You are on page 1of 12

Chinese Journal of Chemical Engineering, 19(5) 833844 (2011)

Prediction of Distillation Column Performance by Computational Mass Transfer Method*


SUN Zhimin (), LIU Chunjiang (), YU Guocong (K. T. YU )** and YUAN Xigang ()

State Key Laboratory for Chemical Engineering (Tianjin University) and School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China Abstract A computational mass transfer model is proposed for predicting the concentration profile and Murphree

efficiency of sieve tray distillation column. The proposed model is based on using modified c2 - c two equations formulation for closing the differential turbulent mass transfer equation with improvement by considering the vapor injected from the sieve hole to be three dimensional. The predicted concentration distributions by using proposed model were checked by experimental work conducted on a sieve tray simulator of 1.2 meters in diameter for desorbing the dissolved oxygen in the feed water by blowing air. The model predictions were confirmed by the experimental measurement. The validation of the proposed model was further tested by comparing the simulated result with the performance of an industrial scale sieve tray distillation column reported by Kunesh et al. for the stripping of toluene from its water solution. The predicted outlet concentration of each tray and the Murphree tray efficiencies under different operating conditions were in agreement with the published data. The simulated turbulent mass transfer diffusivity on each tray was within the range of the experimental result in the same sieve column reported by Cai et al. In addition, the prediction of the influence of sieve tray structure on the tray efficiency by using the proposed model was demonstrated. Keywords simulation, concentration field, computational mass transfer, computational fluid-dynamics, tray efficiency, sieve tray, turbulent mass transfer diffusivity

INTRODUCTION

The tray column has been widely used in many chemical separation processes, such as distillation, absorption, humidification etc. The common design method of tray column nowadays has been advanced from the equilibrium tray concept with empirical tray efficiency to the non-equilibrium tray calculation with mass transfer rate for evaluating the extent of separation. Nevertheless, many researches [1-4] have reported that the separation efficiency was frequently lowered due to the maldistribution of flow on the tray, such as recirculation in the near wall region and the flow channeling etc., which was hardly predicted in advance. However, this problem can not be solved by the conventional design method, as the conditions of flow and concentration on a tray are ignored in the model calculation. The lacking of in-depth understanding of the fluid dynamics and mass transfer behaviors occurring inside the distillation column is the major barrier to the accurate design and evaluation of the tray efficiency. Following the successful application of computational fluid dynamics (CFD) to the engineering field, such methodology has been used to predict the flow behaviors on the tray column. Zhang and Yu [5] employed the k- two equations model of CFD for simulating the liquid phase flow pattern on a sieve tray. Yuan et al. [6, 7] used two phases k- model with new

model constants to give better simulation for sieve tray. Mehta et al. [8] predicted liquid velocity on sieve tray using the steady-state model. Krishna et al. [9] and van Baten and Krishna [10] simulated flow behaviors for a rectangular and a circular trays by the transient two-phase flow model. Liu et al. [11] modified the k- model by considering the bubbling effect on the sieve tray. Gesit et al. [12] predicted velocity distributions on a commercial scale sieve tray. Wang et al. [13] simulated a 1.2 m diameter column with 10 sieve trays with consideration of more influential effect. The understanding of the flow behaviors on a tray is indeed useful to the designer, yet the prediction of tray efficiency may even be more important. As the tray efficiency is dependent both on the flow condition and the concentration distribution, the prediction of the latter becomes a pending problem to be investigated. The concentration distribution of liquid flowing through a tray can be obtained by solving the following turbulent differential mass transfer equation for the liquid phase on the column tray: C U i C C + = D uic + Sn (1) t xi xi xi where C is the mass concentration of separating component, kgm3, U i is the time-averaged velocity of the liquid phase which can be calculated by using CFD, D is the molecular diffusivity, uic is the fluctuating

Received 2011-06-20, accepted 2011-08-25. * Supported by the National Natural Science Foundation of China (20736005). ** To whom correspondence should be addressed. E-mail: ktyu@tju.edu.cn

834

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

mass flux, which is unknown, Sn is the source term representing the mass transfer rate. The usual simplification of above equation is by applying the hypotheC to give the following sis of letting uic = Dt xi form:

C U i C C + = ( D + Dt ) + Sn t xi xi xi

(2)

