1996). Zhu et al. (2000) proposed LSCFB system as a potential candidate for continuous recovery. A typical column adsorption is a transient state process. In a given column of adsorbents, only a part of the column (adsorption zone) is the only active fraction. Using circulating fluidized bed with far lesser quantity of resins (especially for large scale operations) is a useful concept and thus adsorption in continuous mode is a cheaper option. They also provide a consistent product quality and more output. These continuous processes are more convenient for control and optimization (Owen and Chase, 1999). Lan et al. (2000, 2002) have introduced the concept of the LSCFB and modeled the adsorption and desorption processes. Further, they studied the effect of operating conditions on the hydrodynamics of the LSCFB and developed the successful application of the LSCFB for the continuous recovery of bovine serum albumin (BSA). Feng et al. (2003) described the cesium separation using a continuous ion exchange circulating fluidized bed. It is likely that the particle mechanics governing the motion of the adsorbent in this counter-current contactor are complex. There is no direct relevant work in the literature on the behavior of the two phases in a similar environment, so A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 1063 theoretical predictions of the full column hydrodynamics could not be made with confidence. One approach for understanding the hydrodynamics of the solid phase can be made by noting that the counter-current system is similar to a sedimentation process. These two systems essentially consist of a solid phase of discrete particles raining against a liquid. Here we examined the past work on related processes, and made an attempt to predict the hydrodynamics of this operation using our own experimental results. The objective of the current work is to study the effect of various parameters on LSCFB and to model the counter-current adsorption process in the downcomer (vide infra). Then we also compared the performance of the LSCFB with packed bed and expanded bed adsorption (EBA) chromatography in terms of height equivalent to theoretical plates (HETPs) and dynamic binding capacity. 2. Materials and methods 2.1. Materials and instrumentation The adsorbent used in these particular experiments is polystyrene DVB based resins INDION 225H purchased from the Ion Exchange India Pvt. Ltd., India. These are porous resins of particle diameter ranges between 0.8 and 1.2 mm and the particle density is 1100 kg/m3 . Other particles used in the study are ADS 600 (density 1485 kg/m3 ) obtained from Thermax Pvt. Ltd., India. Nickel sulphate (NiSO4 ) was purchased from S. D. Fine Chemicals Ltd., India. All other chemicals are of analytical or commercial grade and obtained from local suppliers. 2% (w/v) of nickel sulphate (NiSO4 ) solution was used as a loading solution. A commercial grade hydrochloric acid (2% v/v) solution was used for elution of bound nickel ions. The NiSO4 concentration was analyzed using a JASCO V-530 UV/Visible spectrophotometer at 394 nm wavelength (). 2.2. Experimental setup and adsorption studies on the LSCFB The LSCFB setup used for the experimentation is constructed of glass, with the major dimensions as shown in Fig. 1. Main components of an LSCFB system include the riser, a downcomer, a liquid–solid separator, a top solids return pipe and a bottom solids return pipe. The riser is 1 cm in diameter and 100 cm tall. The down-comer diameter is 3 cm and the length is 100 cm. Down-comer has two inlet liquid streams D2 and D1 and one outlet at the top. The adsorption takes place in the downcomer and regeneration in the riser. Adsorbing solution (feed) in this case NiSO4 is passed through D2 continuously through the sintered glass plate distributor located inside the down-comer. Water used as a washing liquid that flows through auxiliary stream D1. The unadsorbed products are taken out from the top outlet. The riser primary liquid inlet R1 and the auxiliary liquid inlet R2 are located at the bottom of the riser. The eluent HCl passes through both the streams. Valve 1 and valve 2 are installed for solid sampling and controlling the solid circulation rate. The separator is connected to the riser through valve 1. The separator has one outlet to collect the eluent. The separator is then connected to the down-comer at Fig. 1. Schematic diagram of experimental setup of LSCFB. the top through the top solid return pipe. Valve 2 connects both columns at the bottom by the bottom solid return pipe. Rotameters are installed for controlling all liquid flow rates. Both the outlets of the separator and down-comer have been kept open to regulate the pressure fluctuation