Вы находитесь на странице: 1из 8

Chemical Engineering and Processing 45 (2006) 747754

Experiment and prediction of breakthrough curves for packed


bed adsorption of water vapor on cornmeal
Hua Chang, Xi-Gang Yuan , Hua Tian, Ai-Wu Zeng
State Key Laboratory of Chemical Engineering, Chemical Engineering Research Center, School of Chemical Engineering and Technology,
Tianjin University, Tianjin 300072, PR China
Received 16 November 2005; accepted 5 March 2006
Available online 10 March 2006

Abstract
The adsorption isotherms of water vapor on cornmeal and the breakthrough curves at 82100 C were measured in a fixed-bed apparatus for
ethanol dehydration. Using the water isotherms measured and fitting the experimental data to fixed-bed model for breakthrough curve, the effective
diffusivity of water was obtained. The effective diffusivity was estimated and used to predict breakthrough curves at other adsorption conditions.
The controlling factor for mass-transfer resistance was discussed.
2006 Elsevier B.V. All rights reserved.

Keywords: Adsorption isotherm; Effective diffusivity; Ethanol dehydration; Breakthrough curve; Adsorption model

1. Introduction tions in drying air using PSA (pressure swing adsorption) can
also be found recently [1720,22].
Starchy materials can adsorb and remove water from alcohol A number of studies have been published concerning the
vapors to dry fuel grade ethanol in an energy-efficient man- adsorption behavior of water vapor on starchy material, includ-
ner [14]. The adsorption and desorption of water on starchy ing the measurement of equilibrium isotherms and breakthrough
materials, especially, i.e. cornmeal, corn grits and pure starch, curves [2,4,8,13,16]. However, there has been little attempt to
for ethanol dehydration have been extensively studied [116]. correlate them with equilibrium and kinetic information except
The success of this dehydration method seems related to the work of Hills and Pirzada [23] for prediction of breakthrough
differences in the rate of adsorption, as well as differences in curves on steamed and flaked cornmeal. Our aim is to measure
the strength of interaction between each species and the adsor- the water isotherms in bench-scale apparatus and apply it to
bent. Compounds which exhibit either weak interactions or slow process model for breakthrough curve prediction. By fitting the
rates of adsorption are expected to be readily separated from experimental data to model of breakthrough curve, the effective
those that exhibit strong, relatively fast interactions with the diffusivity of water was estimated and validated by predicting
adsorbent [13,17]. Due to strong polar attraction between water breakthrough curves at some other operation conditions.
molecules and the hydroxyl groups of the adsorbent, water can
adsorb on the adsorbent faster and stronger than ethanol, which 2. Experimental
is known as major mechanism for the selective adsorption of
water [9,13]. Corn grits have been used in large-scale fermenta- The corn meals used as adsorbents are from Wuqing County
tion alcohol plants producing fuel-grade ethanol [1821]. Works in north China, with a granularity of <0.45 mm. All the adsor-
on modified-starch material as desiccant for industrial applica- bents were dried for 8 h in air-convection oven at 105 C. One
hundred and twenty grams of dried cornmeal were used in
all runs. To provide vapor feeds with different concentrations,
waterethanol mixtures with different concentrations were pre-
Abbreviations: ads., adsorbent; concentr., concentration; Temp., tempera-
ture
pared by mixing anhydrous ethanol and water.
Corresponding author. Tel.: +86 22 27404732; fax: +86 22 27404496. Fig. 1 is the experimental apparatus used in this work. In
E-mail address: yuanxg@tju.edu.cn (X.-G. Yuan). Fig. 1a, a transparent column of 0.9 m long and 0.025 m i.d.