Where Dt is defined as turbulent mass transfer diffusivity (sometimes called dispersion coefficient or coefficient of back-mixing). If Dt can be evaluated, the local concentration C as well as its distribution can be obtained. The conventional method to find the unknown Dt is by employing a relevant empirical equation which is obtained by using inert tracer technique, such as injecting dye solution to the equipment. Nevertheless, the governing equation for the inert tracer in the process without mass transfer, i.e. S n = 0 in Eq. (2), is different from that with mass transfer where S n 0 . Thus the Dt obtained by inert tracer technique is not the real one in the mass transfer process and its use is only an approximation and may produce substantial error. Even so, such empirical equation in many cases may not be available in the literature. Another conventional way to find the unknown Dt is by assuming a turbulent Schmidt number [9], such as

profile on sieve tray are available in the literature to verify the prediction by the proposed model, experimental work was conducted for concentration measurement on a sieve tray simulator of 1.2 meters in diameter undergoing the desorption of dissolved oxygen in water by blowing air. In addition, the validation of the proposed model to the process column was also examined by predicting the outlet concentration of each tray and the Murphree tray efficiency under different operation conditions of an industrial scale column, in which the toluene was stripped from its water solution, as reported by Kunesh et al [25]. The comparison between predictions and experimental measurements are shown in subsequent section.
2 THE PROPOSED MODEL

In general, the computational mass transfer (CMT) model is composed of three parts of equations, namely the mass transfer part, the fluid-dynamics part, and the heat transfer part. In some mass transfer processes, if the heat effect can be neglected, i.e. where the temperature is substantially unchanged in the computed domain, such as a single distillation tray, the heat transfer part of equations can be omitted. Then the CMT model for simulating sieve tray can be simplified to involve only two parts of equation. If the heat effect can not be neglected, such as the exothermic chemical absorptions, the computation should consist of all three parts.
Assumptions (1) The pseudo liquid phase modeling is used in order to reduce the computer load. As shown in the subsequent sections, the simulated result by using the pseudo liquid phase model with proper vapor-liquid interacting source term are comparable with that by using vapor liquid two phases model. (2) For the simulation of a single tray, the liquid density is considered to be a constant, but if multi-tray is simulated, the density is changed tray by tray. The mathematical expression of the proposed model is as follows: Mass transfer part of the proposed model The steady state differential turbulent mass transfer equation for pseudo liquid in vapor-liquid flow on a sieve tray is as follows:
C (3) L D x L uic + Sn i Where C is the time average mass concentration, kgm3; Sn is the source term for the mass transfer between vapor and liquid phases: L is the liquid volume fraction in the liquid-vapor mixture, which can be calculated by using the correlation by Bennett et al. [26]: LU i C = xi xi

= 0.7 , but it lacks theoretical basis and Dt therefore is not reliable. Thus it is necessary to investigate a rigorous method to evaluate the Dt without using the foregoing mentioned empirical method. From this viewpoint, a computational mass transfer
(CMT) model, consisting of c2 - c two-equation and relevant k- formulation, was proposed by Liu [14] and improved by Sun et al. [15-17] for solving the unknown parameter Dt. With this model, Sun et al. [15] predicted the tray efficiency of a commercial scale sieve tray column reported by Sakata and Yanagi [18] and a pilot scale sieve tray column reported by Garcia and Fair [19]. Their predicted results, including Dt and tray efficiency, were in agreement with the experimental data. Besides applying to distillation tray column, Liu et al. employed this model to simulating the packed column undertaking distillation, chemical absorption and catalytic reaction [20-23], and Li et al. extended the application to the adsorption and desorption [24]. In this paper, the foregoing mentioned c2 - c model is further improved by using a distributed three dimensional source term for better prediction of concentration distribution on sieve tray. Since no published data on the concentration

Sct =

G L = exp 12.55 Us L G

0.91

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

835

The covariance term uic in Eq. (1) can be postulated, similar to the Boussinesqs hypothesis, by C letting uic = Dt , in which Dt is given below: xi k c 2 Dt = Ct k c
1/ 2

have: U U j uiu j = t i + x j xi 2 k 3 ij (9)

where t = C k 2 / , ij is the Kronecker delta, and the equations representing the kinetic energy k and its dissipation rate are given below [27]:

(4)

where the concentration variance c2 and its dissipation rate c can be calculated by the following equations [15]:
D c 2 LU i c = L D + t xi xi c2 xi
2

Ui

U j k k = + t uiu j xi xi k xi xi
Ui

(10)

C 2 L Dt 2 L c xi
LU i c D = L D + t c xi xi c xi

(5)

= + t x x i xi i U j 2 C 1 uiu j C 2 k k xi

(11)

The model constant are: C = 0.09, C 1 = 1.44, C 2 = 1.92, k = 1.0, = 1.3.