independent of the flow rates. The riser was operated in an entrained flow regime, while the down-comer in conventional fluidized regime. Net flow of the liquid in both the columns was upward. Solids are recirculated in such a way that net movement of the solids in the down-comer is downward and upward in the riser. Particles are circulated continuously between riser and down-comer using pressure difference between the two columns. Solids are removed from the down-comer bottom leg through the bottom solid return pipe and then regenerated continuously in the riser. Then they flow to the separator where elution and solids are separated. The regenerated solids go back to the down-comer just above the expanded solid bed through the top solid return pipe. Another auxiliary stream was injected in the top solid return pipe, which avoids the clogging of the separator due to packing of resin particles. Thus, the major function of auxiliary streams D1 and R2 is to increase the pressure drop across the bottom solid return pipe and maintain the continuous solid circulation. It is very important to maintain a dynamic seal between the riser and the down-comer by keeping the top solids return pipe as well as bottom solids return pipe in the moving packed bed regime. Solid inventory in the down-comer was kept constant. Liquid distribution was kept uniform for avoiding solid back mixing and radial voidage distribution. As for expanded beds, it is important that the characteristic of little mixing be retained in the counter-current unit. This means that carefully designed feed and eluent ports for the continuous and dispersed phases are required which do not affect the nature of the fluidization to a significant extent. 1064 A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 2.3. Adsorption studies on LSCFB Adsorption in the LSCFB system was carried out with a 2% w/v NiSO4 solution as feed and a 2% v/v solution of HCl for regeneration. INDION 225H ion exchange particles were used as adsorbents. Both the NiSO4 and HCl solutions were prepared in deionized water. In the adsorption studies, the feed solution was passed through the sintered glass distributor located in the bottom of the down-comer (D2). The liquid velocity in the down-comer was maintained lower than the terminal settling velocity of the particles, where the ion exchange particles move down counter currently to the rising liquid. As the particles and the feed contacted counter currently the Ni2+ ions adsorbs on the ion exchange particles. The nickel free solution was removed out from the top of the downcomer. After washing in the bottom section of the down-comer by stream (D1), the adsorbed particles transferred to the bottom of the riser through the bottom solid return pipe. 2% (v/v) of HCl solution was passed from the two streams (R1 and R2) of the riser. The total superficial velocity in the riser is the combination of the primary and the auxiliary streams, which was kept higher than the terminal settling velocity of the particles. Particles entering in the riser were continuously regenerated and carried to the separator by the liquid stream. The desorbed nickel solution and the regenerated ion exchange particles were separated in the separator. The nickel solution was taken out from the separator top outlet. Then, the regenerated particles were returned back to the downcomer through the top solid return pipe. After the steady state was achieved it was found that adsorption zone length (green inter phase) in downcomer remains constant. Also, the Ni2+ in the eluent achieves a steady state value for a give feed velocity and concentration (D2). From operating curves and equilibrium curve, the number of transfer units and HETP were determined. This procedure was repeated for different velocities and at different concentrations of NiSO4 . 2.4. Computation and modeling of counter-current adsorption Here we model the adsorption in the down-comer as a continuous counter-current EBA. The EBA is a batch operation. In continuous and counter-current mode, adsorbents are constantly removed from the base of the column and simultaneously supplied at the top of the column. The liquid flows from bottom to top of the column. Conventionally, in the batch operated expanded beds, each adsorbent particle remains approximately stationary at a point which is governed by liquid flow rate, particle density, etc. (Chase, 1994). Conversely, in this system, removal of adsorbent from the base of the bed causes suspended particles to fall with respect to the upward liquid stream, thus resulting in a net downward flow of the adsorbent material (replacement of adsorbent at the top of the column ensures that the adsorbent inventory is retained). The conservation equation based on the liquid phase species transferred from the solute to the sorbents (solid) is given by following Eq. (1) (Owen and Chase, 1999). This equation includes diffusive mass transfer and convective transfer.  