0255-2701/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.cep.2006.03.001
748 H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754

Fig. 1. Diagram of experimental apparatus for fixed-bed adsorption: (1) adsorber; (2) adsorbent; (3) support packing; (4) oil bath; (5) kettle and electro-heater; (6)
and (7) condenser; (8) voltage controller; (9) gas blowout; (10) cold trap; (11) cooling water; (12) rotameter.

with oil bath jacket was used as adsorber and packed with corn- the desorbate to correct the experimental data, assuming that
meal, which is supported from the bottom and covered from the ideal gas condition holds. Note that we assumed that quan-
the top by a layer of random packing, respectively. Thermo- tities of water and ethanol sorbed on the inner surface of the
points are arranged along the axial direction of the column with shells of the column and tubes and on the support packing are
intervals of 0.1 m to measure the temperatures of the bed at the negligible compared to those collected in the cold trap. With the
corresponding positions. The kettle was heated by electro-heater assumption that the adsorbed ethanol and water at the first phase
to provide the ethanolwater vapor feed. The superficial veloc- almost completely desorbed at the second phase, the equilibrium
ity of vapor in the adsorber was controlled through regulating amount of water and ethanol adsorbed can be determined by the
the voltage on the electro-heater with a voltage controller. The quantity and the concentration of the desorbate.
temperatures were monitored by multi-road heat resistance con- The pipelines the vapor went through were heated by heat
tact thermometer, the compositions of samples were analyzed tape to avoid condensing and the pipelines above the adsorber
by HP5890 gas chromatogram workstation, and the mass of the were kept at little high temperature to prevent condensation back
material was weighed by Mettler AE163 electrical scale with the to adsorber.
precision of 0.0001 g. The temperature of the adsorber was kept
well above the dew point of the vapor feed to avoid condensing. 3. Models considered in this work
The experiment was divided into two steps, the adsorption
step (Fig. 1a) and the desorption step (Fig. 1b). Using the appa- 3.1. Linear adsorption isotherm
ratus of Fig. 1a, 120 g adsorbent was packed in the adsorber and
the vapor feed went through the adsorber and the ethanol rich Assuming linear adsorption according to Henrys law for
product was collected at the outlet end of condenser 7. The adsor- dilute concentration of water, the adsorption isotherm for water
ber was operated for 300 min and the breakthrough curves can can be correlated as a linear adsorption isotherm, which is given
be obtained by measuring the concentration of outlet solution at by
intervals of set time.
q = Kc (1)
When the adsorption operation had finished, the experiment
passed to step 2 where the adsorber was flushed by 106 C nitro- where K is the adsorption equilibrium constant for a linear
gen gas for 240 min using the same apparatus but configured as adsorption isotherm and q denotes the mass of water adsorbed
Fig. 1b. The desorbed material was collected in the cold trap that in mol m3 adsorbent and c is the concentration of water in
was cooled by liquid nitrogen, weighed by Mettler AE163 and mol m3 .
analyzed by gas chromatogram. At the moment when the adsorp-
tion operation stopped, the void spaces in the system including 3.2. Model for breakthrough curves
those within the bed and in the pipelines were supposed to be
full of vapor feed, which could be flushed to the cold trap during The aim of the experiments was to attempt to predict mass-
the desorption operation, and then, should be subtracted from transfer diffusivity and breakthrough curves from the measured
H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754 749