Distributed source term The source term SMi in Eq. (8) is an important factor affecting the simulated velocity distribution. Most of the previous works regarding the source terms in the modeling equation are based on the interaction between the rising bubble and the liquid flow and expressed as drag, lift and virtual-mass forces. Among these forces, the drag force is commonly adopted by researchers [11, 12], while the lift and virtual-mass forces are often neglected. Among different ways of calculating the source term, the superficial vapor velocity is commonly adopted for repressing the action of vapor. However, since the vapor velocity leaving the sieve holes is much higher than the superficial and sometimes even forming jet flow; such influential effect can not be ignored, especially under the condition of high F-factor. For the vapor-liquid cross flow on the sieve tray, the three dimensional vapor velocities leaving the sieve hole can be given by the following relationship [28]:

Cc2 c Cc3 L c 2 x k c i c 2 (6) The model constants in foregoing equations are [15]: Ct = 0.14, Cc1 = 1.8, Cc2 = 2.2, Cc3 = 0.8, c2 = 1.0, Cc1 L Dt

c C

= 1.0.
c

Fluid-dynamic part of the proposed model This part of equations is essentially the CFD formulation for simulating the pseudo liquid velocity distribution. The equations of the steady-state continuity and momentum for turbulent flow on the tray are given below:

LU i =0 xi
LU iU j xi 1 P = L + L x j xi

(7)

U Gz = 4.0U h

d 2 Dh exp z 0.1z

U j L uiu j + L g + SMi (8) L xi


U Gr =

U Gx = U Gr cos U Gy = U Gr sin
1 3 M0 1 3 / 4 4 G z z (1 + 2 / 4 )2

where SMi, the distributed source term, represents the momentum exchange between vapor and liquid phases expressed three dimensional and will be given in the subsequent section. To evaluate the velocity covariance uiu j for the turbulent flow, the k- model [27] is adopted as it has been used effectively for the simulation of sieve tray [5 -13]. Applying the Boussinesqs hypothetic relation to represent the covariant term, we

d z The coordinate system is shown in Fig. 1. Consequently, the drag force is calculated as follows [29]:

836

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

Figure 1 Coordinate of a sieve hole

Fdragi = CT G (U Gi U i ) U G U hf

(i = x, y, z)

where CT is 0.1. The resisting force for the liquid flow in the x direction is represented by [29]:
2 Fp = Cp LU x hf

where Cp = 0.4. The froth height hf is usually estimated by hf = hL / L , in which the clear liquid height hL for the sieve tray can be calculated by the Zuiderweg correlation [30] as follows:
0.5 0.25 ( hL = 0.6hW p FP / b ) 0.25

where kL and kG are the film coefficients of mass transfer on liquid side and gas (vapor) side respectively, m is the coefficient of distribution between two phases. The value of m is dependent on the concentration of the system concerned. If the concentration change on a tray is not large, the value of m might be taken at the average concentration. However, for the simulation of a multi-tray column, where the change of concentration in the column is appreciable, the value of m should be re-determined for each tray. In the present simulation, the correlations by Nonhebel [31] were used to estimate kL, kG as follows:
0.5 kL = 8DL

in which
U FP = L L U s G where hW is the weir height; b is the weir length per unit bubbling area. The distributed source term SMi is given by: SMx = Fdragx + Fp
0.5

D kG = 0.625kL G DL

0.5

The effective interfacial area a is calculated by [29, 30]: a = a / hf


a = 43 F 0.3
2 Fbba hL FP 0.53

SMi = Fdragi

(i = y, z)

Furthermore, the structure of the tray is an influential factor to the flow distribution. For instance, the flow field is affected by the number, arrangement and diameter of holes on a sieve tray. In this paper, the positioning and the diameter of sieve holes of the simulated column is in accordance with the published layout.
Mass transfer source term The source term Sn in the mass transfer Eq. (7) is given below:

where the froth height is estimated by hf = hL / L . The clear liquid height hL for sieve tray was given by Zeiderweg [30]:
0.5 0.25 hL = 0.6hW p (FP / b)0.25

FP =

U L L Us G

0.5

Sn = K OL a ( C * C )

where hW is the weir height (m); b is the weir length per unit bubbling area (m1).
Boundary conditions At the inlet, the liquid velocity and concentration are uniformly distributed, i.e. U = U in , C = Cin .