includes the effects of adsorbate (liquid) transfer through the film thickness, (concentration boundary layer surrounding each adsorbent particle) and transfer into the particle pores. Eq. (1) can be written for both solid as well as liquid phase. jc = Dl × ∇ 2 c − ∇{c, f } − . jt (1) The two-phase mathematical model was developed according to the following assumptions. (1) The axial dispersion plug flow model for both solid and liquid phases is applied. (2) The dispersion term in this model assumes that any dispersive transport that takes place within the system can be modeled in terms of a pure diffusion-type expression. (3) The amount of forward and back-mixing taking place within the modeled volume is equal. (4) Free diffusivity in pure molecular diffusion (the magnitude of the dispersion coefficient remains constant throughout the column irrespective of flow rates and starting concentrations). (5) The system obeys the Langmuir adsorption isotherm (this was also observed experimentally). (6) The adsorbate is in thermodynamic equilibrium at the solid–liquid interface. (7) The transfer of the solute from liquid to solid has negligible effect on linear velocities of the solid and liquid phases. (8) Velocity in upward direction was taken as positive. (9) All mass transfer resistances are limited to the film around the particle. Equation for liquid phase is derived from Eq. (1) as follows: Daxl × j 2 Cl Ul jCl jCl − × − Kf × (Cl − Ci ) = . εl jZ jt jZ 2 (2) Equation for solid phase is as follows: Daxs × j 2 Cs Us jCs jCl + Kf × (Cs − Ci ) = . ± × εs jZ jt jZ 2 (3) The overall material balance that couples the two equations is given as (Cli − Clo ) × Ci = Ul Us = (Csi − Cso ) × , εl εs Cs max × Cl , Kd + C l (4) (5) where, Dax(l,s) is the axial dispersion coefficient (m2 /s). C(l,s) is the mobile phase concentration (kg/m3 ). U(l,s) is the superficial velocity (m/s). εl,s is the hold up. (l—liquid, s—solid). Kf is the Film mass transfer coefficient (m/s). Ci is the interfacial concentration (kg/m3 ). Kd is the dissociation constant. Cs max is the solid saturated concentration (kg/m3 ). Clo , Cli , Cso , Csi are liquid and solid outlets and inlets, respectively. The two second order differential equations (2) and (3) were solved simultaneously under following boundary conditions and results were verified by overall material balance across the column (Eq. (4)). At Z = 0 Cl = Cin at Z = L Cs = Csin dcs = 0, dz dcl = 0, and dz and (6) (7) A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 1065 where Cin , Csin are the initial liquid and solid phase concentrations (kg/m3 ). The solution of the model was obtained using MATLABTM (BVP 4c subroutine) for solving simultaneously higher order differential equations. Simulated profiles of dimensionless concentrations versus length of the column were obtained by numerical integration. Kinetic parameters of the selected system such as rate constant (k1 ) and dynamic capacity (qm ) were found experimentally. The parameters of Eqs. (2) and (3), namely, the film mass transfer coefficient and the phase axial dispersion coefficients were determined from available correlations (vide infra). The solutions were obtained as dimensionless plots of concentration versus bed height at different velocities and different particle sizes. Eq. (5) was used for obtaining the interfacial concentration at the surface of the particle. Then, using this value in the overall mass balance (Eq. (4), the concentration at the surface of the film was determined. The length of the adsorption zone was estimated from the solid phase concentration profile. This was then compared with the observed adsorption zone length. The HETP and number of theoretical plates were calculated from the equation discussed below (vide infra). 2.5. Determination of parameters required in the equations for modeling the counter-current system In order to make use of the above-derived models for predicting the performance of counter-current systems, the relevant parameters used in the equations need to be quantified. This can be done by making use of suitable correlations, and carrying out suitable adsorption experiments, as outlined below. 