A THESIS IN CHEMICAL ENGINEERING Submitted to the Graduate ...

PERFORlvIANCE OF A YORK-SCHEIBEL COLUMN

ON AN ACETONITRILE-POTASSIUII

CARBONATE-V/ATER SYSTEM

by

DOUGLAS GEORGE BRESLER JR., B.S.

A THESIS

IN

CHEMICAL ENGINEERING

Submitted to the Graduate Faculty of Texas Technological College in Partial Fulfillment of the Requirements of

the Degree of

MA^ER OF SCIENCE IN CHE-IICAL ENGINEERING

Approved

May, 1969

^\

n 905 T3

dop.^

OitW-ooc^

ACKNa^EDGEtdENTS

The author expresses his sincere appreciation to Dr. A. G, Oberg

for his direction of this thesis and to the other members of the com­

mittee, Dr, R, M, Bethea and Dr, H, R. Heichelheim for their helpful

criticism and encouragement.

Special appreciation.is given to my wife, Pam, for her encourage­

ment, and for her assistance in preparing the final manuscript.

ii

BS.

TABLE OF COIJTENTS

ACKl\IOV/LEDGEiIEr>JTS ii

LIST OF TABLES. . . . . v

LIST OF ILLUSTR/iTIONS v i

Chapter

I. INTRODUCTION 1

II. LITERATUPJi; REVIEl-/ 5

Common Dehydration Techniques • 5

Dehydration by Salting-Out . • . • 5

Performance of York-Scheibel Columns 6

Simplified Calculation Procedures . . . . . 11

III. THEORETICAL DEVELOPI IENT 13

Choosing a System 13

Equilibrium Curve Determination . 13

Operating Line Determination. . , 14-

Mass Transfer Calculations • 19

IV. EXPERII-CENTAL APPARATUS AND MATERIALS 29

Apparatus . 29

Materials 32

V. ANALYTICAL TECHNIQUES 33

Infrared Spectroscopy 33

Gas Chromatography 33

Quantitative Calibration 34

Titration 36

111

••4.U

VI. EXPERIi^IENTAL PROCEDURE 37

Solution Preparation 37

Column Operation 3^

VII, EXPERIl^EIfCAL RESULTS 4-2

Approach to Steady State kZ

Extent of Dehydration 42

Column Performance. 48

VIII. SUl-C-IARY 63

IX, CONCLUSIONS 65

X, RECOM'IENDATIONS 66

LITERATURE CITED 6?

APPENDIX 69

IV

LIST OF TABLES

1. Experimental Dehydration Data for an Average Feed Ratio of E/R = 0.58 4-3

2. Experimental Dehydration Data for an Average Feed Ratio of E/R = 1.18 44

3. Experimental Dehydration Data for an Average Feed Ratio of E/R = 2.39 ^5

4. Experimental Dehydration Data for an Average Feed Ratio of E/R = 3.71 4-6

5. Operating Data of an Eleven-Stage York-Scheibel Column on a CH3CN-K2CO0-H2O System 4?

6. Performance Data of an Eleven-Stage York-Scheibel Column

on a CH3CN-K2CO0-H2O System 50

7. Ternary Binodal Data 72

8. Tie Line Data 73

9. Analysis of Quantitative Chromatographic Data for Acetonitrile-V/ater Solutions. 7k

LIST OF ILLUSTRATIONS

1. Equilibrium (Distribution) Diagram for the System CH3CN-K2CO3-H2O 15

2. Schematic of a Continuous Countercurrent Extraction

System. I6

3. Simple Graphical Dete3:^ination of an Operating Line . . . . 18

4. Modified Graphical Determination of an Operating Line . . . 20

5. Operating Diagram 21

6. Experimental Apparatus. . . . 30

7. Product Purity vs. Extract to Raffinate Ratio 49

8. Raffinate Efficiency vs. Agitator Speed 52

9. Extract Efficiency vs. Agitator Speed . 53

10. Raffinate Efficiency vs. Extract to Raf.finate Ratio . . . . 55

11. Extract Efficiency vs. Extract to Raffinate Ratio 56

12. Raffinate Efficiency vs. Total Column Throughput. . . . . . 58

13. Extract Efficiency vs. Total Column Throughput 59

14. Raffinate Overall Mass Transfer Coefficient vs. Total Column Throughput • 60

15. Extract Overall Mass Transfer Coefficient vs. Total

Column Throughput 6\

16. Chromatographic Calibration Curve 76

17. Rotameter Calibration Curve for the CH3CN-H2O Phase . . , , 77

18. Low Range Rotameter Calibration Curve for the K2CO3-H2O Phase 78

19. High Range Rotameter Calibration Curve for the K2CO3-H2O Phase 79

VI

CHAPTER I

INTRODUCTION

The dehydration of organic liquids containing water has long been

a problem of the chemical industry, since these liquids often come

into contact with water during processing and must be purified before

being used. A dehydration technique v;hich has received limited indus­

trial interest, but has widely reported use in the laboratory, is

salting-out (7). Salting-out involves the preferential attraction of

water by a salt placed in an aqueous solution of an organic liquid.

The salt and water combine to form a new solution v;hich is essentially

insoluble in the remaining organic-water solution. The amount of new

insoluble solution formed depends upon the amount of salt added until

an excess of salt is present and an equilibrium distribution of water

between the salt and the organic liquid has been established. Ternary

diagrams which show the equilibriuin curve at a given temperature and

pressure for all possible combinations of organic liquid, water, and

salt are available in the literature for many sj'-stems. Study of var­

ious ternary diagrams indicates that essentially complete dehydration

of a particular organic liquid-vjater solution can be obtained with

some salts.

The choice of a salt for the dehydration of a particular organic

liquid must include consideration of its dehydrating pov;er, its sol­

ubility in the organic liquid, its reactivity with respect to the

liquid, its corrosiveness with respect to the potential materials of

construction or the equipment presently available for making the

1

necessary contact, and its cost and availability (7). The determi­

nation of dehydrating power and solubility can be obtained from, a

ternary diagram of the system in question. The reactivity of the

salt and organic liquid can also be determined from a diagram at the

desired temperature and pressure since any reactivity would prevent

development of such a diagram. Corrosiveness, cost, and availability

must be determined separately. These three are intimately connected

since corrosiveness would control the ease of production of the salt,

and thus control cost and availability.

The contacting of solid salt material and a solution of an organic

liquid and water normally is done in a batch-wise system. Once the

salt-water solution or phase and the organic liquid-water solution or

phase have reached equilibrium, the phases may be separated physically

by decantation.

The contacting of a saturated salt solution and the organic liquid-

water solution can be accomplished in several types of continuous sys­

tems. The most advantageous system industrially vjould be a continuous

«

countercurrent operation vdth a tower using packing, a tower equipped

with distinct mixing and settling sections, or a tower equipped vd-th

plates commonly used for distillation processes (8). The second type

may have rotating agitator paddles in the mixing sections and wire

mesh packing in the settling sections, as in the case of a York-Scheibel

column. In a continuous system the used salt-water solution v;ould also

have to be dehydrated to achieve a saturated solution for reuse. The

dehydration of the salt-water solution could be accomplished in an

associated evaporation system. The dehydrated organic liquid phase

would also need some additional purification in most cases, either to

remove any remaining water or to remove small amounts of dissolved

salt,

A slightly altered operation for a continuous system vrould involve

the use of a partially saturated salt solution since such solutions

often have considerable salting-out ability (8). A partially satu­

rated salt solution would be easier to handle in a circulating system

since fully saturated solutions often lose some salt in the form of

crystals which would deposit in, and possibly clog, parts of the system.

The use of a salting-out system for del^ydrating purposes can

easily circumvent tvjo problems that are often associated vdth the

dehydration of organic liquids containing water. The lower temper­

atures associated with liquid-liquid extraction systems can avoid

decomposition properties encountered T;d.th heat sensitive organic liquids.

The inability of distillation to achieve high product purities v:hen

constant boiling mixtures, or azeotropes, are formed can be avoided

if the product purity associated with a salting-out system is above

the azeotropic composition. The elimination of these two problems

provides an important reason for studying dehydration by salting-out

in an extraction system.

The purpose of this thesis is to demonstrate the value of using

a York-Scheibel extraction column in conjunction vdth a salting-out

system for dehydrating an orgsjiic liquid. The particular salting-out

system used consisted of acetonitrile-potassium carbonate-water, where

a partially saturated potassium carbonate-water solution was used to

dehydrate an acetonitrile-v;ater solution. The performance of the

York-Scheibel column vras used as the basis for determining the best

operating conditions for dehydrating the acetonitrile-v;ater solution.

The particular points of interest included: (l) product purity as a

function of feed ratio at various agitator speeds, (2) the column

efficiency as a function of agitator speed at various feed ratios,

(3) the column efficiency as a function of feed ratio at various agi­

tator speeds, (4) the column efficiency as a function of total column

throughput at various agitator speeds, and (5) the overall mass trans­

fer coefficients as a function of total column throughput at various

agitator speeds.

CHAPTER II

LITERATURE RBVIEV/

Comraon Dehydration Techniques

In the past the primaiy dehydration technique has been distil­

lation since the vdde range of possible operating temperatures and

pressures makes it adaptable to most liquids of interest (7). Addi­

tional dehydration techniques of lesser importance include crystal­

lization, chemical reaction, and adsorption (7). Crystallization

usually involves the freezing and subsequent removal of water crystals

by filtration. The use of a chemical reaction for dehydration nor­

mally involves the selective reaction of water vdth some other material

in order to obtain a reaction product which may be readily separated

from the organic liquid of interest. The adsorption of v;ater on mate­

rials such as silica gel provides an additional dehydration technique (7).

However, adsorption is not easily adapted to the removal of large quan­

tities of vfater.

Dehydration by Salting-Out

As mentioned previously, the dehydration technique to be used

here is that of salting-out. Salting-out theory has been widely studied

and investigated in the past. A large number of articles concerning

salting-out are available in the literature. A recent reviev: of

salting-out theory was given by Ergin (4).

Meissner and Stokes (7) suggested the industrial importance of

solvent dehydration by salting-out. They discussed the batch-vdse

5

dehydration of an aqueous solution of methyl ethyl ketone by salting-

out with calcium chloride.

A later article by Meissner et al. (8) reported the use of a

continuous countercurrent extraction column for solvent dehydration

by salting-out. The column vras first operated as a spray tov;er and

later as a packed tower with both one-half inch Berl saddles and one-

half inch Raschig rings being used alternately as packing. The sol­

vent to be dehydrated consisted of an aqueous solution of methyl ethyl

ketone, while the salting-out medium was a nearly saturated solution

of calcium chloride and viater. The height of a theoretical transfer

unit and the overall mass transfer coefficient was given for the dis­

persed phase. The column was operated with both the organic and

salt-rich phases being alternately dispersed. The overall mass trans­

fer coefficients for the dispersed phase were reported to vary linearly

with respect to dispersed phase flovn-ate for both the spray and packed

operating conditions. The variation of overall mass transfer coef­

ficient vdth dispersed phase fIovn:»ate was found to be greater for a

packed than for a spray tower with the same liquids being used. The

explanation given was that the packing breaks up the dispersed phase

into fine droplets and thus promotes mass transfer.