where C* is the mass concentration, kgm3, of liquid phase in equilibrium with the vapor passing through the sieve tray; KOL is the overall liquid-phase mass transfer coefficient, which can be expressed by the following relationship : 1 K OL = 1 1 + kL mkG

For the k- equations, the conventional boundary 2 conditions are adopted [32]: kin = 0.003U x in and
3/ 2 in = 0.09kin 0.03

W . 2

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

837

The inlet conditions of c2 - c equations, as presented by Sun et al. [16], are given below:
* 2 = cin 0.082 ( C Cin ) 2

3 EXPERIMENTAL VERIFICATION OF THE PROPOSED MODEL Velocity distribution The simulated velocity distribution on a 1.2 meters in diameter sieve tray reported by Solari and Bell [33] was compared with their experimental measurement. The layout of the reported sieve tray is shown in Fig. 3. They measured the linear velocity of the tracer dye along two lines perpendicular to the liquid flow direction on a plane of 0.038 m above the tray floor.

cin = 0.9

in kin

2 cin

C =0. x At the solid border (tray floor, outlet weir and column wall), the boundary conditions for the mass transfer equation is set to be the concentration flux is equal to zero. The boundary condition of liquid flow is considered to be non-slip, and the conventional logarithm law is employed. At the interface of the vapor and liquid, we set U y U x = 0, = 0 , and U z = 0 . z z At the outlet, we set P = 0 , and
Model computation The simultaneous solution of Eqs. (3) to (11) enables us to obtain the velocity and concentration distributions as well as the turbulent mass transfer diffusivity at once for a sieve tray. The computation was undertaking by the aid of the commercial software FLUENT. The segregated solver and control-volume-based technique were used. To get the convergence, the default of under-relaxation factors and the discretization method are used. The convergence criterions were set to be the residuals less than 103. The selection of proper grid size is essential in the numerical computation. In order to obtain the best simulation, the hexahedral cell is adopted with more grids near the arc area. The simulated results of Ux by taking different numbers of cell to be 22320, 32016 and 58000 are shown in Fig. 2. It can be seen that with number of cells over 30000, stable simulation result can be obtained. Thus the cell number was chosen 32000 for the present computation.

(a) Boundary conditions

(b) Holes arrangement Figure 3 Grid network and sieve hole distribution

Figure 2
22320

Grid size sensitivity of velocity prediction cells; 32016 cells; 58000 cells

Figure 4 shows the liquid velocity Ux predicted by the present model and the experimental data, in which reasonable agreement between them is seen in spite of having some deviations. The discrepancy may be due to the following reason. The experimental measurement is the average linear velocity of the tracer dye from one probe to the neighbor probe, while the present simulation is three-dimensional, and the simulated liquid phase velocities as shown in Fig. 4 are the local velocity components in x direction. Obviously, the comparison between them is only approximate, as it is not exactly on the same basis. The creditability of using pseudo single liquid phase in the present modeling can be further examined by comparing the foregoing simulated results with that using vapor-liquid two-phase model. Gesit et al. [12] simulated the same sieve tray by using two-phase flow model with about 40000 cells. It can be seen in Fig. 4 that the simulated results by the present pseudo single liquid phase model have the same magnitude of accuracy with those by the two-phase model. However, the use of present model is beneficial on less computer time, easier to get convergence and more suitable for the engineering simulation

838

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

(a) Upstream profile


3 3 1

(b) Downstream profile

Figure 4 Comparison between liquid-velocity profiles [QL = 6.9410 m s , FS = 1.464 ms1(kgm3)0.5] experimental; proposed model; gesit et al. model

Figure 5 Schematic setup of the experiment for concentration measurement 1storage tank; 2water pump; 3control valve; 4flow meter; 5downcomer; 6packing; 7sieve tray; 8outlet weir; 9 gas distributor; 10control valve; 11primary control valve; 12blower; 13oxygen cylinder; 14flow meter; 15static mixer

of industrial column with large number of trays.