2.5.1. Determination of stable operation window of LSCFB A circulating fluidized bed operates under more variables than the traditional fluidized bed. LSCFB includes the circulation of solids and its net flux is a strong function of the operating flow rates in the two connected columns. Since riser and down-comer have different diameter and also the liquid–solid flow is co-current and counter-current, respectively, the pressure drops and hydrostatic heads in the two limbs follow different function. Hence, a judicious pressure balance has to be maintained in the two columns for a desirable stable solid flow from down-comer to riser at bottom and riser to down-comer at the top. Thus, for a given LSCFB system, there is a region of operating flow rates in the two columns that will mark the flow rates required for stable LSCFB operation. This is called the operating window. The ‘operating window’ can be estimated from pressure balances as well as determined experimentally. In the present system the calculation of the operating window was done based upon the pressure balance at the four points 1, 2, 3 and 4 as shown in Fig. 2. At all these points major pressure element was hydrostatic heads. So considering only the dispersion densities, pressures at respective points were calculated. This exercise was then repeated for different riser; down-comer velocities and solid flow rates. Then those velocities and solid circulation rates were plotted on X-, Y- and Fig. 2. Schematic diagram of LSCFB showing the pressure heads. Z-axes, respectively. This plot defines the stable operating window. Eqs. (8)–(14) determine the pressure at each point as a function of voidage in the riser and down-comer. For a stable solid flow rate, the pressure at point 1 (P1 ) > pressure at point 2 (P2 ), and pressure at point 3 (P3 ) > pressure at point 4 (P4 ). Assumptions here are (1) pressure due to the velocity head is neglected; (2) frictional pressure drop is not considered. Pressure at points 1 and 2 (P1 and P2 ): P1 = Hr1 ∗ L ∗ g + Hs ∗ r1 ∗ g, (8) r1 = s ∗ s + L ∗ l , (9) P2 = Hd1 ∗ L ∗ g, (10) where Hr1 is the clear liquid height in the separator, L is the liquid density, Hs is the solid hold up height in the separator, r1 is the dispersion density of solid hold up in the separator, g is the acceleration due to gravity and Hd1 is the clear liquid height in the down-comer. s and L are the solid and liquid densities, respectively. s and l are the solid and liquid fractions, respectively. As we know the total height of the downcomer Hd = Hd1 + Hd2 + Hd3 , (11) 1066 A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 where Hd2 is expanded bed height in the downcomer and Hd3 is the packed bed height in the down comer. Since Hd2 = f (velocity, ) and so the Hd and Hd1 . Thus, P2 = f (velocity, ). Pressure at points 3 and 4 (P3 and P4 ): P3 = P2 + Hd2 ∗ d2 ∗ g + Hd3 ∗ d3 ∗ g, (12) d3 = s ∗ s3 + L ∗ l3 , (13a) d2 = s ∗ s2 + L ∗ l2 , (13b) P4 = Hr1 ∗ L ∗ g + Hr2 ∗ r2 ∗ g, (14) where d3 and d2 are the dispersion densities of the expanded region and the packed region, respectively, in the downcomer section. s3 and l3 are the solid and liquid fraction in packed bed region. s2 and l2 are the solid and liquid fraction in the expanded bed region. r2 is the dispersion density in the riser. Pressure at four points can be calculated by above method for different particles. Since the pressures at points 1 and 2 are almost constant, pressures at points 3 and 4 were calculated at different down-comer and riser velocities for the particles with different densities to plot the operating window (note that all heights are in m and densities in kg/m3 ). 2.5.2. Determination of HETP and dynamic capacity in packed and EBA Adsorption studies were carried out in packed and expanded bed using a 2.5 cm diameter column and INDION 225H strong cation exchange resin as adsorbent. NiSO4 (2% w/v solution) was loaded by a peristaltic pump at different flow rates in packed bed as well as expanded mode and 2% v/v solution of HCl was used for elution. NiSO4 gives a green color and can be physically observed in the glass column. The colored effluent sample was analyzed spectrophotometrically at a wavelength of 394 nm. The breakthrough curves were found at different velocities. From these the breakthrough curves HETP and the dynamic binding capacities were calculated using equations given by Lali (2002). In the frontal adsorption chromatography, the column length L is the active adsorbing zone for the solute at any given time. Therefore, the height of the active adsorption zone La in the column is given by the following equation: La = L V E − VB , VE − (1 − f )(VE − VB ) (15) where f is the fractional ability of adsorption, which is nothing but the ratio of the amount adsorbed to the amount loaded. VE is the exhaust concentration and VB is the breakthrough concentration. All these parameters were obtained from breakthrough curves for different velocities. The number of theoretical plates, N were calculated by using equation as  2 VB N = 16 . (16) VE − V B The HETP is then given by the following equation as HETP, H = La . N (17) 2.5.3. Determination of equilibrium parameters Batch studies have been carried out to find the maximum capacity of the resin and the dissociation constant. Various concentrations of NiSO4 solution were prepared and contacted with the 1 ml of washed and degassed INDION 225H ion exchange resins. The supernatant was removed after 24 h and analyzed for NiSO4 spectrophotometrically. The equilibrium concentration of NiSO4 was plotted against the adsorbed concentrations. The Langmuir adsorption isotherm was fitted and from this the maximum adsorbed concentration (qmax ) and dissociation constant (kD ) were determined. 