adsorption isotherm. In the case of isothermal adsorption, the where = (kKZ/u) ((1 b )/b ) is the dimensionless distance
differential equations can be written as follows [24]. coordinate, = k(t(Z/u)) the dimensionless time coordinate,
and erf(x) is the error function that is defined as
3.2.1. Mass balance on water on an element of column
erf(x) = erf(x) (7)
A mass balance on the solute for the flow of fluid through  x
a differential adsorption-bed length, dZ over a differential time 2
e d
2
erf(x) = (8)
duration, dt, gives: 0
2 c (uc) c (1 b ) q and coordinate transformations for Z and t, convert the equa-
DL + + + =0 (2) tions to a much simpler form of erf(x). The approximation given
Z2 Z t b t
by Eq. (6) is known to be acceptable and the error could be
where the first term accounts for axial dispersion with eddy dif- within 0.6% for > 2.0.
fusivity DL , the second term permits an axial variation in fluid Klinkenberg model [24,25] has also included the following
velocity, and the fourth term that is based on q, the mass-average approximate solution for the profiles of solute concentration in
adsorbate loading per unit mass accounts for the variation of q equilibrium with the average sorbent loading:
throughout the adsorbent particle. Eq. (2) gives the concentra-   
c q 1  1 1
tion of solute in the bulk fluid as a function of time and location = 1 + erf (9)
in the bed. cF qF 2 8 8
where c = q/K and c /cF = q/qF , where qF is the loading in
3.2.2. Linear driving force model for mass transfer [24] equilibrium with cF .
q
= k(q q) = kK(c c ) (3) The simple model of Klinkenberg, giving c/cF as a function
t of dimensionless time , and dimensionless bed length , was
where q* is the adsorbate loading in equilibrium with the solute not worse than more sophisticated models, and could be known
concentration c, in the bulk fluid. c* is the concentration in equi- as adequate for preliminary design purpose [26].
librium with average loading q; k is the overall mass-transfer The equilibrium constant K for a given temperature can be
coefficient in s1 , which includes both external and internal correlated by experimental data. By fitting the experimental
transport resistances; and K is the adsorption equilibrium con- breakthrough curves to Eq. (6), the overall mass-transfer coeffi-
stant for a linear adsorption isotherm in Eq. (1). cient k can be estimated. Then by Eqs. (4) and (5), the effective
diffusivity De can be evaluated. In addition, the breakthrough
3.2.3. Relationship for the factor Kk [24] curves at various conditions can be predicted by De which has
R2p been obtained.
1 Rp
= + (4)
kK 3kc 15De 4. Results and discussion
where kc is the external mass-transfer coefficient in m/s, De is
effective diffusivity in m2 /s and Rp is adsorbent particle radius All the results of the experiment were listed in Table 1. From
in m. The first term in Eq. (4) is the overall mass-transfer resis- the data in Table 1, considerable ethanol was also adsorbed on
tances, the second and third term is external one and internal the cornmeal in all runs, which must be recovered by recycling
one, respectively. to a distillation stage if it is used for the separation process.

3.2.4. The correlation for external mass-transfer coefcient 4.1. Linear model of adsorption isotherm for water
The external transport coefficient of particles in fixed-bed can
be correlated by [24]: The adsorption isotherms of water are shown in Fig. 2. By
fitting the experimental results of adsorption isotherm to the lin-
1/3
Sh = 2 + 1.1Re0.6 Sci (5) ear isotherm represented by Eq. (1), the adsorption equilibrium
constant at different temperatures was correlated and listed in
where Sh = Shrewoodnumber = Kc Dp /Di , Re = Reynolds num-
Table 2. From Fig. 2, it can be seen that the linear correlation
ber = Dp G/, and Sci = Schmidt number = /Di .
represent the isotherm fairly well with the value of the correla-
By Eq. (5), external mass-transfer coefficient kc can be esti-
tion coefficient R2 as 0.9666 for 91 C at dilute concentration of
mated from Sh.
water. The linear adsorption isotherms for water were used in the
The analytical solution of a simplified form of Eq. (2), in
prediction of breakthrough curves in after mentioned section.
which negligible axial dispersion, constant fluid velocity u, and
the linear driving force mass-transfer model were assumed, was 4.2. Prediction of effective diffusivity
summarized by Ruthven and discussed in detail by Klinkenberg
[25]. An adopted approximate solution is that of Klinkenberg By least square correlation:
[24]:
 2
 

   ci   ci 
c 1 1 1 min f (k) = (10)
1 + erf + + (6) cF cF experimental result
cF 2 8 8
750 H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754

Water adsorption capacity


q (mol H2 O/m3 ads.)

888.8398
1938.1697
817.5243
1318.8561
1374.6437
2118.8588
2636.8131
637.1892
1219.7844
capacity (g/g ads. 102 )
ethanol adsorption

Fig. 2. Adsorption isotherm for water and its linear model (solid line).
7.47
7.01
3.01
3.18
2.52
2.73
2.77
1.56
1.59
Water adsorption capacity

A fitted value of parameter k can be obtained as (ci /cF ) in Eq.