Performance of York-Scheibel Columns

The majority of the work done with York-Scheibel type extraction

columns has been on systems that do not have a salt as one of the com­

ponents. This type of column has been used to study both simple and

fractional extraction. Although most studies do not involve salting-

out, some of these will be cited for the benefit of their many

7

qualitative and quantitative observations.

Scheibel (l6) suggests that in this type of column an efficiency

of greater than one theoretical stage may be possible if,an actual

stage includes both a mixing and a calming or settling section. The

combined mixing and calming sections provide ideal conditions for

extraction since the initial flov; into the packing is highly dispersed.

Performance data on the fractional liquid extraction of ortho-

from para-chloronitrobenzene vdth Skellysolve C as one solvent and

methanol containing 15 volume percent vrater as the other solvent was

presented (l6) for varying mixing and calning section heights. Effi­

ciencies of betvjeen 37 and 115 percent vjere obtained in the 31-stage

York-Scheibel extractor used. The size of a mixing section v;as not

found to influence the performance of the column. The smallest prac­

tical mixing section v;as preferred because it gave the lov;est power

input to the agitator and also decreased the height of the stage.

The effect of packing height on stage efficiency was studied (l6)

and found to be a major factor in determining the efficiency of the

column. Observation of the column indicated that for packing heights,

or calming sections, of less than two inches the mixing and calming

sections had the same milky appearance. Packing heights of two inches

and greater permitted formation of a clear zone just prior to the next

mixing section where the lighter dispersed phase droplets were seen

rising through the heavier continuous phase. The maximura efficiency

of the combined mixing and calming sections occurs vihen the beginning

of the clear zone coincides with the beginning of the next mixing sec­

tion. Thus, the best conditions for mass transfer can be maintained

8

at all times. Adequate mixing v/as indicated visually by a ndlky

appearance throughout the column vxhere individual dispersed phase

droplets are no longer discemable. Low agitator speeds often allovred

large and easily visible droplets to remain.

The purpose of the mesh packing, as explained by Scheibel, v;as

that of entrainment separation rather than promotion of finer droplet

formation. Once the droplets coalesce, the purpose of the packing had

been accomplished and the material was ready for another mixing. With­

out complete separation of the phases, the efficiency of the stage vras

limited by the entrainment of the solvent.

Scheibel (17) subsequently reported the extraction of acetic acid

from water vdth methyl isobutyl ketone in a one-inch inside diameter

(I,D.) laboratory column. The results x-;ere in general agreement with

his earlier fractional extraction v;ork and provided further support

for his qualitative and quantitative observations. Variations in

efficiency betvjeen approximately 62 and 114 percent occurred while

changing the packing height from one to tv;o inches. This variation

indicated the controlling effect of packing height on efficiency.

Since column capacity increased with the free space of the packing,

Scheibel suggested that the packing used should have a free space of

greater than 98 percent.

Data on a seitdcommercial 12-inch column m t h a 400 gallon per

hour capacity have been presented by Scheibel and Karr (18), The

systems investigated with this relatively large column include acetone-

o-3grlene-water, acetic acid-o-2cylene-x^ater, and acetic acid-methyl

isobutyl ketone-water. The stage efficiencies vrere found to increase

in all runs to a maximum as agitator speed was increased. The increase

in efficiency continued up to the flooding point in some cases. If

flooding conditions v:ere not exceeded, the efficiency vjas found ,to

decrease beyond the maximum with increased agitator speed, rather than

leveling off and remaining constant. The decrease in efficiency when

flooding vras not reached was assumed to be due to the formation of an

emulsion that could not be broken in the amount of packing present.

Stage efficiencies also showed a maximum near the flooding point v;hen

plotted versus total column throughput for a given agitator speed.

The range of agitator speeds giving maximum efficiency was foimd to

vary vd.th the properties of the solvents and vjas generally longer for

the more readily separable solvent phases. In solvent phases that

separate slowly, the agitator speed was more critical, and greater

packing height was required for the same efficiency. The report stated

that the data and visual inspection of the column indicated that the

dispersion in the mixing section increased with throughput at lovrer

throughputs and appeared to decrease at higher throughputs. The higher

throughputs were assumed to cause too large a quantity of liquid to

pass through the mixing section to be completely dispersed by the

agitation provided.

Scheibel (20) found that there v;as an optimum power input per

unit volume of solvents flowing in a column. At low agitator speeds

the stage efficiency was greater at low throughput, and at high agita­

tor speeds the efficiency was found to be greater at high throughput.

The pov;er input to the agitator shaft vxas calculated from the speed

and torque and was found to vary as the cube of the speed. The type

10

of agitator used by Scheibel vias found to have a general correlation

for the power consumption of

P = 1,85 5 £ e N 3 _ 6

where D is the outside diaraeter of the agitator in ft, ^ is the

average density in the mixing section based on one-third holdup of

the dispersed phase, N is the speed of the agitator in rev/sec, g is

the gravitational constant of 32.2 ft/sec^, and P is the power input

per agitator in ft-lb^/sec. This correlation may be used for York-

Scheibel extraction columns provided by the York Process Equipment

Corporation (26). Although the amount of pov;er that could be put into

a particular system without producing an emulsion or a flooding con­

dition was found to be a function of fluid properties, it also vras a

function of vjhich phase was dispersed. In a mixing section the tend­

ency of the mass of liquid to rotate will force the lighter liquid

to the middle. Thus, vrfien the problem v;as that of dispersing the

light phase, this effect facilitated the mixing. V\Tien dispersion of

the heavy phase was desired, this effect opposed mixing.

Ergin (4) presented data obtained id.th a York-Scheibel extraction

column on a system using salting-out for dehydration. The system was

acetonitrile-potassium carbonate-w-ater where the potassium carbonate-

water phase was dispersed. He proposed empirical correlations for

determining mass transfer coefficients for the continuous and dis­

persed phases. The correlations present the mass transfer coefficient

as a function of both continuous and dispersed phase mass velocities.

The equation proposed for the continuous, or acetonitrile-water phase

11

was

Kya = 170.07 exp [0.244 X 10--5(V/S)(L/S)] (2-2)

where the correlation coefficient v;as 0.823 for 31 points. The equa­

tion proposed for the dispersed, or potassium carbonate-water phase

was

K^a = 0.02170 exp [0.1103 X 10-^(V/S)(L/S)] (2-3)

where the correlation coefficient V7as 0.421 for 30 points. The mass

transfer coefficients are in Ibs/hr ft- (mass fraction driving force),

while the mass velocity terms of V/s and L/S are in Ibs/hr ft^. These

correlations appear to be the only ones proposed for use vdth a salting-

out system in a York-Scheibel extractor.

Simplified Calculation Procedures

Scheibel and Othmer (l4) presented simplified methods of calcu­

lating the niimber of theoretical transfer units required in an extrac­

tion system. The simplifications were developed to eliminate the need

for a graphical integration step and greatly reduce the time required

to make such calculations. Formulas are proposed for both immiscible

and partially miscible systems vdth an estimated error of tvro percent

with respect to the graphical solution.

Scheibel and Othmer (15) also presented nomographs for use in

deterndning mean driving forces in diffusional operations including

extraction. The formula used was based on the assumption that the

curvature of the equilibrium line V7as such that the value of the slope

12

varied linearly vdth the concentration of the distributed component

in one phase, V/hen calculating the number of theoretical transfer

units, this assumption led to 8,5 percent error when compared to

graphical integration techniques.

A summary of calculation procedures for liquid-liquid extraction,

including a section on continuous countercurrent systems, has been

presented by Scheibel (19). Graphical representations of equations

pertaining to batch-vdse and continuous countercurrent systems are

presented. The advantage of both contacting methods can readily be

compared for various operating conditions by use of these graphs.

CH;VPTEK III

THEORETICAL DEVELOPIIENT

Choosing a System

The ternary system of acetonitrile-potassiura carbonate-water is

a Type I system as indicated by the formation of one pair of partially

miscible liquids (22; page 15). Several factors v;ere considered in

choosing this system for use in a dehydration study. These factors

included: the strong salting-out characteristics exhibited by the

potassium carbonate-v;ater solution, the industrial importance of

acetonitrile, the availability of a ternary diagram at a convenient

temperature and pressure (12), and the previous use of the same system

by Ergin (4). Ergin operated the coluron vdth the agitator at 1300 RPM

and with the potassium carbonate-water phase dispersed.

Equilibrium Curve Determination

The reported ternary data are shovjn in Table 7* The experi­

mentally determined tie lines vjere insufficient to obtain a complete

equilibrium curve for extraction calculations. The correlation of

tie line data for use in interpolation was investigated by Renard

and Heichelheim (13) for systems involving water-acetonitrile-salts.

All correlations attempted yielded poor results for the ternary systems

used. These poor correlations indicated that the use of this technique

would be unsatisfactory. An older method of obtaining additional tie

lines involves graphical construction using a ternary diagram along

with the experimentally determined tie lines (22; page 29). The

13

14

graphical construction method was used to obtain a sufficient number

of tie lines to establish an equilibrium curve. The tie line data are

listed in Table 8.

A plot of the vjeight fraction vjater in the potassium carbonr.te-

water phase versus that in the acetonitrile-vrater phase gives an

equilibrium, or distribution, curve for the system. The ratio of the

weight fraction v;ater in the salt-rich phase to that in the organic-

rich phase at any point on the curve is the distribution coefficient,

A plot of the distribution curve is shoT.m in Figure 1.

Operating Line Determination

An operating line for each run vras determined graphically by the

method of Varteressian and Fenske (23). More recent discussions of

this technique are presented by Oliver (9; page 221) and Treybal (22;

page 35^). The method for obtaining the operating line, as used here,

consists of first plotting the extract feed composition and the raffinate

feed and product compositions on a ternary equilibrium diagram. The

extract is the' potassium carbonate-v7ater phase while the raffinate

is the acetonitrile-water phase. The compositions of these tv7o feed

solutions and one product solution were used as a starting point since

they are all considered to be binary rather than teimary solutions.

The extract feed and raffinate feed compositions and flowrates permit

the calculation of a hypothetical mixing point composition and quan­

tity. The ndxing point composition v;as established by the follovdng

material balances with respect to Figure 2.

0.8 15

1 r

0 0,1 0.2 0.3 0.4 0.5 0.6

Wt. Frac. H2O in the Raffinate (CH3CN-Rich Phase)

Figure 1 - Equilibrium (Distribution) Diagram for the System CH3CN-K2CO3-H2O

R2 A

^R2

E.

AL

X E2

"ST

16

R A

R^

' ^

E

H

i dH

T

% 1

Rl

X El

E,

.S.

