Carbon adsorption of phenol from aqueous solutions in the presence of other adsorbates

Carbon adsorption of phenol from aqueous solutions in the presence of other adsorbates

Carbon adsorption of phenol from aqueous solutions in the presence of other adsorbates John E. Ceresi, Jr.* and Chi Tien Department of Chemical Syracu...

954KB Sizes 0 Downloads 49 Views

Carbon adsorption of phenol from aqueous solutions in the presence of other adsorbates John E. Ceresi, Jr.* and Chi Tien Department of Chemical Syracuse, NY, USA

Engineering

and Materials

Science,

Syracuse

University,

Experiments offied-bed carbon adsorption of phenol from aqueous solutions in the presence of three other adsorbates were made for the purposes of: (a) validating the multicomponent adsorption algorithm of Wang and Tien and (b) exploring the application of the method developed by Tien and coworkers in characterizing the adsorption af$nity of background organics infixed-bed calculations. The experimental results were found to give good agreement with predictions based on the Wang-Tien algorithm. Predictions made with the use of the characterization results were found to be marginally accurate. It is apparent that aj%tity characterization based on batch experiment results is not adequate if there is a strong competitive adsorption behavior among the various adsorbates.

Keywords: adsorption;

activated carbon; phenol; multicomponent;

Introduction Fixed-bed carbon adsorption is often applied to remove impurities or contaminants present in water or wastewater. As the impurities or contaminants are composed of a variety of species with different adsorption affinity, carbon adsorption in water or waste treatment is a problem of multicomponent adsorption. The problem is further complicated by the fact that the identities of the species to be adsorbed are not completely known. A problem of this type, which is of particular concern presently, is the removal of a target compound (i.e., toxic substances or carcinogen) by carbon adsorption from natural water in the presence of certain unknown background substances. Frick and Sontheimer’ suggested that one may group the unknown species present in natural water into three categories: strongly adsorbable, weakly adsorbable and non-adsorbable. Extending this hypothesis, Tien and coworkers” developed a rational procedure for characterizing the adsorption affinity of solutions of unknown compositions. A solution of un-

* Present Address: National Semiconductor, 10810 Gilford Rd., Suite 111, Annapolis Junction, MD 20701, USA Address reprint requests to Prof. Tien at the Department of Chemical Engineering and Materials Science, Syracuse University, Syracuse, NY 13244, USA. Received 10 January 1991; accepted 3 June 1991

0 1991 Butterworth-Heinemann

aqueous solution

known composition is approximated as one with an arbitrary number (N) of species and the adsorption affinity of these pseudo species is represented by the Freundlich constants of the respective pure component isotherms. The concentration values and Freundlich constants of the pseudo species are determined from the adsorption equilibrium concentration data of the solution in question. If this approach is applied for analyzing carbon adsorption of target compound from aqueous solutions in the presence of background substances, the background substances may be represented as N number of pseudo species. The aqueous solution to be treated by adsorption may be considered to contain (N + 1) adsorbates, namely, the target compound plus the N pseudo species. One can then apply one of the existing algorithms to determine the extent of adsorption of the target compound. The principal objective of the present study is to determine the feasibility and validity of such an approach. For this purpose, fixed-bed adsorption experiments using granular activated carbon as adsorbents and aqueous solutions, containing phenol, acrolein, cyclohexanone and 2,4 dichlorophenol were conducted with phenol as the target compound and the other three as the background substances. Furthermore, since the existing multicomponent adsorption algorithms such as that developed by Wang and Tien have not been tested against data involving more than three adsorbates, these experimental data can also be Separations

Technology,

1991, vol. 1

273

Carbon adsorption of phenol: J. E. Ceresi, Jr. and C. Tien used to further validate the algorithm that constitutes the second objective of this work.

Fixed-bed multicomponent adsorption algorithm A brief outline of the multicomponent fixed-bed adsorption algorithm used in this work is outlined below. This algorithm was developed earlier by Wang and Tien* and used the ideal adsorbed solution (IAS) theory for adsorption equilibrium. The governing equations are:

aci usz + a4i

Z

f&i

o

pb 3

As used in most fixed-bed adsorption calculations, it is assumed that adsorption equilibrium is maintained at the exterior adsorbent-fluid interface that is given by Equation 3. If one uses the IAS theory for adsorption and, furthermore, the pure component adsorption isotherms of the individual adsorbates as given by the Freundlich expression, from Equations 2 and 3 one can express the interface concentrations cSi and qsi in terms of ci and qi by the following equations: 7T

csi= Zi [ Aini

(1)

=

3k,. -- L a,p, (ci - cS$ = k,i(qs; - qi)

48, =

ms, csl

fori=

1,2,.