Concentration distribution As no published data regarding the concentration distribution on a sieve tray are available in the literatures, experimental work was conducted for the purpose of comparison with the model prediction. The experimental installation [34] is shown schematically in Fig. 5. The simulator is a single-pass sieve tray of 1.2 m in diameter with 4.6 mm holes and having 4.6% opening of the tray area. The length of the outlet weirs is 0.79 m. The clearance under the inlet downcomer is 60 mm. The height of the outlet weir is set separately to be 60, 80 or 100 mm. The air rate, ranging from 2600 to 4000 m3h1, was fed to the column by a blower and flow through a distributor to ensure uniform inlet condition. The water from the storage tank at the rate from 10 to 20 m3h1 was pumped to the downcomer after saturated with oxygen in the static mixer. The water was circulated back to the storage tank after flowing through the tray. The oxygen was supplied by an oxygen cylinder. The local concentration of dissolved oxygen in the water was measured by using a measuring probe. The range of measurement is from 0 to 100 mgL1

dissolved oxygen with the accuracy of 0.1 mgL1. The temperature compensation was automatic. The probe was fixed to a slider, which was attached to a truss with cross guide ways on the top of the tray. The probe, submerged in the liquid, could be moved in three directions. The position of the measuring points is shown in Fig. 6. The depth of submergence for the measurement was at 10 and 20 mm above the tray deck. To ensure reliable experimental results, the operation was run until reaching the steady state where the fluctuation of measured concentration was reduced to very small, and the average value was taken as the measuring data. The measurement was point by point with one probe in order to minimize the disturbance to the flow field. Although the concentration data over the whole tray was not taken simultaneously, it is the convenient way to provide an experimental basis to verify the predicted concentration at any point on the sieve tray. As the model prediction is three-dimensional, the planar concentration measurement was conducted at the depth of 10 and 20 mm above the tray deck in order to allow the comparison in three dimensions. As seen from Figs. 7 and 8, the model predictions are reasonably agreed with the experimental measurement in consideration of some inaccuracies involved

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

839

Figure 6 Arrangement of measuring points on the experimental tray

(a) Line I

(b) Line II

(c) Line III Figure 7 Predicted concentration vs. experimental measurement (QL = 17.2 m3h1, G = 4000 m3h1, hW = 100 mm, z = 10 mm) simulated results; experimental data

(d) Line IV

in both simulation and experiment. The obvious discrepancy between the experimental and simulated results is seen in the middle region of Line II, it may be attributed to the fact that this area is around the border between forward and reversed or vortex flow, in which the flowing condition is in transition and appears high fluctuation as found in our experiment.

4 APPLICATION OF CMT MODEL TO THE INDUSTRIAL SCALE COLUMN

The validation of applying the proposed model to the industrial scale column is also verified. Kunesh et al. [25] reported the experimental data for an industrial

840

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

(a) Line I

(b) Line II

(c) Line III Figure 8 Predicted concentration vs. experimental measurement (QL = 17.2 m3h1, G = 4000 m3h1, hW = 100 mm, z = 20 mm) simulated results; experimental data

(d) Line IV

scale column of 1.2192 m (4 ft). in diameter with 6 sieve trays for the stripping of toluene from dilute water solution. They gave the outlet composition and the tray efficiency of each tray under different operating conditions. The simulated outlet concentration expressed in area-weighted average versus tray number for a typical run 16552 is shown in Fig. 9 and compare with the

Figure 9 Simulated outlet concentration vs experimental measurement, Run 16552, QL = 76.3 m3h1, FS = 1.8 ms1(kgm3)0.5 experimental; Sc = 0.7; proposed model

experimental data. Since the concentration concerned is very low, according to the Fenske-Underwood equation under constant relative volatility, a plot of logarithmic concentration versus tray number should yield a straight line. In Fig. 9, both simulated and experimental points are shown on a line and the agreement between them is acceptable in considering the inaccuracy involved in the predictions and the measurements at such low concentrations. The comparison between predicted result and the conventional method of assuming turbulent Schmidt number, Sct, equal to 0.7 is also shown in Fig. 9, the proposed model is seen obviously better than the conventional. The simulated concentration distribution on a sieve tray is given in Fig. 10, in which the non-uniform progress of stripping on the tray is clearly seen. Based on the simulated concentration distribution as shown in Fig. 10, the local tray efficiencies can be obtained. The simulated tray efficiency by area average for run 16552 is 33.4% in comparison with the experimental value of 36%. More simulated tray efficiencies at different mV/L, a self defined constant, are compared with the experimental measurements as shown in Fig. 11, in which reasonable agreement is seen between them. In addition, a feature of the present model is able

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

841

conditions on the tray. In practice, the mass transfer diffusivity is expressed macroscopically by the volume average. For instance, the predicted volume average value of Dt for three runs under different operations were 0.035, 0.030 and 0.021 m2s1 respectively, which are within the range of the experimental measurements from about 0.01 to 0.09 m2s1 in the same sieve tray column as reported by Cai and Chen [32].
5 PREDICTION OF TRAY EFFICIENCY FOR DIFFERENT TRAY STRUCTURE
Figure 10 Similation of concentration distribution, Run 16552, QL = 76.3 m3h1, FS = 1.8 ms1(kgm3)0.5, z = 20 mm