Figure 2 - Schematic of a Continuous Countercurrent Extraction System

M = R, + E^

_

(3-1)

(3-2)

(3-3)

The units of all terms are given in the nomenclature section. The

calculated composition of the mixing point X, can then be located on

a line connecting Xo and XT;, on the ternary diagram. If only steady ^1 ^2

state conditions are assumed during the collection of operating data,

17

material balances including the exit strearn flovn-ates and compositions

must yield the same value of M and X, as

M = R2 + E^ (3-4)

^ = ^2\ '' ^l\ (3-5)

h = 2^R2 •'' ^l^Ei (3-6)

M

Graphical construction vdll then permit the determination of Xr^ and ^1

the subsequent location of the operating point 0 (2; page 44, 6; page

145, 22; page 226). The material balance represented by the above

described graphical construction can be vjritten as

0 = E^ - R^ (3-7)

0 = E2 - R2 (3-8)

The location of points for an operating line can be accomplished by

draidjig lines at some arbitrary increment apart from 0 which inter-

sected both sides of the binodal curve (22; page 35^). An abbreviated

example of this graphical technique can be seen in Figure 3. A modi­

fication of this technique permits the location of X-p, and XR points

even when point 0 lies at extreme distances from the ternary diagram.

An additional line is constructed parallel to the extract or raffinate

side of the ternary diagram. This line crosses the converging Xj .-Xg

and Xj o-Xg lines. The distance between these lines at that point

and the distance between the lines on the other side of the diagram

are divided into an equal number of increments. V/hen the respective

18

o

•H

W

c •H -P U o

&

o

!>i.5

19

points are connected, values of X^^ and Xj^^ are determined in the

same manner as before. These lines will slope toward point 0 at the

same angle that would exist if they were actually extended to that

point. An example of this modified technique can be seen in Figure 4.

When enough sets of these Xg^ and XR^ points have been determined,

an operating line can be plotted on an equilibrium or distribution

diagram as shown in Figure 5. Plots such as these are required for

making the calculations which lead to the number of theoretical trans­

fer units, or the value of the overall mass transfer coefficient,

for an extraction system. The equations that are used with diagrams

such as Plgure 5 vd.ll be developed next.

Mass Transfer Calculations

The recognition of laminar and turbulent flow as two distinct

types of fluid motion leads to a similar recognition of two types of

diffusion. The molecular diffusion occurring during laminar flow

exists because of concentration differences between adjacent liquid

layers. The eddy diffusion of turbulent flow causes small quantities

of liquid to be moved from place to place maintaining a high concen­

tration difference for promoting mass traiisfer. The more commonly

encountered theories and equations proposed for describing diffusional

processes are discussed by Treybal (22; page 150) and Bird et al. (1;

page 495)

In a York-Scheibel extraction coluinn, turbulent flow is maintained

in all mixing sections while the calming or settling sections serve as

a transition from turbulent to laminar flovr. As mentioned previously,

the ideal situation occurs vrhen laminar flow is achieved in the calming

20

W C CM

n « « O X X •d •P «H TJ

g ° § w w > 0 >rH 0) pi ft; •g'^J'^ •H > C

(1) O t j o s '-< ^ © © +>

^U .H Q> EH >iXi

O 1

<aj

o •P H ©

H H ^ JH rsJ

P H

21 0.80

0.75 © w rt

Xi

xi o •A

I o o

CM

• P O

U

©

• H

O CM

O

S

0.70

0.65

0.60

0.55

0.50

Equilibrium Line

Operating Line (Run 3)

0.0 0.10 0.20 0.30 0.40 0.50

Wt. Frac. H2O in the Raffinate (CH3CN-Rich Phase)

Figure 5 - Operating Diagram

22

section just prior to the beginning of the next misdng section.

The velocity distribution and role of eddy diffusion are not

knovTn in normal ex-traction situations. This lack of information pre­

vents the direct calculation of m.ass transfer rates based on diffu­

sional theories. The usual approach is to measure mass transfer

rates for a specific situation and then correlate the results to

extend their usefulness. Mass transfer data are correlated by use

of the mass transfer coefficient V7hich was originally defined for

this purpose. The mass transfer coefficient, V7hen defined for V7eight

fraction driving force, can be expressed as

N; = k A X ^ (3-9)

where N^ is a flux in (lbs/(hr)(ft~')), k is the mass transfer coef­

ficient in (lbs/(hr)(ft )(weight fraction)), and A X^ is the total

change in concentration along a diffusional path in weight fraction.

When the above equation is applied to the raffinate and extract phases

betv7een the bulk liquid and the interface, the follovdng equations

result

The kp and kg terms are the mass transfer coefficients for the raf­

finate and extract respectively. The Xp. and Xv). terms are the inter­

face concentrations of the respective phases. Thus, the mass transfer

from the bulk raffinate to the interface is seen to be equal to the

mass transfer from the interface to the bulk extract.

The tv7o-film or tv7o-re si stance theory as discussed by Treybal

23

(22; page 174) and Perry (10; page 14-18) suggests that the values

of X^^ and Xg^, V7hich cannot be detemined experimentally, are really

equilibrium values. The addition of the reciprocal values of the

mass transfer coefficients for the tv7o phases can then represent

the overall resistance to mass transfer. Since this approach requires

that equilibrium exist at the interface, there is no resistance to

mass transfer at the interface itself. Incorporating the concept

of tv7o resistances and the assumed equilibrium at the interface per­

mits an overall mass transfer coefficient to be defined for each

phase. The overall mass transfer coefficients are defined by the

follovdng equations

% = % ( % - % * ) = % A X Q K (3-11)

"A = KE(XE-%*) = % ^ % (3-12)

The overall mass transfer coefficients Kj and Kg are in (lbs/(hr)(ft )

(weight fraction)), and the equilibrium interfacial compositions

X^ and Xg are in V7eight fraction. The use of these overall mass

transfer coefficients provides a necessary simplification in correla­

ting mass transfer data.

A schematic of an extraction column vdth continuous countercurrent

contact has been shown in Figure 2. If N is the total transfer of

solute, (Ibs/hr), then the rate of mass transfer for a differential

section dH can be described as

dN = d(RXR) = KK(XR-Xj *)dA (3-13)

24

where R is the raffinate rate in (Ibs/hr) and dA is the interfacial

transfer area in (ft^) associated with dH, a differential height in

(ft). Since dA cannot be conveniently measured, it is expressed as

dA = aSdH (3-1^)

where a is the interfacial area per unit volume of packing in (ft^/ft-^)

and S is the column cross sectional area in (ft^). The original

equation can now be reind.tten as

dN = d(RXR) = Ki aS(Xj -XR*)dH (3-15)

The total raffinate rate R varies throughout the column due to the

loss of solute. A solute free raffinate remains constant and is

therefore easier to use since the way in which R varies is unknovTn.

The solute free raffinate is incorporated as follows (22; page 3^5)

X^ \ _ R(l-Xj )dXj _ R dXj^

1-XR ( 1 - X R ) 2 (I-XR)

d(RXR) = R(l-XR)d — i L = ^ ^ = " (3-16)

Combining this alteration vdth the previous equation gives

5 ^ ^ = KpaS(XR-XR*)dH (3-17) (1-Xj^) R R R

or dH = -^ ^ -. (3-18) K^aS (l-Xj )(Xj -Xj *)

The resulting equation can be integrated and solved for the column

height as

25

%aS (l-Xg)(XH-XR-).

The second quantity vdthin the integral, including the concentration

difference (XR-XR ), is a measure of the difficulty of extraction.

The difficulty of extraction is designated the number of transfer

units, N.(. The height of a transfer iinit or U^ is represented by

the first quantity d.thin the integral. The product of the height

of a transfer unit and the number of transfer units is seen to be

equal to the height of a column required to perform a specified ex­

traction. The equation may be vjritten in simplified form as

«toR<™toa (3-20)

Many simplified methods of evaluating the above right hand integral

have been proposed in the past (l4, 15, 21; page 393, 22; page 3^7).

Some of these methods involve enough simplifying assumptions to elim.-

inate the integration step V7hile the more rigorous ones merely modify

the integral to make the necessary graphical integration easier.

Since the mass transfer coefficient contains a term (1-XR)-.

that varies throughout the column, this term is incorporated in the

final equation to make the product KR(1-XR)2^ approximately constant

(5; page 274, 22; page 3^6). The resulting equation appears as

26

dH R (l-%)lm^R

KRaS(l-XR):i , (l-XR)(Xr -XR') (3-21)

Now that the first term on the right hand side has been made approx­

imately constant, it can be moved outside the integral. An average

of the term is used to elindnate some of the loss of idgor experienced

by its removal from inside the integral. The average is obtained from

the terminal values since only these are knovjn in a continuous system.

The resulting equation appears as follov7s

dH = R

%^S(l-%)lm av

(^~%)lm^R

(1-XR)(XR-XR') (3-22)

where

H, tOR R

K^aSd-X^)!^ av

(3-23)

and

N tOR (l-XR)l^dXR

(1-XR)(XR-XJ^*)

(3-24)

The normal method of calculating values of ^-^Q^ involves graphically

integrating the above expression using the operating diagrams like

those discussed previously. With the equation in its present form,

a graphical integration would involve plotting ((1-XJ^)]J>I/(1-XR)(XJ^-XR ))

27

versus Xj^ and detenrdning the area under the curve betvjeen X R and Xp •

Another simplification in the integral is often possible. If the

value of (1-XR ) and (1-XR) differ by less than a factor of tv:o, an

arithmetic average rather than a logarithmic average m.ay be used for

(1-XR)2J^. Investigation of this method by V. eigand (25) indicated that

an error of no more than 1.5 percent v7ould occur. Evaluation of

(l-Xo)-, as an arithmetic average is seen as

(l.X,)3^ - ^'-^^^^ ; ^^-^H> (3-25)

Substitution of this term into the equation for N.J QR gives the fol­

lovdng equation

*°« "^ (x,-x,*) 2 — '

The graphical integration of this equation is somev7hat easier than

for the previous case. This latter form of the equation V7as chosen

for determining values of N.t QR from experimental data. A similarly

derived expression for N.| Qg \Ja.s also used as

B ^^E 1 I n r ' El f .

Once N+oR ^^^ ^tOE ^^^® been obtained, the respective overall mass

transfer coefficients may be evaluated. The equation must be rearraiiged

to isolate K^a or K a tei-ms as follov7s E'

28

% ^ = R

HS(l-Xj^)3^ av

N tOR (3-28)

and

V = E

^HS(l-Xg);^_ av

N tOE (3-29)

where H is the height of the extraction column used. The values of

KRa and Kga obtained can then be used to determine the effect of

changing conditions in an extraction column for a given system.