9

3

*

.

.

cs,)

qsizz;

[

1

=+tb -’ j=I

T

(3) z; = [

&

4i

7Tni 1 +LE Aini +i

particle radius coefficient in the Freundlich expression of the irhspecies initial concentration CO equilibrium liquid phase concentration CCC solution phase concentration of species i Ci liquid phase concentration of species i in the Cp single species state solution phase concentration of the P specs; ties at the exterior carbon pellet surface total liquid phase concentration initially CTo liquid phase diffusion coefficient of species i Di surface phase diffusion coefficient of D,; species i equilibrium relationship fi liquid phase mass-transfer coefficient kli solid phase mass-transfer coefficient ksi mass of activated carbon it4 N number of pseudo species qb&gro adsorbed phase concentration of background species equilibrium adsorbed phase concentration qm adsorbed phase concentration of species i in qP the single species state adsorbed phase concentration of species i 4i

274

Separations

Technology,

1991, vol. 1

(5)

S

where +i = 3k,i/(a,p,kSj and r and S are the solution of the following pair of nonlinear algebraic equations

(64

Notation 2

(4b)

1 Ci +

. .N

where the dependent variables are the solution phase concentration of the ifh adsorbate ci and that at the exterior of the adsorbent pellet, csi, the average adsorbed phase concentration of the ifh adsorbate, qi, and that at the interface, qs,. The independent variables are z, the axial distance, and 8, the corrected time, defined as r - z&/u,. k,, and k,y,are the external and particle phase mass transfer coefficients, respectively. u, , E, pp and pb denote the superficial velocity, bed porosity, the density of the adsorbent pellets, and the bulk density of the pellet.

1

40 4si

4r R s t T US V Xi

Z

Zi

initial adsorbed phase concentration adsorbed interphase concentration total adsorbed phase concentration gas constant quantity appearing in Equations 6a and 6b time absolute temperature superficial liquid velocity volume of solution mass fraction of the ifh pseudo species axial distance mole fraction of species i in the adsorbed phase

Greek letters E 8 ?r pb PP

bed porosity corrected time quantity defined by Equations 6a and 6b bed density particle density defined as 3kli/(apppkSi) objective function C value Veptl experimental concentration Vc&d calculated concentration value

Carbon adsorption of phenol: J. E. Ceresi, Jr. and C. Tien

l

(6b)

l l

Principles of the adsorption affinity characterization procedure The procedure developed by Tien and coworkers for characterizing solutions of unknown composition is based on the following two assumptions: 1. Adsorption equilibrium between the solution and adsorbent in question can be described by the IAS theory. 2. The single component adsorption isotherm of the pseudo species may be expressed by the Freundlich expressions, namely, Equation 7. To approximate a solution of unknown composition as a solution of a number of pseudo species, one must specify the number of pseudo species, the Freundlich constants of the pseudo species and the concentration of the individual pseudo species. Since one does not have any idea about the molecular weight of the pseudo species, the pseudo species concentration must be given by certain surrogate quantities, for example, the total organic carbon (TOC) concentration. The total concentration of the solution can be determined by determining the TOC concentration of the solution. To select the values of N and to specify the corresponding values of Ai and ni, Jayaraj and Tien4 proposed the following: Consider that there are present in a solution a large number of adsorbates with their sin-

* Generally speaking, the Freundlich expression is not valid for low concentrations. A more correct expression is to use the Freundlich expression on a piecewise basis with ni + 1 as cp + 0 as discussed elsewhere.2*6

l l

l

b

Y--p-

(7)

where the subscript o denotes the pure component adsorption states. The Wang-Tien algorithm solves Equations 1 through 2 together with 4a through 6a for specified initial conditions [i.e., 4i = qi(z), i = 1,2, . . . N, 19< 01 and boundary conditions [i.e., ci = (ci), at z = 0, 8 2 0] to obtain values of ci, c,, , qi and qSias functions of z and 8. The information required are the operating variables, namely, the pellet size, the superficial velocity, and initial state of the pellet at the influent concentration. In addition, the adsorption equilibrium data [Ai and ni for i = 1, 2, . . . N] and the rate parameter values [kl and kSifor i = 1, 2, . . . N] are also needed.