Figure 11 Simulated and experimental tray efficiencies under different conditions, QL = 76.3 m3h1, FS = 1.8 ms1(kgm3)0.5 experimental data; simulated results

to predict the turbulent mass transfer diffusivity Dt, which is commonly regarded as representing the extent of back-mixing and also is an influential factor to the tray efficiency. As an example, the simulated distribution of Dt across the tray for run 16552 is shown in Fig. 12. The diverse distribution of Dt is chiefly due to the complicated non-uniform flow and mass transfer

By means of the simulated concentration distribution on a tray, the tray efficiency can be evaluated for different tray structure. Sun [15] simulated a sieve tray distillation column of 1.2 m in diameter, reported by Sakata and Yanagi [18] for separating cyclohexane (C6) and n-heptane (n-C7) mixture with different tray structure, including sieve hole arrangement, heights of inlet weir and outlet weir. As an example of illustration, we can compare the tray efficiency with different height of outlet weir, for instance, at 20mm and 100mm respectively. The simulated concentration distributions are correspondingly shown in Fig. 13. The inlet concentration of C6 to both trays was 0.482 mole fraction, while the simulated outlet concentrations for the case of outlet weir height hW equal to 20 mm and 100 mm were found to be 0.393 and 0.383 respectively. More stripping of C6 on the hW = 100 tray may be due to deeper liquid layer resulting more interacting time between vapor and liquid and therefore enhance the mass transfer. The Murphree tray efficiencies obtained were 86.7% and 89.5% respectively. The simulated Dt for both cases are shown in Fig. 14. It can be seen that their distribution is somewhat different. However, such simulated results do not mean that higher outlet weir is a good choice, as in this case several drawbacks are also appeared, such as higher pressure drop etc. However, the feasibility of evaluating the influence of tray structure on tray efficiency by using present model may be interested to

Figure 12

Distribution of turbulent mass transfer diffusivity, Run 16552, QL = 76.3 m3h1, FS = 1.8 ms1(kgm3)0.5, z = 20 mm

842

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

(a) hW = 20 mm

(b) hW = 100 mm Figure 13 Simulated concentration profile of a sieve trays (xin = 0.482, QL = 30.66 m3h1, G = 5.75 kgs1, P = 165 kPa, z = 20 mm, total reflux)

(a) hW = 20 mm

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011

843

(b) hW = 100 mm Figure 14 Simulated turbulent mass transfer diffusivity profile of sieve trays (xin = 0.482, QL = 30.66 m3h1, P = 165 kPa, z = 20 mm, total reflux)

the designers for the optimization of tray structure.


6 CONCLUSIONS

Engineering (Tianjin University).


NOMENCLATURE
a a b C C* Cp CT Ct, Cc1, Cc2 C, C1, C2 c c 2 D Dh Dt d F Fbba Fdrag FP Fp hf hL hW KOL k kG kL L M0 m P p QL effective vapor-liquid interfacial area, m2m3 interfacial area per unit bubbling area weir length per unit bubbling area, m1 time average liquid phase mass concentration, kgm3 time average liquid phase mass concentration in equilibrium with vapor phase, kgm3 constant in resisting force correlation for the liquid flow in the x direction constant drag force correlation turbulence model constants for c2 - c model turbulence model constants for k- model fluctuating mass concentration, kgm3 concentration variance, kg2m6 molecular mass transfer diffusivity, m2s1 diameter of the sieve hole, m turbulent mass transfer diffusivity, m2s1 distance between the center of the sieve hole and the designated point in the liquid, m F factor, F = U s G vapor F-factor on bubbling area, (ms1) (kgm3)0.5 drag force, kgm2s2 flow parameter resisting force, kgm2s2 froth height, m clear liquid height, m weir height, m overall liquid-phase mass transfer coefficient, ms1 turbulent kinetic energy, m2s2 gas-phase mass transfer coefficient, ms1 liquid-phase mass transfer coefficient, ms1 molar liquid flowrate, kmols1 momentum of gas phase from sieve hole, kgms1 distribution coefficient time average pressure, kgm1s2 pitch of holes in sieve tray, m liquid rate, m3h1