(6) is only a function of overall mass-transfer coefficient k at
(g/g ads. 102 )

the condition that other variables such as bed depth Z, equilib-


rium isotherm constant K, the bed void fraction b , and constant
velocity u, are known. The error for the model prediction with
1.89
4.11
1.74
2.80
2.92
4.50
5.60
1.35
2.59

optimum k can be expressed by


n
The relative humidity


((ci /cF ) (ci /cF )experimental result )2
of ethanol, Pi /Pis


i=1
Sy = (11)
n1
0.858
0.764
0.596
0.563
0.547
0.504
0.477
0.446
0.410

where Sy is standard deviation for the prediction to experimen-


tal result. The smaller the value of Sy , the better the effect of
The relative humidity

prediction.
Changing the value of k to minimize Eq. (10), the opti-
of water, Pi /Pis

mal k is obtained as 2.7813 103 s1 with Sy being 0.101


for the operation condition of bed temperature of 91 C, bed
0.133
0.294
0.091
0.163
0.208
0.304
0.381
0.065
0.153

depth 43 cm, superficial velocity of 1.64 cm/s and 6.2 wt.%


water concentration feed. The comparison of correlation break-
through curve and experimental one is shown in Fig. 3. The
Water concentration,

deviation from zero for experimental result at low can be


c (mol H2 O/m3 )

Note: temp.: temperature; ads.: adsorbent; concentr.: concentration.

explained by the measurement error from the humidity of


air. Using the same method, the overall mass-transfer coeffi-
cient k at different temperature can be estimated, and it was
2.3169
5.1122
2.1978
3.9232
4.9922
7.3120
9.1567
2.1142
5.0116

found that the values so obtained for k almost keeps con-


stant at 2.7813 103 s1 , with Sy being 0.057 and 0.086
for 82 and 100 C, respectively. The breakthrough curves for
Vapor feed concentr. of

the other two temperatures of 82 and 100 C are shown in


Figs. 4 and 5.
ethanol (%)

97.4
93.8
97.4
95.2
93.8
90.5
87.8
97.5
93.8
Run conditions and results

Table 2
Adsorption equilibrium constant in linear adsorption isotherms for water on
Bed temp.

cornmeal
Temperature ( C)
( C)

K
82
82
91
91
91
91
91
100
100

82 379.89
Run no.
Table 1

91 293.07
100 252.15
1
2
3
4
5
6
7
8
9
H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754 751

Fig. 6. Prediction of breakthrough curves for different bed depths (solid line)
and comparison with experimental results (points) (bed temperature of 91 C,
Fig. 3. Comparison of breakthrough curves between experimental results (() superficial velocity is 1.60 cm/s, water concentration feed of 6.2 wt.%).
points) and Klinkenbergs model (solid line) (bed temperature of 91 C, bed
depth 43 cm, superficial velocity of 1.64 cm/s, water concentration feed of
6.2 wt.%). 4.3. Predictions of breakthrough curves and comparison
with experimental results

Generally, Klinkenbergs [24] models, Eqs. (6) and (9), could


be used respectively to predict the breakthrough curves and pro-
files of solute concentration in equilibrium with the average
sorbent loading for different bed depth at same other operational
conditions. In this paper, we try to predict the breakthrough
curves for different bed depth, as well as different velocity
and different water vapor concentration by using the value of
k obtained in the foregoing section.
The breakthrough curve predictions for different bed depths
and the comparison with experimental data are shown in Fig. 6.
From Fig. 6, it can be seen that with the value of k as
2.7813 103 s1 , the breakthrough curves for similar velocity
but different bed depth can be well predicted with the calcu-
lated Sy from Eq. (11) being 0.052, 0.046, 0.045 and 0.051 for
Fig. 4. Comparison of breakthrough curves between experimental results (()
points) and Klinkenbergs model (solid line) (bed temperature of 82 C, bed
22, 28, 43 and 46 cm bed depths, respectively. This indicates that
depth 43 cm, superficial velocity of 1.26 cm/s, water concentration feed of Klinkenbergs [24] model is quite adequate for our experimental
6.2 wt.%). conditions, and that it is valuable for industrial applications. The
profiles of solute concentrations in equilibrium with the average
sorbent loading for different bed depths are shown in Fig. 7.
The shapes of the profiles of solute concentration in the bed
for different bed depths are similar with those of breakthrough
curves.
The breakthrough curve for large superficial velocity of
4.31 cm/s at 91 C was predicted in Fig. 8, and different water
concentration of vapor feed in Fig. 9. As seen from Fig. 8,
the model of Klinkenberg [24] gives a good fit to the experi-
mental points with Sy of 0.060 for the same concentration of
vapor feed but a different superficial velocity. Thus, in spite
of the sweeping assumption involved in Klinkenbergs [24]
model, direct measurement on a small sample can be used to
predict the breakthrough curves of a larger adsorber by means
of the numerical solution of the model. However, from Fig. 9,
Klinkenbergs [24] model cannot give perfect fit for different
Fig. 5. Comparison of breakthrough curves between experimental results (()
points) and Klinkenbergs model (solid line) (bed temperature of 100 C, bed vapor concentration. The corresponding values of Sy are 0.141,
depth 43 cm, superficial velocity of 1.70 cm/s, water concentration feed of 0.143, and 0.242 for 9.5, 12, and 20 wt.% water concentration
6.2 wt.%). feed, respectively. This indicates that with the water concentra-
752 H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754