l

-t

In the above equations, Ai and ni are the Freundlich constants of the irhadsorbate; in other words, the single component adsorption isotherm of the ithadsorbate is given as* 44 = A,&“”

8

e

l

l

I

l l

l

le Amin

AI

A2

l

b

Amax

A3

FREUNDLICH CKXFFICIENI; A

Figure 1 A-n Network for the Jayaraj and Tien characterization procedure

gle component adsorption isotherm data described by the Freundlich equation. Each adsorbate is represented by a particular point on the A - IZplane (see Figure 1). By applying the species grouping principle of Calligaris and Tien,’ the adsorbates near a given point of the A - n network of Figure I may be combined to form a pseudo species with A and n values corresponding to those of the grid point. One can, therefore, first set up an A - IZnetwork based on the knowledge of the maximum and minimum values of A and n. The number of the pseudo species, therefore, depends upon the finess of the network. Experimentally, one may conduct batch adsorption measurements using the solutions to be characterized and the selected adsorbent and obtain the equilibrium concentrations. Similarly, if the solution is properly characterized, the equilibrium concentration can be readily calculated using the equations developed before.6,7 Thus, one may assign the concentration values to the individual pseudo species by minimizing the objective function defined as

where v denotes the total equilibrium concentration achieved by contacting a solution of known concentration with adsorbent for a fixed value of v/M where v is the volume of the solution and M the mass of adsorbent. The superscript, k, refers to the kthmeasurement result. The minimization is to be carried out under the constraint that the sum of the initial concentrations of the individual pseudo species is known, or Separations

Technology,

1991, vol. 1

275

Carbon adsorption of phenol: J. E. Ceresi, Jr. and C. Tien

2 (do = (cd0

(8)

i=l

where cr, is the total concentration be characterized.

of the solution to

Experimental work Three types of experiments were performed in this work: (a) fixed-bed adsorption experiments, (b) batch adsorption experiments for determining the pure component adsorption isotherms and rate parameters of the individual adsorbates, and (c) batch adsorption experiments for characterizing the adsorption affinity and rate parameters of the pseudo species approximating the background substances. The test solutions used for the fixed-bed experiments were aqueous solutions containing phenol, acrolein cyclohexanone and 2,4 dichlorophenol. Aqueous solutions containing only one of the adsorbates were used for the second type of experiments. For the last kind of experiments, the test solutions used contained the last three adsorbates. The preparation of carbon and test solutions and the equipment and procedures used in the experiments are briefly described below. Experimental

systems

Filtrasorb 400 activated carbon (F 400, Calgon Corporation, Pittsburgh, PA) was first classified by sieving and those retained on the U.S. no. 20 sieve (but passed through no. 14 sieve) were used in this study. After classification, the carbon was extracted with boiling distilled water in a Soxhelt apparatus for two d in order to remove any leachable impurities present in the carbon. Afterwards, the carbon was further washed with distilled water and dried in an oven for two d at 110°C. The properties of the carbon are listed in Table 1. Chemicals (phenol, acrolein, cyclohexanone and 2,4 dichlorophenol) of Regent Grade and distilled water treated by carbon adsorption and deionizing columns were used for preparing the test solutions. The concentrations of the test solutions used varied from experiment to experiment. The relative composition of the background species was fixed with the mass ratio acrolein : cyclohexanone : 2,4 dichlorophenol = 1:1:2. Batch adsorption

experiment

Batch experiments were made to determine the equilibrium data (for either pure component adsorption