(1) An improved computational mass transfer model with distributed source term is presented for simulating distillation sieve tray column. The present model is able to predict the concentration distribution on a tray and the turbulent mass transfer diffusivity as well as the Murphree tray efficiency. (2) For improving the simulation, the vapor velocity flowing out from the sieve hole is considered to be three dimensional for better evaluation of the vapor-liquid interaction. (3) The predicted concentration distribution was confirmed by the experimental measurement on a sieve tray simulator of 1.2 m in diameter undertaking desorption of dissolved oxygen in the feed water by blowing air. (4) The validation of applying the present model to the industrial scale column is examined by simulating a distillation column of 1.2192 m (4 ft). in diameter with 6 sieve trays for the stripping of toluene from water solution. The simulated outlet concentration and efficiency of each tray is in agreement with the experimental data reported in the literature. (5) The predicted turbulent mass transfer diffusivity for the sieve tray is within the range of the published experimental result. (6) The influence of tray structure on tray efficiency can be evaluated based on the concentration distribution by using the present model.
ACKNOWLEDGEMENTS

The authors wish to acknowledge the assistance by the staffs in the State Key Laboratories for Chemical

844

Chin. J. Chem. Eng., Vol. 19, No. 5, October 2011


source term of inter phase momentum transfer, ms2 time average source term of mass transfer, kgm3s1 time, s time average liquid velocity, ms1 time average liquid velocity vector, ms1 time average gas velocity vector, ms1 time average gas velocity leaving the sieve hole, ms1 time average liquid velocity on bubbling area, ms1 time average radial velocity, ms1 time average superficial vapor velocity, ms1 fluctuating velocity, ms1 molar gas flow rate, kmols1 outlet weir width, ms1 liquid volume fraction Kronecker delta turbulent dissipation, m2s3 dissipation rate of c2 , kg2m6s1 time average liquid velocity vector, ms1 angle, () turbulent viscosity, m2s1 density, kgm3 surface tension, kgs2 turbulence model constants for c2 , c , k, gas tensor coordinates inlet liquid radial direction Cartesian coordinates 143-151 (2000). Liu, C.J., Yuan, X.G., Yu, K.T., Zhu, X.J., A fluid-dynamics model for flow pattern on a distillation tray, Chemical Engineering Science, 55, 2287-2294 (2000). Gesit, G., Nandakumar, K., Chuang, K.T., CFD modeling of flow patterns and hydraulics of commercial-scale sieve trays, AIChE J., 49, 910-924 (2003). Wang, X.L., Liu, C.J., Yuan, X.G., Yu, K.T., Computational fluid dynamics simulation of three-dimensional liquid flow and mass transfer on distillation column trays, Industrial and Engineering Chemistry Research, 43, 2556-2567 (2004). Liu, B.T., Study of a new mass transfer model of CFD and its application on distillation tray, PhD Thesis, Tianjin university, Tianjin (2003). Sun, Z.M., Study on computational mass transfer in chemical engineering, PhD Thesis, Tianjin university, Tianjin (2006). Sun, Z.M., Liu, B.T., Yuan, X.G., Liu, C.J., Yu, K.T., New turbulent model for computational mass transfer and its application to a commercial-scale distillation column, Industrial & Engineering Chemistry Research, 44, 4427-4434 (2005). Sun, Z.M., Yu, K.T., Yuan, X.G., Liu, C.J., A modified model of computational mass transfer for distillation column, Chemical Engineering Science, 62, 1839-1850 (2007). Sakata, M., Yanagi, T., In performance of a commercial scale sieve tray, In: Institution of Chemical Engineers Symposium series, 56, IChemE, USA (1979). Garcia, J.A., Fair, J.R., A fundamental model for the prediction of distillation sieve tray efficiency. 1. Database development, Industrial and Engineering Chemistry Research, 39, 1809-1817 (2000). Liu, G.B., Yu, K.T., Yuan, X.G., Liu, C.J., Guo, Q.C., Simulations of chemical absorption in pilot-scale and industrial-scale packed columns by computational mass transfer, Chemical Engeering Science, 61, 6511-6529 (2006). Liu, G.B., Yu, K.T., Yuan, X.G., Liu, C.J., New model for turbulent mass transfer and its application to the simulations of a pilot-scale randomly packed column for CO2-NaOH chemical absorption, Industrial and Engineering Chemistry Research, 45, 3220-3229 (2006). Liu, G.B., Yu, K.T., Yuan, X.G., Liu, C.J., A computational transport model for wall-cooled catalytic reactor, Industrial and Engineering Chemistry Research, 47, 2656-2665 (2008). Liu, G.B., Yu, K.T., Yuan, X.G., Liu, C.J., A numerical method for predicting the performance of a randomly packed distillation column, International Journal of Heat and Mass Transfer, 52, 5330-5338 (2009). Li, W.B., Liu, B.T., Yu, K.T., Yuan, X.G., A rigorous model for the simulation of gas adsorption and its verification, Industrial and Engineering Chemistry Research, 50 (13), 8361-8370 (2011). Kunesh, J.G., Ognisty, T.P., Sakata, M., Chen, G.X., Sieve tray performances for steam stripping toluene from water in a 4-ft diameter column, Industrial and Engineering Chemistry Research, 35, 2660-2671 (1996). Bennett, D.L., Rakesh, A., Cook, P.J., New pressure drop correlation for sieve tray distillation columns, AIChE J., 29, 434-442 (1983). Launder, B.E., Spalding, D.B., The numerical computation of turbulent flows, Compt Methods Appl Mech Eng, (3), 269-289 (1974). Dai, G.C., Chen, M.H., Fluid Dynamics in Chemical Engineering, Chemical Industry Press, Beijing (1988). Leboeuf, F., Huang, G.P., Kulisa, P., Perrin, G., Model and computation of discrete jets in crossflow, European Journal of Mechanics, B/Fluids, 10, 629-650 (1991). Zuiderweg, F.J., Sieve traya view on the state of the art, Chemical Engineering Science, 37, 1441-1464 (1982). Nonhebel, G., Purification Processes for Air Pollution Control, Butterworth & Co., London (1972). Cai, T.J., Chen, G.X., Liquid back-mixing on distillation trays, Industrial and Engineering Chemistry Research, 43, 2590-2597 (2004). Solari, R.B., Bell, R.L., Fluid flow patterns and velocity distribution on commercial-scale sieve trays. AIChE J., 32, 640-649 (1986). Sun, Z.M., Liu, C.J., Yuan, X.G., Yu, K.T., Measurement and numerical simulation of concentration distribution on sieve tray, J. Chemical Industry and Engineering (China), 57 (8), 1877-1883 (2006). (in Chinese)

