Page 1
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
^\
Page 2
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.
Page 3
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
Page 4
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
Page 5
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
Page 6
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
Page 7
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
Page 8
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
Page 9
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
Page 10
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.
Page 11
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
Page 12
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
Page 13
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
Page 14
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
Page 15
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
Page 16
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
Page 17
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
Page 18
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.
Page 19
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
Page 20
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.
Page 21
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
Page 22
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,
Page 23
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
Page 24
18
o
•H
W
c •H -P U o
&
o
!>i.5
Page 25
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
Page 26
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
Page 27
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
Page 28
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
Page 29
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)
Page 30
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
Page 31
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
Page 32
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 ))
Page 33
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
Page 34
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.
Page 35
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
Page 36
30
Figure 6 - Experimental Apparatus
Page 37
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
Page 38
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.
Page 39
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
Page 40
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
Page 41
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
Page 42
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.
Page 43
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
Page 44
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
Page 45
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
Page 46
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
Page 47
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.
Page 48
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
Page 49
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
Page 50
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
Page 51
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
Page 52
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
Page 53
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
Page 54
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
Page 55
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
Page 56
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
Page 57
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
Page 58
52
^figure B . Raff^^^^^®
^ f f i c i eticy vs. ^ S ' t a t o r Spe ed
Page 59
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
Page 60
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
Page 61
^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
Page 62
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
Page 63
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
Page 64
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
Page 65
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
Page 66
60
5.0 6.0
(E + R)/S (ft^/hr ft^)
Figure 14 - Raffinate Overall Mass Transfer Coefficient vs. Total Column Throughput
Page 67
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
Page 68
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.
Page 69
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
Page 70
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.
Page 71
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
Page 72
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
Page 73
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
Page 74
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
Page 76
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
Page 77
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.
Page 78
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
Page 79
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)
Page 80
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
Page 81
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
Page 82
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
Page 83
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
Page 84
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
Page 85
__ 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