Table 1 Properties this work

of filtrasorb

400 activated

Separations

Fixed-bed

adsorption experiment

The apparatus used for conducting the fixed-bed experiments is shown schematically in Figure 2. The main feature of this apparatus was a microprocessorbased control system that allowed for the simultaneous operation of four columns with automatic sampling. For each column, three operating variables (the superficial velocity, bed height and influent concentration) may be varied. The procedure used is as follows: Test solutions of appropriate concentration levels in 200 L quantities were prepared and placed into the feed tank and fully wetted activated carbon placed into the column to a specified height. For the height and the weight of the carbon, the bed porosity was determined. Pre-wetting of the carbon particles was found necessary in order to eliminate the presence of air bubbles within the column. Once the column was fully prepared and the test solution was pumped into the overhead tank, the flow through the columns and the microprocessor were started simultaneously. The microprocessor automatically took influent and effluent samples from each column at preselected time intervals ranging from 0.5 to 4 h. A fixed-bed experiment may last as long as ten d. The total adsorbate concentrations of the samples taken were determined using a Beckman TOC analyzer as stated before. In addition, the phenol concentration of the samples were determined using a Hewlett Packard Model 57 10A Gas Chromatographer with a six-foot column packed with polyphenyl ether (PPE 6 ring) on a Tenax-GC Support.

Results Pure component

0.92 0.70 0.44 0.90 1100

Mean particle diameter Particle density Bulk density Pore volume Total surface area

276

carbon used in

isotherms of the four adsorbates or equilibrium concentration data for characterizing the three adsorcyclohexannom and 2,4 dichlobates-acrolein, rophenol-as background species) and the rate information. The procedures used were essentially the same; namely, a small amount of carbon was introduced in a flask containing the appropriate test solution with M/V (grams carbon/liter of solutions) varying from 0.5 to 30. The duration of each experiment ranged from several to 20 d. For the kinetic experiments, liquid samples were taken periodically to obtain the concentration histories. In all the batch experiments, the adsorbates (or total) concentrations (in terms of TOC) were determined using a Beckman Model 915 B TOCAMASTER Total Organic Carbon (TOC) Computational System.

Technology,

1991, vol. 1

(mm) (gm/cc) (gm/cc) Wgm) (m*/gm)

adsorption isotherms

As stated before, batch adsorption experiments were conducted to obtain equilibrium concentrations in both the solution and adsorption phases by contacting a known mass of adsorbents (M) with a given volume of test solutions (V). Let co be the initial concentration

Carbon adsorption

FIXEDBED

of phenol:

J. E. Ceresi, Jr. and C. Tien

COLUMNSW THKE-W&V SOLEND0WILvE%S)

MICRDP~SDR B INTERFACE

COLLECTION TABLE

JOG ClRCUlT

TOP VIEW OFCOUECTlDN TABLE PUMP Figure 2

Fixed-bed adsorption

apparatus with microprocessor-based

and qmand cm the equilibrium concentration and qm is given as qm = $

achieved,

(c, - cm)

control system

TOCII, the relative adsorption affinity of the four compounds, in the ascending order, is acrolein, cyclohexonone, phenol and 2,4 dichlorophenol. The Freundlich constant of the pure component adsorption isotherms are given in Table 2.

The batch experiments yield results of c, vs. M/V at a given temperature. Consequently, 4% can be readily calculated from the data. The results of 4% vs. cmfor the four adsorbates studied are shown in Figure 3. It is clear that all the data can be expressed by the Freundlich equation [i.e. Equation 71. Figure 3 also shows that over the concentration range, c < 300 mg Table 2 Experimentally determined values of Aj, ni of the individual adsorbates and pseudo species Adsorbates Phenol Acrolein Cyclohexanone 2.4 Dichlorophenol Background species approximation (a) Represented by one species (b) Represented by two species Pseudo species 1 Pseudo species 2

*

units

of

Ai

lrng‘,“m”,‘-““’

t dimensionless

Ai*

Ilit

41.88 1.12 6.02 139.4

3.12 1.27 1.94 6.37

15.00

2.2

6.00 88.00

3.00 2.2 Fluid phase concentrotion.cbng/ll

(,)1/n,

Figure 3 Single component adsorption bates used in this work (7 = 25°C)

Separations

Technology,

isotherms

of adsor-

1991, vol. 1

277

Carbon adsorption of phenol: J. E. Ceresi, Jr. and C. Tien the two species representation are shown. Good agreement is observed in all cases. Both the one-species and two-species representation results were used to predict fixed-bed adsorption performance to be discussed later. Intraparticle

diffusion coefficient

The algorithm of Wang and Tien considers mass transfer effects in both the fluid and particle phases. The