2.5.4. Determination of the film transfer coefficient The film mass transfer coefficient in column chromatography was determined from the Chung and Wen (1968) correlation. This was found to be very well in accordance for countercurrent adsorption also. The value of film transfer coefficient was found to be 1.8 × 10−8 m/s, which is approximately to the reported value. 2.5.5. Determination of inter-particle voidage in the counter-current contactor The inter-particle voidage in the contactor is calculated from the Richardson Zaki equation, given by  = (u/ut )1/n . (18) Coefficients used in the above equation describing the fluidization characteristics of the adsorbent were obtained from McCreath et al. (1995). 2.5.6. Determination of solid-phase dispersion coefficient The aim is to quantify deviations from plug flow of the adsorbent as it rains down the upwardly flowing process solution in the counter-current contactor. However, because of the practical difficulty of labeling individual adsorbent particles and then measuring their residence time in the contactor, we evaluate dispersion coefficients for the adsorbent, by fitting the model derived above to the experimental data obtained for the continuous adsorption of NiSO4 in LSCFB. 3. Results and discussions 3.1. The stable operating window Experiments were carried out with particles of different densities, fluidized by water in LSCFB. The overall recirculation of the particles to the riser at bottom is governed by the overall pressure balance. In order to maintain the system under stable operating conditions, the auxiliary and primary liquid flow rates need to be maintained. The equilibrium flow velocities required for the stable operation depends on the loading and the density of particles (Zheng et al., 1999). The down-comer auxiliary velocity (D1 = minimum fluidization of particles) was kept constant. Two streams of the riser (R1 and R2) were operated at the same flow rates. First for a constant down-comer velocity (D2) different riser velocities were determined for a A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 1067 Fig. 3. Stable operating window for particles of different densities. Windows: outermost—2500 kg/m3 (), inner—1485 kg/m3 (), innermost—1000 kg/m3 (). Fig. 5. (Top) Variation of solid flow rates with respect to down-comer auxiliary flow rates. (Bottom) Variation of solid flow rates with respect to riser auxiliary flow rates in absence of riser inlet flow rate. Fig. 4. Shows solid circulation rates (particle density—1485 kg/m3 ) vs riser velocity at down-comer velocities; (•) 0.4 cm/s and () 0.8 cm/s. stable solid circulation. Fig. 3 shows the stable operating window at different riser velocities versus down-comer velocities for particles of different densities. As the density of the particle increases the size of the window increases, i.e., the system can be operated in larger velocity ranges. The left side of the window represents the minimum fluidization of the particle. Lower side of the window represents the limitation due to pressure i.e., P3 > P4 . Right side of the window shows the velocity corresponding to maximum 0.6 voidage in the down-comer. The upper side represents the maximum velocity in the riser to attain 0.8 voidage (note that the riser is in the entrained flow rigime). For higher density particles the pressure drop across points 3and 4 increases, this reduces the instabilities in the operation and allows stable solid circulation over a wide range of velocities (see the pressure balance at points 3 and 4, Eqs. (12) and (14)). Since denser particles have a greater terminal settling velocity, the system can be operated at higher flow rates, and the size of the operating window is enlarged. Fig. 6. (Top) Inlet flow position configurations in the riser bottom (R1). (A) earlier systems and (B) modified system. (Bottom) solid circulation rates at two different riser inlet positions A (•) and B (). 