Fig. 7. Prediction of profiles of solute concentration in equilibrium with the Fig. 10. Comparison of breakthrough curves between experimental results (()
average sorbent loading for different bed depths (bed temperature of 91 C, points) and Klinkenbergs model (solid line) with the updated k value (bed
superficial velocity is 1.60 cm/s, water concentration feed of 6.2 wt.%). temperature of 91 C, bed depth 43 cm, superficial velocity of 1.76 cm/s, water
concentration feed of 20 wt.%).

tion increases, the prediction is getting worse. If we assumed


that Sy larger than 0.11 is unacceptable, it can be concluded
that the value of k as 2.7813 103 s1 cannot be applied to
predict the breakthrough curves for different vapor concentra-
tions. A larger slope in the beginning of S-shape breakthrough
curves suggests a much larger mass-transfer rate than that in
the end. While at the end of the breakthrough curves, the small
slope in the end of S-shape breakthrough curves indicates that
adsorption rate becomes very small when adsorbent approach-
ing to equilibrium. In fact, the breakthrough curve prediction
from Klinkenberg model [24] is symmetrical. Even with an
updated value of k, which was obtained as 1.1125 103 s1
by re-minimize the criterion of Eq. (10) with the experimen-
Fig. 8. Prediction of breakthrough curves for large velocity (solid line) and tal results for 20 wt.% water concentration feed, the value of
comparison with experimental results (() points) (bed temperature of 91 C, Sy was as high as 0.177. The key here is that, the break-
bed depth 43 cm, superficial velocity of 4.31 cm/s, water concentration feed of through curve demonstrated under relative higher water con-
6.2 wt.%). centration are becoming far from being symmetric and cannot
be predicted well with Klinkenberg model [24], as seen in
Fig. 10.

4.4. Analysis of mass-transfer resistance

According to Eq. (5), the external mass-transfer coefficient


for different operation conditions can be estimated. The esti-
mation results are listed in Table 3. From Table 3, the overall
mass-transfer resistance 1/kK increases with the increasing of
temperature, and the internal mass-transfer resistance is con-
trolling factor due to (R2p /15De ) (1/kK)  (Rp /3kc ) even
for large velocity at the end of breakthrough curves. This is con-
sistent with the fact that the overall mass-transfer coefficient
obtained at low superficial velocity can be used to predict well
the breakthrough curves for large superficial velocity, that is, the
overall mass-transfer coefficient is similar for both low velocity
Fig. 9. Prediction of breakthrough curves for different water concentrations
and high velocity due to the fact that internal mass-transfer is
(solid line) and comparison with experimental results (() points) (bed temper- controlling factor. As a result, the estimated value for De is in
ature of 91 C, bed depth 43 cm, superficial velocity of 1.80 cm/s). the order of 2.5 109 m2 s1 .
H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754 753

Table 3
Values of parameters for different operation conditions

T ( C) K k ( 103 s1 ) u0 ( 102 m/s) b u ( 102 m/s) Rp ( 106 m) (kg/m3 ) ( 105 Pa s) Di ( 105 m2 /s)

82 379.89 2.7813 1.26 0.3 4.20 225 1.486 1.085 1.642


91 293.07 2.7813 1.64 0.3 5.47 225 1.443 1.111 1.736
91 293.07 2.7813 4.31 0.3 14.37 225 1.443 1.111 1.736
100 252.15 2.7813 1.70 0.3 5.67 225 1.421 1.137 1.809