Table 3 Adsorption ground species Search no. 1

nit

xi (mass fraction)

1 2 1 2 1 2 1 2 1 2

6.00 88.00 6.00 88.00 6.00 88.00 6.00 88.00 6.00 88.00

2.20 2.20 2.20 2.20 2.20 2.20 2.20 2.20 2.20 2.20

0.367 0.633 0.369 0.631 0.367 0.633 0.365 0.635 0.367 0.633

2 FluId phase concentration

Figure 4

Background

species equilibrium

3

C(mgTOC/I)

4

data

Final average results

Characterization

of background

species

The simplest way of characterizing the background species is to assume that all the background species may be combined to form a single pseudo species. In other words, there exists a unique relationship between the total adsorbate equilibrium concentration in the solution phase, cb&rd, and that of the adsorbed phase, qbckgrd,obtained from the batch experiment. To test such a possibility, experimental data of qb&& vs. cb&grdobtained using test solutions with the same relative amount of acrolein, cyclohexanone and 2,4 dichlorophenol (namely, a mass ratio of 1 : 1 : 2 as stated earlier) but at different concentration levels are plotted against each other (in logarithmic scales as shown in Figure 4). The best fitted linear relationship for these data points is found to be qbckgrd

=

(10)

(l~.oo)(cb,kg,)“2’2

This relationship, of course, is not an adequate expression of the equilibrium data since there exists a significant and consistent difference among data obtained using test solutions of different concentration levels. To improve the characterization, a two-species representation was considered. Both of the pseudo species were assumed to have the same n values (and equal to the result of one-species representation) while the A values are 6.00 and 88.00 so that these two pseudo species have significantly different adsorption affinity. The concentration of the pseudo species (or their mass fraction) obtained from four different searches are shown in Table 3. It is clear that essentially the same results were obtained. In Table 4, the batch adsorption data obtained using testing solutions containing background species and predictions using 278

Separations

Technology,

1991, vol. 1

*

units

of

Ai

results of back-

A/*

loo

IO

characterization

Species

1

Ojm

affinity

(mgT°C)‘-““’

(I)““;

cm t dimensionless

Table 4 Comparison of total adsorbate equilibrium concentration calculated using characterization results with experiments

Data Test solution #l 1 5 4 5 6 Test solution #2

MIV b-n CII)

0.513 2.206 6.510 10.02 15.01 25.02

235.29 104.95 54.21 29.31 16.05 10.07

92.0 56.4 25.3 11.8 5.54 4.29

87.48 51.50 26.50 11.72 4.98 4.29

0.513 1.25 2.52 5.00

35.2 22.3 14.0 5.86

34.28 23.88 13.87 5.10

0.51 1.52

10.5 4.84

10.55 4.80

7

0.510

8

1.51 4.01 7.02 15.02 22.52

9 10 11 12 Test solution #3 13 14 15 16 Test solution #4 17 18

253.0 105.0 47.0 27.9 15.7 9.7

Average % error = 4.87% Test solution no. 1 2 3 4

Initial concentration, CT0 (mgTOCI/) 401 195 100 39.6

Carbon adsorption Table 5 Mass transfer fixed-bed adsorption

coefficients

Adsorbate or pseudo species Phenol Acrolein Cyclohexanone 2.4 Dichlorophenol Pseudo species #l Pseudo species #2

results used in predicting

OS, (m2/s) 6.5 x lo-” 3.0 x 10-l’ 3.9 x lo-” 6.0 x lo-” 1.8 x lo-” 1.8 x 10-l’

k,; (s-l 4.61 x 2.13 x 2.77 x 4.25 x 1.28 x 1.28 x

1 10m3 1O-3 1O-3 10m3 1O-3 10m3

fluid phase mass transfer is characterized by k,, , which can be obtained from a number of established correlations. The intraparticle mass transfer effect is described by ksi. If the intraparticle diffusion is controlled by surface diffusion, ksiis related to the surface diffusion coefficient, Dsi, by the following expression k,; =

15 Dsi a;

(11)

The surface diffusion can be determined by matching concentration histories of batch adsorption experiments with predictions using different values of D, .8 Once D,; is known, the corresponding ksi can be found from Equation 10. The value of ks, for the individual adsorbates so evaluated are given m Table 5. For the case where the background species are treated as unknowns and approximated by pseudo species, prediction of fixed-bed adsorption based on the Wang-Tien algorithm requires that the value of ksimust be known for the pseudo species. On the other hand, batch adsorption experiments only yield the total concentration (i.e., both pseudo species) history. To determine the ksi value, (or D,; values) from the total concentration histories, it was assumed that the surface diffusivities of the two pseudo species are the same. Based upon this assumption and the two species-representation results (see Table 3), the surface diffusivity of the pseudo species can be estimated by matching experimental concentration histories of batch adsorption experiments with predictions. For the estimated value of D, found in Figure 5, k,, was determined to be 1.28 x 1O-3 s-l.