1068 A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 Fig. 7. Three dimensional window at riser configuration B with particle density 1485 kg/m3 . Once the stable flow velocities were determined then the solid circulation rates were measured with respect to following different parameters. 3.1.1. Variation in solid circulation rate with respect to riser and down-comer velocity The operation of LSCFB is greatly depending on the net solid circulation rate. The solid flux is in turn governed by the two column velocities. Fig. 4 shows the solid flow rate experimentally determined with respect to the riser velocities (R1 and R2) at two different downcomer velocities (D2) and at the riser inlet (R1) configuration A as shown in Fig. 6 (top). The solid flow rate was measured by momentarily opening valve 1. At this riser inlet configuration and at constant D2 velocity, the increase in R1 and R2 (R1 and R2 were kept at the same velocity) decreases the solid circulation rate. This may be because at higher rise velocities P4 increases. Then due to the reduced pressure driving force (P3 − P4 ) the solid flow rate goes down. This study also indicates that an increase in feed flow rate (D2) results in the expansion of the bed in the down-comer. Due to this the dispersion density and ultimately P3 goes down. This decreases the pressure driving force across the bottom of the two columns. 3.1.2. Variation in solid circulation rates with respect to down-comer and riser auxiliary flow rates The solid circulation rate varies with auxiliary flow rates R2 and D1, also in addition to feed flow rate D2 and riser flow rate R1. The results for the variation in solid flow rate at various down-comer and riser auxiliary flow rates are plotted in Fig. 5 top and bottom, respectively. As the down-comer auxiliary flow rate increases, the particles tend to flow freely to the riser and increases the solid circulation rate (Fig. 5). Increase in the riser auxiliary flow rate (R2) increases the velocity at point 4 (Fig. 2) and according to Bernoulli’s equation decreases P4 . This causes an increase in the solid flow rate. It is also observed that the solid circulation rate is stronger function of the riser auxiliary flow rate (R2) than that of the downcomer (D1). 3.1.3. Solid circulation studies with respect to riser inlet configuration System configuration can play a major role in the solid circulation flow rates. It was observed that the solid flow rate is a function of auxiliary liquid inlet position of the riser (Feng, 2003). Keeping all other flow rates constant, solid circulation rates were investigated at two different locations of the riser inlet, which are shown in Fig. 6 (top). The results obtained by changing the inlet geometry of the riser liquid inlet (R1) are plotted in Fig. 6 (bottom). With the riser inlet configuration A, as the velocity increases the solid circulation rate goes down (discussed in earlier section). In the modification as shown in Fig. 6, at inlet configuration B, the riser inlet was placed at the center of the down-comer return leg. This configuration results in better conversion of the pressure in the velocity head at the A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 1069 Fig. 8. Equilibrium adsorption isotherm of NiSO4 on INDION 225H. solid entry point and results in steady increase in solid flow with riser flow velocity. A complete 3-D window (Fig. 7) was made by incorporating the solid flow rate into the 2-D window (Fig. 3) presented earlier with respect to operating velocities. 3.2. Dynamic capacity and HETP studies 3.2.1. HETP and dynamic capacity studies in packed bed and EBA First, we performed equilibrium adsorption studies. The adsorption isotherm of NiSO4 on INDION 225H is shown in Fig. 8. As it can be seen, the shape of the isotherm is approximately rectangular. So it is appropriate to represent adsorption by an empirical fit of the Langmuir adsorption isotherm. The fitted parameters of Langmuir equation are qmax , 91.0 mg/ml and kD , 0.23 mg/ml. HETP is an important parameter that decides the required bed height for a given separation. It is reported that HETP in counter-current systems is lower than fixed beds or cocurrent systems (Owen and Chase, 1999). Fig. 9 shows the breakthrough curves in packed and expanded bed. HETP was estimated from these breakthrough experiments using NiSO4 adsorption on INDION 225H ion exchange columns in both packed and expanded bed mode. These HETP values are then compared with counter-current adsorption in the LSCFB downcomer. 