SMi Sn t U U UG Uh UL Ur Us u V W L ij

11 12 13

14 15 16

17 18 19 20

c , , k ,
2

Subscripts
G i, j, k in L R x, y, z

21

REFERENCES
1 2 3 Bell R.L., Residence time and fluid-mixing on commercial scale sieve trays, AIChE J., 18, 498-505 (1972). Bell R.L., Solari, R.B., Effect of nonuniform velocity fields and retrograde flow on distillation tray efficiency, AIChE J., 20, 688-695 (1974). Yu, K.T., Huang, J., Simulation and efficiency of large tray (i) eddy diffusion model with non-uniform liquid velocity field, J. Chemical Industry and Engineering (China), 32, 11-19 (1981). (in Chinese) Yu, K.T., Some progress of distillation research and industrial applications in China, In: Distillation and Absorption92, Birmingham, England, A139-A166 (1992). Zhang, M.Q., Yu, K.T., Simulation of two dimensional liquid phase flow on a distillation tray, Chin. J. Chem. Eng., 2, 63-71 (1994). Yuan, X.G., You, X.Y., Yu, K.T., Velocity field simulation of gas-liquid two-phase flow on sieve tray, J. Chemical Industry and Engineering (China), 46, 511-515 (1995). Yu, K.T., Yuan, X.G., You, X.Y., Liu, C.J., Computational fluid-dynamics and experimental verification of two-phase two-dimension flow on a sieve column tray, Chemical Engineering Research and Design, Transactions of the Institute of Chemical Engineers, Part A, 77, 554-560 (1999). Mehta, B., Chuang, K.T., Nandakumar, K., Model for liquid phase flow on sieve trays, Chemical Engineering Research and Design, Transactions of the Institute of Chemical Engineers, Part A, 76, 843-848 (1998). Krishna, R., van Baten, J.M., Ellenberger, J., Higler, A.P., Taylor, R., CFD simulations of sieve tray hydrodynamics, Chemical Engineering Research and Design, Transactions of the Institute of Chemical Engineers, Part A, 77, 639-646 (1999). van Baten, J.M., Krishna, R., Modelling sieve tray hydraulics using computational fluid dynamics, Chemical Engineering Journal, 77,

22 23

24 25

4 5 6 7

26 27 28 29 30 31 32 33 34

10

You might also like