J. E. Ceresi, Jr. and C. Tien

of phenol:

liquid phase mass transfer coefficients were estimated to the suggestion of Vermeulen et al9 or 3k, A=-UP

2.62 DiU, 1 - E 2ai.5 P

(12)

For the second type of predictions, the effluent histories were calculated by using the pseudo species representation for the background species. If the background species was approximated as a single pseudo species, the problem becomes that of fixed-bed adsorption involving two species, phenol and the pseudo species. Similarly, if the background species are approximated by a two-species representation, three species of adsorbates (two pseudo species plus phenol) were present in the fixed-bed adsorption. The Ai, ni, Xi and ksivalues for the pseudo species are those given in Table 3. The liquid phase mass transfer coefficient was calculated from Equation 10 using Di = 7 X 10e6 cm*/s based on the fact that the bulk diffusivity values of relatively small molecule organics in water are in the order of 7.0 x lO-‘j cm2/s. A large number of comparisons of the effluent histories with both types of predictions were made and the

I2

20

16

Time (ham)

Figure 5 Estimation of pseudo species surface diffusion coefficient from experimental data

Fixed-bed adsorption experiment The fixed-bed adsorption experiments yielded effluent concentration histories (both phenol and trial adsorbates). The experimental variables included the superficial velocity, the influent concentrations as well as the relative concentrations between the target compound (phenol) and background species. The experimental effluent concentration histories were compared with predictions of effluent concentration histories that were made in two ways. First, the problem was considered exactly, namely, by fixed-bed adsorption with four adsorbates (phenol, acrolein, cyclohexanone and 2,4 dichlorophenol). The predictions were made by using the Wang-Tien algorithm with appropriate Ai, tti and ksi values (given in Table 3). The

0

4

6

12

16

20

24

28

32

36

40

Tlme(hoursl

Figure 5 Experimental breakthrough curve and its comparison with predictions using a single pseudo species representation for background species (Run 1)

Separations

Technology,

1991, vol. 1

279

Carbon adsorption

0

4

of phenol:

I2

16

20 Time

J. E. Ceresi, Jr. and C. Tien

24

28

32

36

40

lhourrl

Figure 7 Experimental breakthrough curve and its comparison with predictions based on a two-pseudo species representation of background species (Run 1)

I

I 20

I 30

adsorbed phenol by background species is not manifested in the calculation results. If the background species is approximated by a twospecies representation, it is obvious that the adsorption affinity of phenol is less than the second pseudo species that can be seen from the results given in Tubles 2 and 3. A comparison of the experimental effluent phenol concentration history with calculations based on the two-species representation of background species is shown in Figure 7. The calculated results indeed exhibit the overshoot behavior. However, the extent of the predicted overshoot is much less than that exhibited experimentally. This difference is not surprising in view of the fact that the second pseudo species is assigned with an adsorption affinity significantly less than that of 2,4 dichlorophenol. Similar discrepancies in phenol concentration histories were observed in other cases where the flow rate, influent concentration or the relative concentration between the target compound and background species are changed from those used in Run 1 as shown in Figures 8, 9, and 10. As the relative amount of the target compound to that of the background species increases, the calculated results were found to be in better agreement with experiment. This can be seen by comparing Figure 10 with Figures I1 and 12.