T ( C) Re Sci Sh kc (m/s) 1/kK Rp /3kc ( 104 ) R2p /15De De ( 109 m2 /s)

82 2.589 0.445 3.486 0.127 0.946 5.898 0.946 3.568


91 3.194 0.444 3.684 0.142 1.227 5.278 1.226 2.752
91 8.395 0.444 5.007 0.193 1.227 3.883 1.226 2.752
100 3.186 0.443 3.680 0.148 1.426 5.071 1.425 2.368

5. Conclusions Re Reynolds number = Dp G/


Rp radius of adsorbent (m)
The adsorption isotherms for water at 82100 C was mea- Sy standard deviation for the prediction
sured in a fixed-bed apparatus. The water isotherms were lin- Sci Schmidt number = /Di
earized and applied to the prediction of breakthrough curves. Sh Sherwood number = kc Dp /Di
By fitting the experimental results of breakthrough curves to the u effective velocity of vapor (m/s)
model of Klinkenberg, the overall mass-transfer coefficient was u0 superficial velocity of vapor (m/s)
estimated as 2.7813 103 s1 and successfully used for the
prediction of breakthrough curves at different superficial veloc- Greek letters
ity and different bed depth, but cannot predict well the break- viscosity of vapor (Pa s)
through curves for different vapor concentration. The simple dimensionless distance coordinate
model of Klinkenberg, giving c/cF as a function of dimension- density of vapor (kg m3 )
less time , and dimensionless bed length , was not worse than dimensionless time coordinate corrected for displace-
the more sophisticated model, and could be known as adequate ment
for preliminary design purpose. The analysis of mass-transfer b bed void fraction
resistance indicated that water adsorption on cornmeal was con-
trolled by the internal mass-transfer resistance for both the low
and high velocity at the end of the breakthrough curves. References

[1] M.R. Ladisch, K. Dyck, Dehydration of ethanol: new approach gives


Appendix A. Nomenclature positive energy balance, Science 205 (1979) 898900.
[2] M.R. Ladisch, M. Voloch, J. Hong, P. Bienkowski, G.T. Tsao, Cornmeal
adsorber for dehydrating ethanol vapors, Ind. Eng. Chem. Process Des.
c concentration of water (mol m3 ) Dev. 23 (3) (1984) 437443.
cF concentration of water in feed vapor (mol m3 ) [3] P.R. Bienkowski, A. Barthe, M. Voloch, R.N. Neuman, M.R. Ladisch,
Breakthrough behavior of 17.5 mol% water in methanol, ethanol, iso-
c* concentration in equilibrium with average loading q
propanol, and T-butanol vapors passed over corn grits, Biotechnol.
(mol m3 ) Bioeng. 28 (7) (1986) 960964.
Di molecular diffusivity (m2 s1 ) [4] A.A. Hassaballah, J.H. Hills, Drying of ethanol vapors by adsorption on
Dp diameter of adsorbent (m) corn meal, Biotechnol. Bioeng. 35 (6) (1990) 598608.
erf(x) error function [5] G.H. Robertson, L.R. Doyle, A.E. Pavlath, Intensive use of biomass
G mass velocity of vapor (kg m2 s1 ) feedstock in ethanol conversion: the alcoholwater, vapor-phase separa-
tion, Biotechnol. Bioeng. 25 (12) (1983) 31333148.
k overall mass-transfer coefficient (s1 ) [6] J. Hong, M. Voloch, M.R. Ladisch, G.T. Tsao, Adsorption of
kc external mass-transfer coefficient (m/s) ethanolwater mixtures by biomass materials, Biotechnol. Bioeng. 24
K adsorption equilibrium constant for water in a linear (3) (1982) 725730.
adsorption isotherm [7] V. Rebar, E.R. Fischbach, D. Apostolopoulos, J.L. Kokini, Thermo-
dynamics of water and ethanol adsorption on four starches as model
Pi partial pressure for component i at the adsorption tem-
biomass separation systems, Biotechnol. Bioeng. 26 (5) (1984) 513517.
perature (Pa) [8] R. Neuman, M. Voloch, P. Bienkowski, M.R. Ladisch, Water sorption
Pis vapor pressure for component i at the adsorption tem- properties of a polysaccharide adsorbent, Ind. Eng. Chem. Fundam. 25
perature (Pa) (1986) 422425.
Pi /Pis the relative humidity for component i [9] J.Y. Lee, M.R. Ladisch, Polysaccharides as Adsorbents: an Update on
q water adsorption capacity (mol m3 ) Fundamental Properties and Commercial Prospects, Henniker, NH, USA,
Acad of Sciences, New York, USA, 1987, pp. 492498.
q average loading of adsorbent for water (mol m3 ) [10] A. Ostroff, E. Hatzidimitriu, J.L. Kokini, Thermodynamics of water and
q* adsorbate loading in equilibrium with the solute con- ethanol adsorption on model biomass systems, Biotechnol. Bioeng. 31
centration c in the bulk fluid (mol m3 ) (8) (1988) 880884.
754 H. Chang et al. / Chemical Engineering and Processing 45 (2006) 747754