CHAPTER IV

EXPERB4ENTAL APPARATUS AND MATERIALS

Apparatus

The countercurrent liquid-liquid extraction column used in this

experiment was a York-Scheibel model XA-2. The column V7as of two inch

I.D. pyrex glass six feet high. A variable speed agitator capable of

speeds up to I3OO revolutions per minute (RPM) V7as provided. The

agitators and shaft were of stainless steel. The column had eleven

stages with one stage being made up of a mixing and a settling or

calming section. The mixing sections vjere one inch high and the set­

tling sections wore four inches high. Each mixing section V7as equipped

with one centrally located impeller seven-eighths inch in diameter.

The settling sections V7ere of stainless steel mesh vdth 97 to 98 per­

cent void space. It was necessary to replace the two glass liquid

feed vessels normally provided vdth a York-Scheibel column since they

would not permit feeding at a constant head pressure. Ten gallon

galvanized steel cans vjere used for all feed, mixing and storage

vessels, vdth the exception of the acetonitrile-v7ater solution make-up

tank, V7hich V7as of stainless steel. The column was provided with two

rotameters for measuring the feed rates. Each rotameter V7as equipped

vdth a glass and a steel bead, m.aking calibration possible over a

wider range of flowrates. The tubing was either stainless steel or

polyethylene to avoid corrosion problems. Sample valves were also of

polyethylene.

The experimental apparatus is sho\m in Figure 6. A large stainless

29

30

Figure 6 - Experimental Apparatus

31

steel mixing tank, which is not shovjn, v;as used for making up large

quantities of acetonitrile-vjater feed solution. The acetonitrile-

rich feed solution V7as added to Vessel 1 as needed. The solution V7as

pumped from Vessel 1 to Vessel 2, the constant level feed tank. An

overflow point on Vessel 2, mth a return line to Vessel 1, insured

a constant level for the feed. All of the acetonitrile-rich feed

solution equipment v;as located at a sufficient height to insure an

adequate head to overcome the approximately six feet of more dense

salt solution in the column. A polyethylene line led to the loxTer

level, where the rest of the apparatus V7as located. The sample point

on the acetonitrile-rich feed line was also located on the lower level

for convenience. The acetonitrile-rich feed solution entered near the

bottom of the column after passing through a flow control rotameter.

Vessel 3 served as an accumulator for the dehydrated acetonitrile-rich

solution or product. The line from the column to the product accumu­

lator was equipped vdth a sample point.

Vessel 4, vjhich was equipped vdth a mixer, served as the make-up

point for the potassium carbonate-water salt solution. Salt solution

from Vessel 4 was allov7ed to flov; to Vessel 5 as required. Salt sol­

ution V7as pmaped from Vessel 5 to the constant level feed tank, Vessel

6. The arrangement of the overflow point and return line vias the same

as for the acetonitrile-rich feed system. A sample point V7as provided

on the line between the salt solution feed tank and the flov: control

rotameter. The accumulator for the vraste salt solution was Vessel 7.

An interface control leg was located betv7een the bottom of the colunin

and the v;aste accumulator. The interface control leg was adjusted to

32

maintain the interface between the salt solution feed point near the

top of the column and the acetordtrile-rich product overflov; point a

few inches higher. A sam-ple point was located on the waste salt solu­

tion line follovdng the interface control leg.

Materials

The acetonitrile used was technical grade, purchased from the

Sohio Chemical Company. The potassiura carbonate V7as also technical

grade and vras purchased from the Industrial Chendcals Division of

Allied Chemical Corporation.

CHAPTER V

ANALYTICAL TECHI^IQUES

Infrared Spectroscopy

Investigation of the use of infrared spectroscopy for quantita­

tively analyzing acetonitrile-vjater solutions did not yield a sat­

isfactory calibration curve. The difficulty encountered \d.th infrared

spectroscopy V7as that the particular instrument available did not have

an attenuator which would allov7 the peak heights to be m.aintained

approximately constant while varying the composition of the samples.

The inability to read peak heights accurately from the vddely varying

peak sizes resulted in an unsatisfactory calibration curve.

Gas Chromatography

The next technique investigated for analyzing acetonitrile-v;ater

solutions V7as gas chromatography. A search of the literature revealed

only tv7o chromatographic columns that v;ould separate both V7ater and

acetonitrile under the same operating conditions (11). Unfortunately

the reference cited did not give the ratio of stationary phase to

solid support. Several columns v;ere tried with considerable success

in that the fourth column tried proved to be satisfactory. The column

packing V7as 80-100 mesh Porapak-Q in six feet of one-fourth inch stain-

loss steel tubing. The optimum operating conditions of the chromato-

graph, vd-th respect to peak shape, retention time, and the elimination

of the tailing effect of water, proved to be as follows: an injection

port temperature of approximately l63°C; a carrier gas flowrate of

33

3k

65 milliliters of helium per minute at ambiont conditions; a column

temperature of 135°C; a detector temperature of 200°C; and a bridge

power setting of I50 milliamperes. The retention times for the water

and acetonitrile were 1,5 and 6.4 minutes respectively. The chro-

matograph used was an F & M model 5OO.

Quantitative Calibration

In referring to Renard*s binodal data (12) on the system

CH3CN-K2CO3-H2O it V7as evident that he v;as unable to detect meas­

urable amounts of salt in the acetonitrile-rich layer after the

concentration reached 82 weight percent CH3Cni. The data thus indi­

cated that although there may be minute amounts of salt present above

the 82 weight percent CHoCN point, the presence of the salt may be

reasonably ignored.

All experimental dehydration runs v;ere intended to give con­

centrations of the acetonitrile-rich product streami in excess of the

azeotropic composition of 83.7 vreight percent acetonitrile and I6.3

weight percent water. Thus, all acetonitrile-rich product streams

obtained at steady state v7ould have greater than 82 V7eight percent

acetonitrile and should have had sufficient time to approach equilib­

rium. The relative absence of salt in the acetonitrile-rich product

stream permitted the use of a chromatographic calibration curve for

analyzing both the acetonitrile-rich feed and product streams.

The calibration curve prepared from experimental data consisted

of 27 points representing different knovjn compositions of acetonitrile-

water solutions. There are points located appro^dmately every 1 weight

percent up to a composition of 10 V7eight percent water. Beyond the 10

35

percent point, points are located approxi.mately every 5 percent up to

95 weight percent V7ater. The 1 percent interval in the low V7ater con­

centration region was to improve the accuracy there since it v;as

believed that most acetonitrile-rich product compositions v7ould exceed

90 weight percent acetonitrile.

All data points are the result of analyzing each sample five

times on the chromatograph in order to obtain an average value of

percentage peak height of water. The average percentage peak height

values and the weight fractions of water vjere correlated by a step­

wise polynomial regression program (3; page 258). The criterion for

obtaining an acceptable fit was that the predicted calibration curve

be at least as good a representation of the data as could be obtained

graphically. Some difficulties were encountered xd.th the curve fit

equations provided by the regression program. It V7as necessary to

divide the data for the calibration curve into regions in order to

obtain the desired accuracy. The equation predicted for the region

of 0 to 68 percent relative peak height of water and 0 to 50 V7eight

percent V7ater was Y = 1.659 x 10'^ + 1.009 x 10" X^. The equation

predicted for the region of 68 to 100 percent relative peak height of

water and 50 to 100 weight percent V7ater was Y = -6,675 x 10"- +

1,696 X 10 X. Values of X are in percentage relative peak heights

of v;ater, V7hile values of Y are in weight fractions of V7ater. A plot

of the calibration curve with the different regions indicated can be

seen in Figure I6.

Average percentage relative peak heights and the standard devia­

tion (24; page 63) calculated for five analyses of the same sample are

36

given in Table 9. The V7eight percentage water predicted by the curve

fit equations is given along vdth the actual value for each experimental

point. As can be seen from the table, the curve fit equations do not

include every experimental point, but these equations do represent

the best curve that can be statistically dravTn through the points in

the respective regions.

The usefulness of the curve fit equations V7as extended by writing

a program which used the equations to calculate values of V7eight frac­

tion V7ater from values of percentage relative peak height in incre­

ments of 0.01 from 0 to 100. The tabulated results served as a ready

reference for determining vreight fractions of water from values of

percentage relative peak height while analyzing actual experimental

samples.

Titration

It was found that 25 milliliter samples of waste salt solution

could be titrated vdth less than 50 milliliters of 4 N or greater HCL.

The inciicator used was 0.1 percent methyl orange in alcohol. Standard

titration procedures V7ere somev7hat altered, in that the salt solution

was not heated to remove dissolved CO2 which v;ould cause an early end-

point to be reached. Several tests showed that agitation of the flask

during titration eliminated enough CO2 as it was formed to obtain the

same endpoint as through heating. The tests consisted of simply

titrating wdth agitation until an endpoint was reached and then heating

the sample. If the color remained, not enough CO2 was present to alter

the endpoint, which was true for all cases tested. Therefore, the

more expedient method of agitation V7as used to remove COp.

CHAPl'ER VI

EXPERIt4ENTAL PROCEDURE

Solution Preparation

A solution of approximately 55 weight percent acetonitrile and

45 weight percent water was prepared using largo containers and a

scale. This solution was then analyzed and its strength adjusted as

required to obtain exactly 55 weight percent acetonitrile. The potas­

sium carbonate-water solution was prepared by adding the salt to

Vessel 4 in Figure 6 which was partially filled with V7ater until

excess salt appeared on the bottom of the vessel. Excess salt v:as

maintained on the bottom of Vessel 5 while operating the column. The

presence of this excess salt in contact with the extract feed solu­

tion did not result in a saturated solution as had been previously

expected. Subsequent analysis of the effluent extract solution indi­

cated that not enough salt v;as present to achieve saturation of the

extract feed solution. Thus, partially saturated extract feed solu­

tions were found to result in all cases. The salt concentration of

the extract feed solution remained approximately constant during each

experimental run. Additional acetonitrile-water and potassium car-

bonate-v7ater solutions V7ere made by the same techniques as required.

The rotameter calibration curve for the acetonitrile-rich feed

solution is shovm in Figure 17. The calibration curves for the two-

float rotameter used vdth the salt feed solution are shown in Figures

18 and 19. All points on the three rotameter calibration curves were

the result of averaging three runs at each major setting. Each run

37

38

lasted between five and ten minutes depending on the volurae of solu­

tion collected.

Column Operation

After assembly of all parts of the apparatus, it was tested for

leaks with water. The initial filling of the column vd.th salt solu­

tion was started by raising the interface control leg to its highest

point to prevent any continuous phase from flowing out of the column.

Then the salt solution V7as allox red to flox into the column until it

covered the top stage. The dispersed, or acetonitrile-v7ater, solu­

tion was then allowed to flov; into the column until it overflov7ed

into the acetonitrile-rich product accumulator. The location of the

interface was adjusted to the desired level and then the agitator V7as

started. Agitator speed adjustments were made with the aid of a rota­

tion counter and a timer. After the desired agitator speed v;as obtained,

both dispersed and continuous flov7s were restarted. Continuous adjust­

ment of the interface control leg was necessary until the holdup of

the colurun stabilized. Once the coluriin holdup stabilized, ver '- little

difficulty was encountered in maintaining the interface at the desired

level. Shut down was accomplished by turning off both feed control

valves and the agitator.