I 40

Timelhcurrl

Figure 8 Experimental breakthrough with predictions (Run 2)

curve and its comparison

conclusion can be summarized as follows: The comparison of the phenol concentration histories are shown in Figures 6-10. In Figure 6, the effluent phenol concentration results of Run 1 are compared with predictions with the background substances represented by a single pseudo species. The phenol concentration history displays the so-called “overshoot” behavior; namely, at certain times, the effluent phenol concentration exceeds the influent value. This overshoot behavior occurs because of the displacement of adsorbed phenol by species with higher adsorption affinity. The results of the exact calculation (namely, consider the influent in its actual composition) are in good agreement with the experimental data. On the other hand, calculations using the one-species representation for the background species completely failed in predicting this behavior. In using a one-species representation for the background species, the background species as a whole has an adsorption affinity less than that of phenol. (This can be seen by comparing Equation 10 against the Freundlich expression of phenol which is 4 = 41.8 c1/3.12 .) Consequently, the displacement of 280

Separations

Technology,

1991, vol. 1

Figure 9 Experimental breakthrough with predictions (Run 3)

curve and its comparison

I.6 _ 1.6 -



Time( hours)

Figure 10 Experimental breakthrough son with predictions (Run 5)

curve and its compari-

Carbon adsorption

of phenol:

J. E. Ceresi, Jr. and C. Tien

Comparisons of total effluent concentration histories between experiment and calculation were also made. Two examples are given in Figures 13 and 14. In both cases, good agreement between exact calculation and experiments are seen. However, with the use of the two-species representation, significant differences were observed at later times.

Conclusions The present study demonstrates that the fixed-bed multicomponent adsorption algorithm of Wang and Tien can indeed be used to predict fixed-bed adsorption performance. On the other hand, if the background species are approximated by the pseudo spethe complete breakthrough cies representation, behavior of the target compound may not be predicted with sufficient accuracy. The lack of accuracy does not necessarily imply that the affinity characterization procedure developed previously has no value in design calculations. For, in most situations, one is primarily interested in the occurrence of breakthrough. If the breakthrough concentration is relatively low (say a few percent of the influent value or less), application of the affinity characterization procedure certainly yields good predictions. I.6

Time(houn1

Figure 13 Comparison between experiments and predictions of total effluent concentrations at short times (Run 3)

I

w

OO



20 1



40“1

60

1



80



IO0

*

120 !

J

Tome (hours)

Figure 14 Comparison between experiments and predictions of total effluent concentrations at long times (Run 3)

References 1.

0

20

40

60

2.

Time (hours1

Figure 11 Experimental breakthrough son with predictions (Run 7)

curve and its compari-

3.

4.

5.

6.

7.

8. 0 0

I

I

I

4

e

I2

I I6

I 24

I 32

I 40

I 48

70

9.

Time(houro)

Figure 12 Experimental breakthrough son with predictions (Run 9)

curve and its compari-

Frick, R.B. and Sontheimer, H. Adsorption equilibria in multisolute mixtures of known and unknown composition. In Activated Carbon Adsorption of Organics from Aqueous Phase, Vol. 1. I.H. Suffet and M.J. MC&ire, eds. Ann Arbor: Ann Arbor Science, 1980 Wang, S.-C. and Tien, C. Further work on multicomponent liquid phase adsorption in fixed beds. AIChE J. 1982,28,565572 Jayaraj, K. and Tien, C. Characterization of adsorption affinity of solutions. Proc. Envir. Eng. Conf., Los Angeles, 1984, 394399 Jayaraj, K. and Tien, C. Characterization of adsorption aflinity of unknown substances in aqueous solutions. Ind. Eng. Chem. Process Des. Dev. 1985, 24, 1230-1239 Kage, H. and Tien, C. Further development of the adsorption affinity characterization procedure for aqueous solutions with unknown compositions. Ind. Eng. Chem. Res. 1987,X, 284292 Tien, C. Incorporation of the IAS theory in multicomponent adsorption calculations. Chem. Eng. Commun. 1986,40, 265279 Calligaris, M. and Tien, C. Species grouping in multicomponent adsorption calculations. Can. J. Chem. Eng. 1982, 68, 772-780 Larson, A.C. and Tien, C. Multicomponent liquid phase adsorption in batch, Parts I & II. Chem. Eng. Commun., 1984, 27, 339-379 Vermeulen, T., Klein, G., and Heister, N.K. Adsorption and ion exchange. In Chemical Engineers’ Handbook, 5th Ed. R.H. Perry and C.H. Chilton, eds. New York: McGraw-Hill, 1973, pp. 16-20

Separations

Technology,

1991, vol. 1

281