[11] J.P. Crawshaw, J.H. Hills, Sorption of ethanol and water by starchy [19] K.E. Beery, M. Gulati, E.P. Kvam, M.R. Ladisch, Effect of enzyme mod-
materials, Ind. Eng. Chem. Res. 29 (2) (1990) 307309. ification of corn grits on their properties as an adsorbent in a skarstrom
[12] J.P. Crawshaw, J.H. Hills, Experimental determination of binary sorption pressure swing cycle dryer, Adsorption 4 (34) (1998) 321335.
and desorption kinetics for the system ethanol, water, and maize at 90 C, [20] K.E. Beery, M.R. Ladisch, Chemistry and properties of starch based
Ind. Eng. Chem. Res. 31 (3) (1992) 887892. desiccants, Enzyme Microb. Technol. 28 (78) (2001) 573581.
[13] J.Y. Lee, P.J. Westgate, M.R. Ladisch, Water and ethanol sorption phe- [21] S. Al-Asheh, F. Banat, N. Al-Lagtah, Separation of ethanolwater mix-
nomena on starch, AIChE J. 37 (8) (1991) 11871195. tures using molecular sieves and biobased adsorbents, Chem. Eng. Res.
[14] P. Westgate, J.Y. Lee, M.R. Ladisch, Modeling of equilibrium sorption Des. 82 (7) (2004) 855864.
of water vapor on starch materials, Trans ASAE 35 (1) (1992) 213219. [22] M.R. Ladisch, Biobased adsorbents for drying of gases, Enzyme Microb.
[15] P.J. Westgate, M.R. Ladisch, Sorption of organics and water on starch, Technol. 20 (1997) 162164.
Ind. Eng. Chem. Res. 32 (8) (1993) 16761680. [23] J.H. Hills, I.M. Pirzada, Analysis and prediction of breakthrough curves
[16] X. Hu, W. Xie, Fixed-bed adsorption and fluidized-bed regeneration for for packed bed adsorption of water vapour on corn-meal, Chem. Eng.
breaking the azeotrope of ethanol and water, Sep. Sci. Technol. 36 (1) Res. Des. 67 (5) (1989) 442450.
(2001) 125136. [24] J.D. Seader, E.J. Henley, Separation Process Principles, John Wiley &
[17] M.R.L. Paul, J. Westgate, Air drying using corn grits as the sorbent in Sons, Inc., 2002, p. 814.
a pressure swing adsorber, AIChE J. 39 (4) (1993) 720723. [25] K. Adriaan, Numerical evaluation of equations describing transient heat
[18] L.E. Anderson, M. Gulati, P.J. Westgate, E.P. Kvam, K. Bowman, M.R. and mass transfer in packed solids, Ind. Eng. Chem. 40 (10) (1948)
Ladisch, Synthesis and optimization of a new starch-based adsorbent for 19921994.
dehumidification of air in a pressure-swing dryer, Ind. Eng. Chem. Res. [26] D.M. Ruthven, Principles of Adsorption and Adsorption Processes, John
35 (4) (1996) 11801187. Wiley & Sons, Inc., New York, 1984.

Вам также может понравиться