The colximn V7as alv;ays operated vdth the interface at the top.

The salt solution was the continuous phase and the acetonitrile-rich

solution was the dispersed phase. Four agitator speeds were chosen

for investigation. The speeds were 500, 750, 1000, and 1250 RPM.

All runs were made at a room temperature of 25^C + 2°C, and since

the distribution data for the system do not vary appreciably with

39

temperature, no attempt V7as made to correct for temperature varis,-

tions.

The approdmate flowrates and feed ratios v;ere chosen for the

first experimental run from data collected by Ergin (4). In later

runs, several other combinations of flox%Trates and feed ratios V7ere

attempted in an effort to locate a suitable range for systematic

investigation.

A data-taking scheme was established folloid.ng the sixth run.

Data V7ere collected at four feed ratios for the four agitator speeds

previously selected. The feed rate of the acetonitrile-rich solution

was held constant at 27 (ml/nln) with the feed rate of the salt solu­

tion being varied. The four salt solution feed rates vjere 10, 20,

40, and 60 (ml/min).

One combination of feed ratio and agitator speed constituted

one run. The most efficient V7ay to collect data V7as to maintain the

feed ratio and thus the feed rates constant, until runs had been made

at all agitator speeds. Changes in agitator speed did not delay the

approach to steady state to the same extent as did changing the feed

ratio.

At the beginning of a run, the level of the acetonitrile-rich

feed solution in Vessel 1 was checked and adjusted as required. The

circulation pump which maintained a constant head in Vessel 2 V7as also

started. Then the level of the salt solution in Vessel 5 was checked

and adjusted as required. The salt solution circulation pump was

started. After both feed systems V7ere circulating, the feed control

valves V7ere adjusted to give the desired rotameter settings. The

40

agitator was then started and adjusted to the desired speed. The

interface control leg V7as adjusted to maintain the interface at the

desired level.

The column V7as allox%Ted to operate at the desired feed rates for

55 minutes. At this time floxirate determinations were started on

both the acetonitrile-rich product and x-7aste salt solution streams.

The flowrates x ere determined by collecting knoxjn volumes of liquid

over a timed interval x hich x as generally greater than five minutes.

It xms necessary to control the location of the interface X7hile

checking the effluent flovirates since any alteration in its position

would alter the rate of both effluent streams by decreasing one and

increasing the other. Thus, if the location of the interface fluc­

tuated during a floxn:*ate determination, it was necessary to continue

the test until the interface could be restored to its original level

or the volumes of liquid collected in a given time v7ould not be

representative of the indicated flowrate.

As soon as the flowrate determinations were complete, samples of

both effluent streams were taken for analysis. Analysis of the sam­

ples was carried out immediately on most runs to determine if steady

state had been reached. Sample analysis and column operation could

not be carried out concurrently by one person. Thus, it was necessary

to have tv7o people present x hen operating the column and analyzing

solutions simultaneously. After some experience was gained, a fev;

runs V7ere made in which samples Xi7ere collected for a fixed length of

time and analyzed as soon as possible. Each sample was analyzed twice

and an average V7as taken to establish the V7eight percentage V7ater at

41

that time.

Flowrate determinations vdth the subsequent taking of samples

were repeated at 30~minute intervals follovdng the first detemd-

nation. A particular run was continued until analysis of the aceto­

nitrile-rich product stream indicated that steady state had been reached,

The attainment of steady state conditions is normally assumed when the

deviation from steady state conditions is less than the variations

caused by other factors in the system. It has been found that for

small one-inch columns of the York-Scheibel type, variations of up to

10 percent are encountered in the feed streams in spite of V7ell-cali­

brated rotameters (26). In such a system, an approach to vdthin 10

percent of steady state is as close as could be expected.

Steady state V7as assumed when the composition of the samples

analyzed either leveled out, or began to vary slightly rather than

to show a tendency to go in one direction. The slight variations

were considered to be produced by the difficulty in operating the

column consistently for extended lengths of time.

CHAPTER VII

EXPERH IENTAJ. 11ESULTS

Approach to Steady State

A listing of experimental data shovdng the approach to steady

state is given by feed ratio and agitator speed in Tables 1, 2, 3,

and 4. Experimental runs were generally three to four hours in

duration. An early upset in the column operation during Run No. l4

was responsible for the longer time required to reach steady state.

The reported feed ratios are averages obtained from operating con­

ditions xd.thin the steady state region. The steady state region

for each run was determined by considering the extent of variations

in composition for both the raffinate and extract streams.

The average steady state operating conditions for the column

are shovTn in Table 5* The V7eight percentage xfater is given for all

inlet and exit streams to shov; the exchange of X';ater that occurred

betxv een the raffinate and extract phases. The apparent volumetric

flovrrates of the feed streams are indicated as they resulted from

the set point feed rates mentioned previously. The apparent volu­

metric flowrates of the effluent streams are averaged values obtained

from floX'7rate determinations that were made while the column V7as in

operation. The run numbers, agitator speeds, and feed ratios are

also listed to shox-7 their relationship to the other operating data.

Extent of Dehydration

The effect of the feed ratio on product purity at constant

42

k3

TABLE 1

EXPERH^'NTAL DEHYDRATION DATA FOR AN AVERAGE FEED RATIO OF E/R=0.58

Run No.

16

3

15

14

Time Hr.

1 ll 2 ^2

3

1 •••2

2 ^2

3 3i

1 Ii 2 ^2

3 3i 4

1 •12

2 2l ^2

3 31 4 5

t^ 6| 7 7i 8 8|

Raffinate Product H2O Wt. j>

14.62 14,62 14.56 14.61 14.58

14.48 14.44 13.87 13.90 14.00 13.72

13.9^ 15.69 1 .35 14.23 14.40 14.05 14.03

8.72 9.k9 10.13 10.99 11.63 12.18 12.73 12.93 12.97 13.38 13.69 14.02 15.kl 14.45 14.60 14.67

Agitator Speed RPM

500

750

1000

1250

Feed Ratio, E/R Wt. K2CO3-H2O

Wt. CH3CN-H2O

0.57

0.58

0.60

0.58

44

TABLE 2

EXPERDffiNTAL DEHYDRATION DATA FOR Al AVERAGE FEED RATIO OF E / R = 1 . 1 8

Ron No. tN V .

17

18

19

20

Time Hr.

1 1-1 2 2l '^z 3 3i 4

1 •'•2

2 2l ^2 3

1 •*2

2 ^2 3

1 ii 2 ^z 3

4

Raffinate Product H2O VJt.

10.84 10.52 9.92 9.50 9.87 9.75 9.70

9. 1 9.87 9.80 9.83 10.24

9.77 . 9. 2

9.60 9.81 10.10

7.56 8.69 8.09 8.11 7.99 8.05 7.97

Agitator Speed RPM

500

750

1000

1250

Feed Ratio, E/R Wt. K2CO3-H2O

Wt. CH3CN-H2O

1.17

1.19

1.18

1.17

k5

TABLE 3

EXPERIMENTAL DEHYDR TION DATA FOR AN AVERAGE FEED RATIO OF E / R = 2 . 3 9

Run No.

9

8

7

6

Time Hr,

1 J-2

2 ^2

3 3i 4

1

2 ^2

3 3i 4

1 ii •^2

2 2i ^2

3 3i

1 Ii • 2

2 ?i ^2

3 3i 4 ^

Raffinate Product H2O Wt. 'J3

5M 5. 9 5. 2 5.51 5.38 5.63 5.68

5.88 5.91 6.02 6.04 6.22 5.95 6.03

4.85 5.05 5.18 5.15 5.23 5. 9

6,28 5.93 5.82 5.79 5.73 5.81 5.99 5.51

Agitator Speed RPM

500

750

1000

1250

Feed Ratio, E/R Wt. K2CO3-H2O

Wt. CH3CN-H2O

2.39

2.40

2.39

2.39

TABLE 4

46

EXPERIMiENTAL DEHYDRATION DATA FOR Alsl AVERAGE FEED RATIO OF E / R = 3 . 7 1

Run No.

10

11

12

13

Time Hr.

1 Ii ^Z

Z 2i ^z 3 3i 4 1 li ^Z

Z ?i ^2

3 3i 1 li ^Z

2 ?i ^z 3 %

4 1 li ^Z

2 ?i ^2

3 3i 4

Raffinate Product H2O Wt. 5

4.89 4.83 4.69 4.85 4.99 5.13 4.88

5.1^ 5.10 5.06 5.25 5.36 5.26

6.95 6.56 7.16 6.38 7.32 7.00 6.01

8.72 7.26 7.26 7.26 7.51 7.51 8,13

Agitator Speed RPM

500

750

1000

1250

Feed Ratio, E/R Wt. K2CO3-H2O

Wt. CH3CN-H2O

3.67

3.70

3.71

3.77

kl

k—4

6 t H CO

o CM

1 CA

o o CM •s i

i S

< A HI o <

s o

§ ^ \A O O

i-:i

w ^ pq M

W M

\A S <q CO t ^ 1

STA

GE

Y

OR

I

1

s w § tC4 O

< ; EH ^ O O

E H

S M O^ o

( i i

- o \-A CM W

.^ 1 O CA

•H O +3 O c\} CM

©

0 CM

W J

"^ 0 C ^

0

i fc

O CM CM +3

1 © C ^ - P ^ O rt ^ o p i - - ^ CM 0 ^

! ^ - P «H

-p g

c M

O CM CM - P

W «H 1 ©

^ - P J-t

C ^ C t i " - ^ W c^ O - P

«H

-P g

f j M

^ ©

- P

\f ^

40 :s

0 CM

W 1 - P

S'B, 0

CM

CM

1 CA C

0 H 0 CM

0 CM

^ ^

C ^

m 0

0 CM

1 ^ f^ 0 H <A

W 0

u 0 TlJ • P © >E ciJ © O i -P P./S •H CO bD

<

c . ^ <^ ( ^ i ^

I N - 0 0 0 CO U ^ V ^ A N O U ^

. . . . 0 0 0 0

0 0 O N CM C ^ v r ^ v A 0 0 {>-

. . . . T H ^H rH "TH

{ N - £ > - { > - O -O N O N O N ON

. . . . 0 0 0 0

v A U > > o CM

^ .;t.=^ ^ xH T H V H T H

C ^ C A C A C A NO NO NO NO

. . . . CM CM CM CM

C M N O T S C5N

0 O N C ^ 0 -C ^ N O N O N O

NO C J N C ^ ^

^ CM 0 ( A NO NO NO NO

NO 0 CM NO . . . .

^ CA.::^- - : t •»H T H T H -^-l

0 00 0 0 . . . .