3.2.2. HETP studies on LSCFB system Nickel adsorption from 2% w/v NiSO4 solution was carried out with a bed of height 32 cm in the down-comer. For different feed velocities (D2) riser flow rates were adjusted to get a stable solid circulation flow rate of 24 ml/min. Since Ni2+ imparts green color to the resin, the adsorption zone height in the downcomer was visually measured. Fig. 9. (Top) Breakthrough curve of NiSO4 in packed bed mode using INDION 225H resins at flow rate of 6 ml/min (), 9 ml/min (•), 15 ml/min (), 20 ml/min () and 30 ml/min (—). (Bottom) expanded bed mode at flow rate of 10 ml/min (), 16 ml/min (—), 22 ml/min (), 30 ml/min (•) and 40 ml/min (). 3.3. Variation of adsorption zone length with respect to feed velocities and total voidage Fig. 10 shows the profile of the dimensionless mobile phase concentration (y1 ) and adsorbed solid phase concentration (ys1 ) versus the dimensionless column height. As the velocity increases effective transfer may not take place due to which some of the liquid feed leaks out of the column without any adsorption. This increases the adsorption zone length and in turn increases the total volume of required adsorbent. Fig. 11 shows the variation in adsorption zone length with respect to different size of the adsorbent particles. As the diameter of the particle increases the specific area for mass transfer decreases, which in turn increases the adsorption zone length. For 0.0012 m diameter particles the dimensionless adsorption zone length is 0.7, while for 0.002 m diameter particles the bed is exhausted due to inefficient mass transfer. Table 1 shows that the modeled and experimental adsorption zone heights are in good agreement with each other. Since the adsorption process is limited by equilibrium and capacity of the beads the adsorption zone height increases with increase in concentration. From this adsorption zone length and knowing 1070 A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 Fig. 10. (Top) Liquid phase concentration for particles of diameter 0.0012 mm at down-comer feed liquid velocity 0.09 cm/s (1), and at liquid velocity 0.179 cm/s (2). (Bottom) shows the corresponding solid phase concentration. Fig. 11. (Top) Liquid phase concentration at particle diameter Dp = 0.0012 m (1), particle diameter Dp =0.002 m (2) and (bottom) shows the corresponding solid phase concentration. Table 1 Adsorption zone height and HETP at varying velocities the equilibrium and operating curve the HETP of the LSCFB system was determined. Fig. 12 shows the comparison of the HETP of the packed, expanded and LSCFB system. It shows that with increased velocity, HETP increases, which decreases the performance of the given adsorption system. It is known that an increase in velocity increases the radial dispersion and the system starts going away from the plug flow. For a given system above a certain value of flow rate, the mobile phase starts leaking through the column without any transfer between mobile phase and adsorbent. This reduces the effective mass transfer, resulting in reduced performance of operation. It was observed that at higher velocities, the expanded bed HETP increases sharply over packed bed HETP. At lower velocities the HETP of the packed column and the expanded bed column are comparable, because at lower velocities an expanded bed behaves like a packed bed. This is an advantage for the systems, which contain D2, velocity (cm/s) Modeled adsorption zone height (cm) Experimental adsorption zone height (cm) HETP (cm) 0.046 0.09 0.123 0.176 9 12 18 27 5 8.5 21 30 (almost exhausted) 2.5 2.9 5.25 7.5 particulate materials because it reduces the separation steps that are needed before the chromatographic separations. It was observed that the HETP is lower in the case of LSCFB because adsorption takes place in the counter-current mode and this maximizes the driving force at each stage. At each stage particles attain their equilibrium concentration value which increases the number of transfer units. So the performance of the LSCFB is comparatively better than packed and expanded bed mode. For example, at velocity 0.2 cm/s the HETP for a packed A. Gaikwad et al. / Chemical Engineering Science 63 (2008) 1062 – 1071 1071 performance of the LSCFB. Detailed hydrodynamic studies of the riser need to be done. Acknowledgments We would like to thank Peter Verschuren for editing and Dr. Kishore Jahagirdar for helpful suggestions. References Fig. 12. Comparison of HETP of packed bed, expanded bed and LSCFB system (HETP values for LSCFB at higher velocities were not compared because the system goes out of stability window at higher velocities). bed is the highest, i.e., 10 cm, an expanded bed has a HETP of 6 cm and the LSCFB has the lowest HETP, i.e., 2 cm. The convective motion of the solid augments the solid–liquid mass transfer but one should be careful about the effect of convective motion of both solid–liquid on their dispersion. 