0 0 0 0 0 u ^ O VA ^ C ^ O CM

T-J T H

NO c A u - ^ J j -•»H •e-K T^K

O - O N O O t> -%-l t H -v-l .iH

. . . . •r-1 vH vH T H

M3 CM O - V A > ^ { > - CS-NO

. . . . CM C\J CM CM

V A ^ A » A U ^ ON ON ON O N

. . . . T H xH T H -vH

VAVACV2 00 C^CM C ^ CM

•r-\ •v-\ T-K t-K

O ^ C A C A C A NO NO NO NO

. . . . CM OJ CM CM

CJNC^ •iHNO

NO C^£>-U>v NO NO NO NO

C ^ O N O O N

CM 0 vH r H NO NO NO NO

{N.C0 O - "tH . . . .

C T N O N C J N O O

0 0 0 0 . . . .

\ A > A \ A U ^ . : j - ^ ^ . c } -

0 0 0 0 0 V A O U ^ ^no- 0 CM

•JH T H

0 - 0 0 O N O T H T H •r-« CM

CJN 0 O N ON c^-:j- r^c^

. . . . CM CM CM CM

T H CACJN CM .cj- ^ CM u•^

. . . . - : 3 - - ^ ^ . c } -

C J N O N C T N CTN CO 00 0 0 00

. . . . C ^ O ^ C ^ C A

xH T H 0 - . C\2 C A C ^ - : t -=}•

T H vH -cH i H

0 ^ 0 ^ C A C ^ NO NO NO NO

. . . . CM CM CM CM

t S - O > A O

c^r^.c3- ^ NO NO NO NO

O N O 0 - t N -

0 CiN O N ON VO V A V A V A

CM 0 vH O N

. . . . x A v O » A V A

0 0 0 0 . . . .

V A U ^ ^ X A V A ^ ^ . c i - ^

0 0 0 0 0 V A O ^ A U ^ j > - 0 CM

vH T H

C3N00 l > - v O

CN 0 vH O -NO t> - t> - 0--

. . . . C ^ C ^ C A C ^

'

O^CM . : t " ^ C A . ^ CM {> -

• . . . NO NO NO {>-

C ^ C ^ C A O ^ CO 0 0 00 CO

. . . . l A V A u ^ u - ^

CN- CvJ C A C A C^CM O ^ C ^

vH T H "(H i H

C ^ C A C ^ C A NO NO NO NO

. . . . CM CM CM CM

vH U^«vO 0 ^

0 0 0 CO NO \0 ^n>A»A

T-l CM O N O O

IS-NO U>»CA X A V A U ~ ^ v A

O N - : ^ CO {>-. . . .

- ^ V A N O 0 -

0 0 0 0 . . . .

V A V A W ^ V A

-=f -:t -^ -:t

0 0 0 0 0 vno u- U ^ C ^ 0 CM

vH \ H

0 r H CO CA xH T H xH T H

48

agitator speeds can be seen in Figure 7. The best overall feed ratio

corresponds to E/R ^ 2.40. Although higher product purities are

possible for agitator speeds of 500 and 750 RPil, it is doubtfiil that

the necessary increase in the extract phase flovTrato is justified.

An apparent maximum product purity was encountered for agitator speeds

of 1000 and 1250 RPM since lox er purities resulted when the feed ratio

was increased beyond E/R = 2.40. This decrease in product purity at

higher agitator speeds was probably due to insufficient emulsion

breakup in the vdre mesh packed calming sections of the colximn.

Although pacld-ng height V7as not a variable in this study, insufficient

packing height has been observed for systems that have a slox-; rate of

phase separation or emulsion breakup (l6). The inability of the

calming sections to perform satisfactorily is apparently due to both

the increased dispersion possible at the higher agitator speeds and

the increased total flox-; through the column.

Column Performance

The performance data calculated from the average steady state

operating conditions are shovjn in Table 6. The calculated number of

theoretical transfer units are given for each run x-dth respect to both

the raffinate and extract phases. The column efficiencies related to

both types of transfer units are also given, V7here an actual stage is

made up of a mixing and a calming .section. The total pov7er was cal­

culated by use of equation (2-1) which is repeated here for convenience.

p = 1.85 e l ^ i ^ (2-1) g

9

97

95

•p

o

o U

Qt ©

Xi -p •H

93

91 o o • p

©

o © 89

87

85 0.0 1.0 2.0

E/R

3.0 4.0

Figure 7 - Product Purity vs. Extract to Raffinate Ratio

50

NO

3 E-*

65

O CM

tn I CA

O

o CM

W I CA

o f. < © to

<H - P r ^

TO d - H ^^ ^^ O

«H

© EH . H

O W CiH

c(3

S O

u

o H O

•d TJ © -P R5

•P H O tH g

•^

»

o © w «H

o ^ CM H 4i

O <H

^ ^

r.

w o c © •H O •H «H <H W

p T r^ O O

o o vH w o -p

s

o o T-l

Pi o -p

J^

•;;! 0

A-> O

< S

• ^ ;:< -P o < ;

s

w

o JH © .o

s

d •

o

© U O «H (D to

Xi C EH

E H

O -P

O -P

CtJ o

- ^ ON NO CM ( A c ^ i A \ A

-:d- O M3 CM CM T H

CM CM

CO O O-.c}-0 0 - j - - : t O

. . . . O - j - ON T-l CM T-l T H CM

O IN-NO ^ A T-l 0 *^00 T H

. . . . V A O O 0 0 CM CO CM T-l

-Cj- CM ON (N. O CM ^ O

. . . . O C ^ T H 0 0 TH V H -x-t

CJN NO CM C ^ t> - ( A O N CM

O N O N O ONVA. : : ^

CM O N

CO ON T H O C ^ T H 0 ^ { N •

O C^ T H CO NO T H T H CO

C ^ O ^ C ^ C ^ C ^ C M C A ^ » A C A i A C A CO CM CM vH

. : 3 ' O N O C M O O T H V A C A C O C S - ^ A . : ^

CO CM VA T-l T H j j - T-l VA CO {VOO -cj-

T H VACS- O N CM NO O-OO CO O - C N - 4 -

T H CA

t r ^ i A ^ CV CM NO O CM CO (N-CO l A

rt C^

CO ON VA CO

T H t A . c h O CAOO NO O CO { > - 0 0 (N-

T-l C ^

O N O N C ^ C ^ . . . .

ON NO ( A G O T H T - I T - I T - I

- : t O N O N ON . . . .

C J N O C ^ C ^ . . . .

- : t O N O O N . . . .

0 ( N - C 7 N O O O N O N O ^ C N i c A C O r H

T ^ U ^ V A C 0 O N O O N . ; } - O N C J N - C J - O CO T^ C ^ IN-

O 00 TH tS-C^ CO CO CO

NO ^ O -d-

c^ cj r>;3-£>-vr\NONO 0- - : ;h CO O

CM CM Cvl CO

a tN-CO-C}-u-^ V A T - I CO - : j - O

T H CO CO

( A T - I . C J - O -T - l ^ NO ON C A T H ( A O N

t N ^ - : t CO .^ t NO 0 0 ON T H 00 O T-1

T H O T-l T-l

V A V A NO C ^ NO T H O N 0 0 O N CO C^CO

.:J- ON CO O £ > - V A O N C O ON NO NO VA

NO O . : t VA NO ( A C Q O CM C^CM CNj

o o o o o o o o

C ^ O ^ C M C O ( A CO C A . : t

O O O O O V A O VA V A C ^ O CJ

T-l TH

O O O O O V A O VA v A t N - O CO

T-l T-l

^ CvJ T H T H { > - V A O N O C ^ O N T H ON

. . . . NO CA V A C ^

O O O O O V A O VA V A I N - O CM

^ t H O^T- I O N V A V A CO CJNNO - : t CO

. . . . CO Cv2 CO CO

O O O O O V A O VA VACN- O CO

T-l T-l

VO ( A VA-Cj-T-l T H T-l

C^CO O N O T H T-l T H CM

CJNOO C N N O O •cH T H

CO ( A T-l T H

51

The total power is given in Table 6 in (ft-lb^^/sec) for each run.

Although the pov7er term is a function of a composite liquid density

term, the controlling factor is seen to be agitator speed. The

repeating pattern of the pov7er terms for all feed ratios so closely

approximates the repeating pattern of the agitator speeds that the

variations in composite densities encountered betx'7een runs is insig­

nificant. Thus, the pov7er term does not appear to provide any better

approach for correlating column performance data than do the respec­

tive agitator speeds. The final set of performance data consists of

the overall mass transfer coefficients for both the raffinate and

extract phases.

The calculations leading to the number of theoretical transfer

units were performed in terms of V7eight fractions since the ternary

equilibrium data V7ere available in that form. A xinit conversion to

a mole fraction basis was used in the subsequent calculation of the

overall mass transfer coefficients presented in Table 6. This con­

version was employed so that the units of the reported overall mass

transfer coefficients v7ould be consistent xd.th literature values.

The effect of agitator speed on column efficiency at constant

feed ratio can be seen with respect to ^^Q^ and N.J.QJJ in Figures 8

and 9 respectively. The column efficiencies are shox m to decrease

in all cases with increased agitator speed. The decrease in column

efficiency, beyond the maximum, with increased agitator speed has

been previously encountered for systems in V7hich flooding was never

reached (18). The milky appearance normally associated vdth adequate

mixing of the liquid in the colxiron led to the use of high agitator

52

^figure B . Raff^^^^^®

^ f f i c i eticy vs. ^ S ' t a t o r Spe ed

53

O

o o

60

55

50

45

40

35

30

g 25 •p

20

15

10

0

o E/R = D E/R =

A E/R =

V E/R =

0.58 1.18

2.39

3.71

L

400 500 600 700 800 900 1000 1100 1200 1300

RPM

Figure 9 - Extract Efficiency vs. Agitator Speed

54

speeds that caused the roilky appearance to be pronounced. The

decrease in efficiency encountered at high agitator speeds suggests

that the milky appearance is not the controlling factor and that

adequate mixing is obtained for this system at lov7er agitator speeds.

Both Figures 8 and 9 contain one point, X"7hich is indicated by its

being circled, that would prevent the E/R =1.18 lines from fol­

lovdng the decreasing pattern. Thus, it is believed that these

points can be omitted in plotting those two lines if the correct

trend is to be established. In Figure 8 the efficiencies increase

with increased feed ratio at each agitator speed until a decrease is

encountered X'7hen the highest feed ratio is reached. The lox7 effi­

ciencies experienced at the higher feed ratio are normally due to

insufficient emulsion breakup in the calming sections of the colximn

due to the increased total flov; through the column. The efficiencies

based on N.j Qp, which are shovTn in Figure 9, are all lox-;er than those

based on N- QR* ^^^ difficulty of extraction xdth respect to trans­

ferring V7ater to the extract phase is seen to be easier than trans­

ferring V7ater from the raffinate phase. Thus, the number of theoretical

transfer units required for the raffinate phase more closely approxi­

mates the actual number of stages present in the colximn.

The effect of the feed ratio on colximn efficiency at constant

agitator speeds is demonstrated vdth respect to ^\^Q-^ and N. Qg in

Figures 10 and 11 respectively. The best feed ratio for the raffinate

phase is shox jn in Figure 10 to be at a value of E/R = 2.40. This

ratio is the same one that proved to be best for product, or raf­

finate, purity in Figure 7. Increased values of feed ratio cause

^5

1J

O <

60

55

50

45

40

35

30 o o

g 25 -P S

20

15

10

0

0.0

0

•

A

V

500 Rm

750 RR4

1000 RPM

1250 RPM

1 1.0 2.0

E/R

3.0 4.0

Figure 10 - Raffinate Efficiency vs. Extract to Raffinate Ratio

60

55

56

50

k5

40

o 500 RPM

o 750 RFM

A 1000 RH^

V 1250 RPM

^ 35 o

o o 30 _

§25 -P

20

15

10

5

0

0.0

Figure 11 - Extract Efficiency vs. Extract to Raffinate Ratio

51

decreased efficiency for the ortract phase at all agitator speeds

as shox jn in Figure 11, Therefore, the best ratio for the extract

phase x 7ould be the lovrest one investigated, or E/R = 0.58. The dif­

ferences in efficiency levels betv7een the raffinate and extract

phases V70Xild be explained in the same manner as for Figures 8 and

9. The same tv7o data points that were previously considered to be

in error are again ignored for plotting purposes and are indicated

by being circled. These points vdll be indicated in a similar man­

ner on all subsequent figures in V7hich they appear.

The effect of total column throughput on colximn efficiency at

constant agitator speeds is shovTn x-dth respect to N.J QT3 and N+Qg in

Figures 12 and 13 respectively. These figures are essentially the

same as Figures 10 and 11 respectively, since the raffinate rate xras

held constant for all runs x hile the extract rate V7as varied to obtain

differences in the feed ratio and total colurtm throughput.

The effect of total column throughput on overall mass transfer

coefficients at constant agitator speeds is shovTn vdth respect to the

raffinate and extract phases in Figures l4 and 15 respectively. The

highest values of the overall mass transfer coefficients occurred

for a total column throughput of 6.5 (ft- /hr ft ) for both the raf­

finate and extract phases. The differences in magnitude betv7een the

overall mass transfer coefficients for the raffinate and extract

phases x as due to the differing resistances to mass transfer vdth

respect to the tv7o phases.

All of the experimental data indicate one particular feed ratio

and the corresponding total colximn throughput as being the best for

58

60

55

50

k5

40

^

I 35 <

o o 30

^« 25 O •«->

^ 20

15

10

5

0

p 500 RPM Q 750 PuPM A 1000 RPM V 1250 RPM

3.0 4.0

I 5.0 6.0 7.0

(E + R)/S (ft /hr ft^)

8.0 9.0

Figure 12 - Raffinate Efficiency vs. Total Column Throughput

59

60

55

50

k5

40

^

5 35 o <

^ 30 o o

Z^ 25 w o •p ^ 20

15

10

5

0

— !

-

,

'

0

J3

A

V

1

1

500 RPM 750 RPM 1000 RPM

1250 RPM

—^-^f^t^~->^^!^^~>~^

1 3.0 4.0

jo

<r

5,0 6.0 7.0 8.0

(E + R)/S (ft /hr ft^)

Figure 13 - Extract Efficiency vs. Total Column Throughput

60

5.0 6.0

(E + R)/S (ft^/hr ft^)

Figure 14 - Raffinate Overall Mass Transfer Coefficient vs. Total Column Throughput

61

26

24

22

20

18

o . <16

S 14 ^

"©-12 H o B

H * 10

w S^8

4

2

0 3.0

o 750 RPM A 1000 RPM V 1250 RPM

i 4.0

i 5.0 6,0 7.0

(E + R)/S (ft /hr ft^)

8,0 9.0

Figure 15 - Extract Overall Mass Transfer Coefficient vs. Total Coluran Throughput

62

product, or r^iffinate, purity and column efficiency. The saiic values

produce the highest overall mass transfer coefficients for both the

raffinate and extract phases. These results indicate that there x.as

an optimxim set of operating conditions vdthin the range of those

studied and that reasonably high product purities vjere possible with

a partially saturated salt solution as the extract. The calculated

overall mass transfer coefficients shoxild further demonstrate the

feasibility of using salting-out as a means of dehydrating organic

liquids on an industrial level.

CHAPTER VIII

SUl'-H-lARY

A solution of 55 percent acetonitrile and 45 percent V7ater V7as

contacted with a partially saturated solution of potassium carbonate

and water in a York-Scheibel colximn for the purpose of removing V7ater

from the acetonitrile-rich phase by salting-out. Four feed ratios

in the region of E / R = O.58 to E / R = 3.71 x rere used along vdth four

agitator speeds of 5OO, 750, 1000 and 1250 RPM to obtain performance

data on the colximn. The raffinate feed rate vras held constant for all

runs causing the total throughput to vary as the feed ratio vras varied.

Steady state vjas assxiraed in all runs x 7hen the composition of the raf­

finate product became constant as determined by chromatographic anal­

ysis. The nximber of theoretical transfer units and overall mass

transfer coefficients were also calculated from the experimental data.

A summary of the performance behavior of the column includes the

follovding observations:

1, Three to four hours were required for the concentration of

the raffinate, or acetonitrile-rich product stream, to reach

a steady state value.

2, The optimum acetonitrile concentration, or product purity,

occurred at a feed ratio of E / R = 2 . 4 for all agitator

speeds used.

3, The column efficiency \<!Q.S found to decrease at feed ratios

betx-reen E / R = 0,58 and E / R = 3.71 vdth increased agitator

speed in the region of 500 to 1250 RPM for both the raffinate

63

64

and extract phases.

4. The maximum column efficiency for all agitator speeds occur­

red at a feed ratio of E/R =2.4 for the raffinate phase and

E/R = 0.58 for the extract phase.

5. The maximum column efficiency for all agitator speeds occur­

red at a total column throughput of 6.5 (ft- /hr ft ) for the

raffinate phase and 3.6 (ft /hr ft ) for the extract phase.

6. The maximum overall mass transfer coefficients for all agi­

tator speeds were observed at a total column throughput of

6.5 (ft- /hr ft ) for both the raffinate and extract phases.

CHAPTER IX

CONCLUSIONS

This study led to the follovdng conclusions on the performance

of the York-Scheibel column xdth the acetonitrile-potassium carbonate-

water system used:

1. Agitator speed had no effect on product purity up to the

optimum practical purity observed,

2. ColuiTin efficiency decreased with increasing agitator speed,

3. The optimum practical product purity, column efficiency, and

overall mass transfer coefficients for the acetonitrile-rich

phase occurred at an extract to raffinate ratio of 2,4,

65

CHAPTER X

RECavMENDATIONS

The most obvious area for investigation vdth the same ternary

system vrould be at lovrer agitator speeds, since efficiencies vrere

found to decrease as agitator speed was increased. Studies of other

salting-out systems could also provide information about similarities

in dehydration ability or performance characteristics. Eventually

it might be possible to obtain performance-oriented correlations

for salting-out systems, or for all extraction systems, involving

the use of York-Scheibel extraction colximns. If ternary diagrams

were available at higher temperatures, the effect of preheating the

extract feed to permit the use of higher salt concentrations could

be investigated. The salt feed concentrations and the column tem­

perature coxild be controlled to eliminate salt crystal deposits

that would clog the system. Another area in vrhich a great deal of

work could be done is in computerized calculation procedures. Curve

fitting the ternary equilibriura data could lead to complete ana­

lytical solution of the number of theoretical transfer units and

mass transfer coefficients, thereby eliminating the tedium normally

associated xdith such calculations.

66

LITERATURE CITED

1. Bird, R, B., Stev7art, W. E., and Lightfoot, E. N.: Transport Phenomena. John Wiley and Sons, Inc., New York, I965.

2. Bull, F. W. and Coli, G. J.: "Graphical Methods as Applied to Extraction Problems." Bulletin of the Virginia Polytechnic Institute. Virginia Polytechnic Institute, Blacksburg, Virginia. (19^9)

3. Dixon, V/. J, ed: Biomedical Computer Programs. University of California Press, Berkeley", I967.

4. Ergin, S.: "Dehydration of Acetonitrile by Salting-Out." pp. k5-l6, M.S. Thesis, Library, Texas Technological College, Lubbock, Tex. (I966)

5. Foust, A. S., Wenzel, L. A., Clximp, C. W., Maus, L., and Anderson, ^* » • Principles of Unit Operations. John Wiley and Sons, Inc., New York, 19Ebl

6. Henley, E, J. and Staffin, H, K,: StagGxd.se Process Design, John Wiley and Sons, Inc,, Nev; York7~T963.'

7. Meissner, H. P. and Stokes, C. A.: "Solvent Dehydration by Salting Out - Prediction of Maximxim Degree of Dehydration." Ind. Eng. Chem. 36: 816-820 (1944).

8. Meissner, H. P., Stokes, C. A., Hunter, C. M., and Morrow, G. M.: "Solvent Dehydration by Salting Oat - Continuous Countercurrent Dehydration." Ind. Eng. Chem. 6: 917-921 (19^).

9. Oliver, E. D.: Diffusional Separation Processes. John Vdley and Sons, Inc., New York,

10. Perry, J. H., ed.: Chemical Engineer's Handbook. McGrav7-Hill Book Company, Inc., New York,"19^3.

11. Raupp, G.: "V/ahl der stationaren Phase fur die qualitative gaschromatographische Analyse." Fresenius Z. Anal. Chem. l64; 134-146 (1958).

12. Renard, J. A, and Oberg, A. G.: "Ternary Systems? Water-Aceto-nitrile-Salts," J. Chem. Eng. Data 10: 152-155 (1965).

13. Renard, J, A, and Heichelheim, H. R.: "Ternary Systems: V/ater-Acetonitrile-Salts." J. Chem. Eng. Data j^: 33-36 (1967).

67

68

1^. Scheibel, E. G. and Othmer, D. F.: "A General Method for Cal-cxaating Diffusional Operations Such as Extraction, Distillation and Gas Absorption." Trans. Am. Inst. Chen, Eng. 38: 339-364 (19^2), •""

15. Scheibel, E. G. and Othmer, D. F.: "Nomographs for Mean Driving Forces in Diffusional Problems." Ind. Eng. Chem. 34: 1200-1208 (19^2). —

16. Scheibel, E. G.: "Fractional Liquid Extraction, Part I." Chem. Eng. Prog. 44: 68I-690 (1948).

17. Scheibel, E. G.: "Fractional Liquid Extraction, Part II." Chem. Eng. Prog. 44: 771-782 (1948).

18. Scheibel, E. G, and Karr, A. E.: "Semicommercial Multistage Extraction Colximn Performance Characteristics," Ind, Eng. Chem. 42: 1048-1057 (1950).