3.4. Preliminary RTD studies For checking the intermixing of the liquid streams in the riser and down-comer, the RTD experiments were conducted using Erichrome blue dye as a tracer. Non ionic beads (ADS 600) were used to avoid dye adsorption on the beads. The 1% w/v solution of Erichrome blue was injected in the down-comer and the tracer concentration was measured in the riser and separator section, i.e., in the outlet of the separator. Then the tracer was injected in the riser and the dye concentration was measured in the down-comer outlet. Visual inspection of RTD studies showed that there is no intermixing of riser and down-comer liquids when the tracer was passed through the either of the column. 4. Conclusion The LSCFB gives a better performance than any other conventional fluidized operations. Lowest HETP values of the LSCFB system allow the use of reduced height of column and resin amount. Since the pore resistance is neglected, the model results did not match accurately to the experimental values but are nevertheless in good accordance. As there is non uniformity in the solid holdup through out the bed and radial solid distribution, it is better to use local dispersion values. This gives an accurate picture of the process. One needs to look deeper into microstructure of the solid and liquid profiles to get a real idea of the effect of these parameters on the hydrodynamic Byers, C.H., Sisson, W.G., Snyder, T.S., Beleski, R.J., Nayak, U.P., Francis, T.L., 1997. Zirconium and hafnium separation in sulfate solution using continuous ion exchange chromatography. U.S. Patent No. 5,618,502. Chase, H.A., 1994. Purification of proteins by adsorption chromatography in expanded beds. Trends in Biotechnology 12, 296–303. Chung, S.F., Wen, C.Y., 1968. Longitudinal dispersion of liquid flowing through fixed and fluidized beds. A.I.Ch.E. Journal 14, 857–866. Feng, X., Jing, S.W., Chen, J., Song, C., 2003. The hydrodynamic behavior of the liquid–solid circulating fluidized bed ion exchange system for cesium removal. Powder Technology 134, 235–242. Gordon, N.F., Tsujimura, H., Cooney, C.L., 1990. Optimization and simulation of continuous affinity recycle extraction. Bioseparation 1, 9–21. Higgins, I.R., 1969. Continuous ion exchange of process water. Chemical Engineering Progress 65, 59–62. Himsley, A., 1981. Continuous countercurrent ion exchange process. U.S. Patent No. 4,279,755. Lali, A., 2002. Expanded bed affinity chromatography. In: Gupta, M.N. (Ed.), Methods for Affinity Based Separations of Enzymes and Proteins. Birkhauser, Basel, p. 29. Lan, Q., Zhu, J.X., Bassi, A.S., Margaritis, A., Zheng, Y., Rowe, G.E., 2000. Continuous protein recovery using a liquid–solid circulating fluidized bed ion exchange system: modeling and experimental studies. Canadian Journal of Chemical Engineering 78, 858–866. Lan, Q., Bassi, A., Zhu, J.X., Margaritis, A., 2002. Continuous protein recovery with a liquid–solid circulating fluidized-bed ion exchanger. A.I.Ch.E. Journal 48, 252. Liang, W.G., Zhu, J.X., Jin, Y., Yu, Z.Q., Wang, Z.W., Zhou, J., 1996. Radial nonuniformity of flow structure in liquid–solid circulating fluidized bed. Chemical Engineering Science 51, 2001–2010. Lim, K.S., Zhu, J.X., Grace, J.R., 1995. Hydrodynamics of gas fluidization. International Journal of Multiphase Flow 21 (Suppl.), 141–193. McCreath, G., Chase, H., Owen, R., Lowe, C., 1995. Expanded bed affinity chromatography of dehydrogenases from bakers yeast using dye ligand perfluoropolymer supports. Biotechnology Bioengineering 48, 341–354. Owen, R.O., Chase, H.A., 1999. Modeling of the continuous countercurrent expanded bed adsorbers for the purification of proteins. Chemical Engineering Science 54, 3765–3781. Porter, R.R., 1975. Continuous ion exchange process and apparatus. U.S. Patent No. 3,879,287. Rodrigues, M.I., Zaror, C.A., Maugeri, F., Asenjo, J.A., 1992. Dynamic modelling, simulation and control of continuous adsorption recycle extraction. Chemical Engineering Science 47, 263–269. Slater, M.J., Prud’homme, P., 1972. Continuous ion exchange in fluidized beds. Technical Bulletin, vol. 158, Mines Branch, Canada, p. 79. Turner, J.C.R., Church, M.R., 1963. A continuous ion exchange column. Transaction of the Institution of Chemical Engineers 41, 283–288. Van der Meer, A.P., 1985. On counter-current fluidized ion-exchange columns. Ph.D. Thesis, Delft University of Technology, Delft, The Netherlands. Zheng, Y., Zhu, J.X., Wen, J., Martin, S.A., Bassi, A.S., Margaritis, A., 1999. The axial hydrodynamic behavior in a liquid–solid circulating fluidized bed. Canadian Journal of Chemical Engineering 77, 284–290. Zhu, J.X., Zheng, Y., Karamanev, D.G., Bassi, A.S., 2000. (Gas)–Liquid–solid circulating fluidized beds and their potential applications to bioreactor engineering. Canadian Journal of Chemical Engineering 78, 82–94.