19. Scheibel, E. G.: "Calculation of Liquid-Liquid Extraction Proc­esses." Ind. Eng. Chem. 46: l6-24 (1954).

20. Scheibel, E. G.: "Performance of an Internally Baffled Multi­stage Extraction Column." Am. Inst. Chem. Eng. J. 2: 7^-78 (1956).

21. SoX'dstovrsld., H. and Sndth, VJ.: Mass Transfer Process Calculations. Interscience Publishers, Nev; York7~19^3T"

22. Treybal, R. E.: Liquid Extraction. McGraw-Hill Book Company, Inc., New York, I9S3.

23. Varteressian, K. A, and Fenske, M. R.: "Liquid-Liquid Extraction -Performance of a Packed Extraction Coluirm, Using Continuous Countercurrent Operation." Ind, Eng. Chem. 28: 928-933 (1936).

24. Volk, V7.: Applied Statistics for Engineers. McGrax -Hill Book Company, Inc., 1958.

25. V/eigand, J. H.: "Simplified Calculation of the Number of Trans­fer Units for General Absorption Problems." Trans. Am. Inst. Chem. Eng. 26s 679-682 (19^0).

26. York-Scheibel Instruction Manual on York-Scheibel Liquid-Liquid Extraction Unit XA-s. York Process Equipment Corporation, V/est Orange, Nex7 Jersey

APPENDIX

69

70

N Q M E I N I C L A T U R E

R; = raffinate feed rate (Ib/hr)

R2 = raffinate effluent rate (Ib/hr)

Xj = weight fraction H2O in R^

X^ - vreight fraction H2O in R2

E^ = extract effluent rate (Ib/hr)

E2 = extract feed rate (Ib/hr)

Xg. = weight fraction H2O in E^

Xg = weight fraction H2O in E2

M = total flow rate (Ib/hr)

Xj = vreight fraction H2O in M

H = column height (ft)

XR = weight fraction H2O at any point betv7een Xj and X-^

Xv - vreight fraction H2O at any point betvreen Xg-i and Xpj

N^ = distributed phase flux (lbs/(hr)(ft2))

k = mass transfer coefficient (lbs/(hr)(ft )(v;eight fraction))

XA = total change in concentration (vreight fraction)

kj = raffinate mass transfer coefficient (lbs/(hr)(ft^)(x^7eight

fraction))

k^ = extract mass transfer coefficient (lbs/(hr)(ft )(weight lit

fraction))

Xj^ = vreight fraction H2O in the bulk raffinate

Xg = weight fraction H2O in the bulk extract

Xj . = raffinate interfacial vreight fraction H2O

Xj' = extract interfacial weight fraction H2O

71

% - raffinate overall mass transfer coefficient (lbs/(hr)(ft^)

(weight fraction))

Kg = extract overall mass transfer coefficient (Ibs/(hr)(ft^)

(weight fraction))

X^ - raffinate equilibrixim weight fraction H2O

Xjg - extract equilibrixim vreight fraction H2O

N = total transfer of distributed phase (Ibs/hr)

R = raffinate rate (Ibs/hr)

dA = interfacial transfer area (ft^)

dH = differential column height (ft)

a = interfacial transfer area per unit volume (ft /ft )

S = column cross sectional area (ft2)

N^ = number of transfer units

E.^ = height of a transfer unit (ft)

^tOR " nximber of theoretical transfer units for the raffinate

H- oR = height of a transfer unit for the raffinate (ft)

N-toE ~ number of theoretical transfer units for the extract

^tOE ~ height of a transfer unit for the extract (ft)

* Kj a = raffinate overall mass transfer coefficient (lbs/(hr)(ft-^)

(weight fraction))

* Kp.a = extract overall mass transfer coefficient (lbs/(hr)(ft-^)

(weight fraction))

* Reported overall mass transfer coefficients V7ere converted to

(lb-mole/(hr)(ft^)(mole fraction)) to be consistent vdth

literature values.

72

TABLE 7

TER.NARY BINODAL DATA (13) AT 25^0

CH3CN

wt. ^

1.1 1.4 2.7 ^•7 5.6 6.5 6.8 8.1 8.8 9.1

10.1 10.3 10.9 13.2 13 .^ 15.^ 17.2 17.9 20.5 21.3 24.6 27.6 29.7 36.7 38.1 39.^ ^7.9 ^9.3 51.9 53.5 53.6 5^.8 61.6 64.1 66.1 82.0 96.2 99.0

K2CO3

Wt. $

42.0 39.7 31.2 23.4 21.5 19.9 19.5 17.^ 16.0 15.2 14,4 13.9 13.2 11.2 11.3 9.6 8.3 7.8 6.6 6.4 5.0 4.0 3.7 2.3 2.6 2.0 1.4 1.3 1.0 1.0 1.0 0.9 0.6 0.4 0.3 0.0 0.0 0.0

H2O Wt. $

56.9 58.9 66.1 71.9 72«9 73.6 73.7 7^.5 75.2 75.7 75.5 75.8 75.9 75.6 75.3 75.0 7^.5 7^.3 72.9 72.3 70.4 68.4 66.6 61.0 59.3 58.6 50.7 49.4 47.1 k5.5 45.4 ^ . 3 31.8 35.5 33.6 18.0 3.8 1.0

TABLE 8

73

H20

wt, $

* 1.0 * 3.8

5.0 7.5

11,4 14.3 16,2

*18.0 20,5 21,9 23.5 25.9 27.8 31.0 32.7 3^,6 37.7 41.2 k5.1 51.7 55.4 m^ m/ •

59.6

Raffinate

CHoQ^-rich phase

CH3CN

Wt, i

99.0 96.2 95.0 92.5 88.6 85.7 83.8 82.0 79.^ 78.0 76.3 73.9 72.0 68.8 67.0 65.0 61.8 58.0 53.3 47.0 43.0 38.4

TIE LINE DATA

K2CO3

Wt. $

0.0 0.0 0.0 0,0 0,0 0.0 0.0 0.0 0.1 0.1 0.2 0.2 0.2 0.2 0.3 0.4 0.5 0.8 1.0 1.3 1.6 2.0

Extract

K2C0o-rich phase

H2O Wt. ^

56.9 58.2 59.7 61.5 64.1 66.4 67.2 68.8 69.8 70.8 71.6 72.7 73.7 7^.6 75.0 75.5 75.8 75.2 74.8 72.5 70.5 68.0

CH3CN

Wt. ^

1.1 1.2 1.3 1.5 1.9 2.3 2.8 3.2 3.7 4.2 ^.7 5>5 6.3 7.9 9.0

10.0 12.2 14.8 17.2 21.5 24.5 28,0

K2CO3

Wt. $

42.0 40.6 39.0 37.0 34.0 31.3 30.0 28.0 26.5 25.0 23.7 21.8 20,0 17.5 16.0 14.5 12.0 10.0 8.0 6.0 5.0 4.0

* Renard and Oberg (13)

74

TABLE 9

Al^ALYSIS OF QUMTITATIVE CHRCMATOGR.\PHIC DATA FOR ACETONITRILE-WATER SOLUTIONS

No.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

H2O Avg. ^'/o Peak Ht.

4.65

6.88

13.64

13.30

16.05

19.95

22.51

24.27

27.20

27.18

35.59

41.84

47.62

54.3^

58.85

61.32

66.08

69.03

72.71

75.78

77.73

Std. Dev.

0.05

0.10

0.13

0.06

0.15

0.04

0,08

0,10

0,07

0.15

0.29

0,12

0.27

0.14

0.22

0.10

0.07

0.15

0.11

0.13

0.16

H2O Calc. Wt. $

1.88

2.14

3.53

3.^'

4.26

5.67

6.77

7.60

9.12

9.11

14.44

19.32

24.54

31.^5

36.61

39.60

^5.72

50.32

56.56

61.77

65.07

H2O Actual Wt. $

1.50

2.05

2.70

3.84

5.13

6.49

6,92

7.5^

9.23

10.15

14.86

19.81

24.26

30.96

35.53

39.7^

45.02

50.98

56.49

60.47

65.76

15

TABLE 9—Continued

No.

22

23

24

25

26

27

H2O Avg. ^ Peak Ht.

80.90

83.20

86.85

89.23

92.65

96.11

Std. Dev,

0.10

0.02

0.09

0.06

0.05

0.02

H2O Calc. Wt. ^

70.46

7^.35

80.5^

84.57

90.38

96.25

H2O Actual Wt. i

69.56

74.90

80.66

85.4^

90.83

95.54

100

95

90

85

80

75

70

o 65 CO -

60

55

50

45

40

35

30

25

20

15

10

-1 . . . ..-2 Y = -6.675 X lO'- ^ 1.696 X 10

0.9987

tiO •H ©

M ©

© t> -P

H © Pi -P © o u o a.

0

76

= 1.659 X 10"^ -f- 1.009 X

-4 7 10 X"

= 0.9989

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0,9 1.0

Weight Fraction (H2O)

Figure 16 - Chromatographic Calibration Curve

77

L _ J I I I I I I I I

o ON

o CO

o

o NO

o VA T ^

o -=

o c^ T-l

o © CO -p r-l 0

t H W u ©

O P O Ti ^ H

•H

o d ON T H

o CO

o

o NO

o VA

O

o 0 ^

o CO

© -p

h o El

© W c\5

Xi

o CO

I

(A W o ©

Xi •p u o

©

O •H P n5 U

X» •H

• ^ O

u ©

-p ©

-p o

I

T H

© u

O CJN 00 CN- NO VA -ci- 0^ CO

(p^og ss^xo) 2uxq.q.8S j:8q.8mT?:(.0H

o

78

© -p

w ©

-p •H H •H H H •H

© -P a

o

© w a

Xi Pu,

o CO I r^ o o CO

©

o

<H

©

t 3 c O

•H -P oJ

•H

li O ©

-P o

p o ft;

© bJD

o

CO t H

© u a El

(pTJog SSCX9) SuTiq.' cig joci-oira^oa

__ I ^ l___l^.^l l.

J I 1 _ _ L VA -:t CA CO

O ( A

VA CO

o CO

79

ON tN- NO

VA t H t H

110

VA O t H

100

VA O N

O O N

VA 0 0

O CO

VA C--

O Cv-

VA NO

O NO

VA VA

O VA

vn>

O

VA c->

O CA

VA CO

<<—< ©

- P

g •H ^

w U ©

-P •H H •H H H

^•-^ © -P

O

CTH

Ph

ase

o CO

1 O o CO

© xi -p ^1 o

t H

© > u 8 o

•H -P

U rO •H

o

u © -p ©

-p O

© txO

§

Xi

•H

1

ON t H

©

u a s

(p-eog lOGq.g) Suxq.q.oc j9q.exin2q.0-iT

Tf^.vjP^.