Liquid-Liquid Extraction With Ternary Systems: 8.0 Instructional Objectives

Trim Size: 8.5in x 11in Seader c08.tex V2 - 10/16/2015 10:49 A.M.
Page 231
Chapter 8
Liquid–Liquid Extraction with Ternary Systems

§8.0 INSTRUCTIONAL OBJECTIVES
After completing this chapter, you should be able to:

• List situations where liquid–liquid extraction is preferred to distillation.
• Define the distribution (partition) coefficient and show its relationship to activity coefficients and the selectivity of a
solute between carrier and solvent.
• Make a preliminary selection of a solvent using group-interaction rules.
• Distinguish, for ternary mixtures, between Type I and Type II systems.
• For a specified recovery of a solute, calculate with the Hunter and Nash method, using a triangular diagram, minimum
solvent requirement and equilibrium stages for multistage ternary liquid–liquid extraction.
• Design a cascade of mixer-settler units based on mass-transfer considerations.
• Make an estimate of the size of an extraction column.
In liquid–liquid extraction (solvent extraction), a liquid partially soluble, but solute A is completely or substantially
feed of two or more components is contacted with a second soluble in S. During extraction, mass transfer of A from
liquid phase, called the solvent, which is immiscible or the feed to the solvent occurs, with less transfer of C to the
only partly miscible with one or more feed components and solvent, or S to the feed. Nearly complete transfer of A to
completely or partially miscible with one or more of the other the solvent is seldom achieved in just one equilibrium stage.
feed components. The solvent is selected to partially dissolve In practice, a number of stages are used in one- or two-section
certain species of the liquid feed, effecting at least a partial countercurrent cascades. Most commonly, the stages are
separation of the feed components. The solvent may be a pure provided by mixer-settler units, columns with mechanical
compound or a mixture. If the feed is an aqueous solution, an agitation, or centrifugal devices. A wealth of information on
organic solvent is used; if the feed is organic, the solvent is liquid–liquid extraction is given by Frank et al. in Section 15
often water. Important exceptions occur in metallurgy for the of Perry’s Chemical Engineers’ Handbook [2].
separation of metals and in bioseparations for the extraction
from aqueous solutions of proteins. Solid–liquid extraction
Industrial Example
(leaching) involves the recovery of substances from a solid
by contact with a liquid solvent, such as the recovery of oil Acetic acid is produced by methanol carbonylation, acetalde-
from seeds by an organic solvent. hyde oxidation, or as a byproduct of cellulose acetate manufac-
According to Derry and Williams [1], extraction has been ture. In all cases, a mixture of acetic acid (n.b.p. = 118.1∘ C)
practiced since the time of the Romans, who used molten lead and water (n.b.p. = 100∘ C) is separated to give glacial acetic
to separate gold and silver from molten copper by extraction. acid (99.8 wt% min). When the mixture contains less than
This was followed by the discovery that sulfur could selec- 50% acetic acid, separation by distillation is expensive
tively dissolve silver from an alloy with gold. However, it was because of the high heat of vaporization of large amounts of
not until the early 1930s that L. Edeleanu invented the first water. Thus, a solvent-extraction process becomes attractive.
large-scale extraction process, which involved the removal Figure 8.1 shows an implementation of extraction, where two
of aromatic and sulfur compounds from liquid kerosene distillation operations are required to recover solvent for recy-
using liquid sulfur dioxide at 10 to 20∘ F. This resulted in cle. These additional separation operations are common to
a cleaner-burning kerosene. Liquid–liquid extraction has extraction processes. A feed of 30,260 lb∕h of 22 wt% acetic
grown in importance since then because of the demand for acid in water is sent to a single-section extraction column
temperature-sensitive products, high purity, and availability operating at ambient conditions, where it is contacted with
of better equipment and solvents with higher selectivity. 71,100 lb∕h of ethyl-acetate solvent (n.b.p. = 77.1∘ C), satu-
This chapter introduces liquid–liquid extraction by rated with water. The low-density, solvent-rich stream, called
treating a ternary system consisting of two miscible feed the extract, exits from the top of the extractor with 99.8%
components—the carrier, C, and the solute, A—plus solvent, of the acetic acid in the feed. The high-density, carrier-rich
S, a pure compound. Components C and S are at most only stream, called the raffinate, exiting from the extractor bottom,
231
Trim Size: 8.5in x 11in Seader c08.tex V2 - 10/16/2015 10:49 A.M. Page 232
232 Chapter 8 Liquid–Liquid Extraction with Ternary Systems
Makeup solvent Ethyl acetate-rich
Reflux Two liquid

phases
Decanter
Extract
Ethyl acetate 67,112 Water-rich
Water 6,660
Acetic acid 6,649
Distillation
Feed Glacial
Acetic acid 6,660 acetic
Water 23,600 acid
(99.8%
min)
Recycle
solvent
Liquid–liquid
Water 2,500
extraction
Ethyl
acetate 68,600
Ethyl acetate-rich
Raffinate
Distillation
Ethyl acetate 1,488
Water 19,440
Acetic acid 11
Note: All flow rates are in lb/h Wastewater
Figure 8.1 Liquid–liquid extraction process for recovering acetic acid.
contains only 0.05 wt% acetic acid. The extract is sent to a column, which contains 40 rotating disks, has an inside
distillation column, where glacial acetic acid is the bottoms diameter of 5.5 ft and a total height of 28 ft. The disks are
product. The overhead vapor, which is rich in ethyl acetate but rotated at 60 rpm by a 5-hp motor. Since the 1930s, thousands
also contains appreciable water vapor, splits into two liquid of similar extraction columns, with diameters ranging up to at
phases when condensed. These are separated in the decanter least 25 ft, have been built. As discussed in §8.1, a number of
by gravity. The lighter ethyl acetate-rich phase is divided into competitive extraction devices are suitable for the process in
reflux and solvent recycle to the extractor. The water-rich Figure 8.1.
phase from the decanter is sent, together with the raffinate
from the extractor, to a second distillation column, where Liquid–liquid extraction is a reasonably mature operation,
wastewater is the bottoms product and the ethyl-acetate-rich although not as mature or as widely applied as distillation,
overhead is recycled to the decanter. Makeup ethyl-acetate absorption, and stripping. Procedures for determining the
solvent is provided for solvent losses to the glacial acetic acid number of equilibrium stages to achieve a desired solute reco-
and wastewater. very are well established. However, for liquid–liquid phase
Six equilibrium stages are required to transfer 99.8% of equilibrium, no simple limiting theory, like Raoult’s law for
the acetic acid from feed to extract using a solvent-to-feed vapor–liquid equilibrium, exists. Frequently, experimental
mass ratio of 2.35; the recycled solvent is water-saturated. data are preferred over predictions based on activity-coefficient
A rotating-disk contactor (RDC), described in §8.1.5, is correlations. Such data can be correlated and extended by
employed within the column to disperse the organic-rich activity-coefficient equations such as NRTL or UNIQUAC,
phase into droplets by horizontal, rotating disks, while the discussed in §2.7. Also, considerable laboratory effort may
water-rich phase is continuous throughout the column. The be required to find an optimal solvent. A variety of industrial
§8.1 Equipment for Solvent Extraction 233
Table 8.1 Representative Industrial Liquid–Liquid Extraction 2. Removal of a contaminant present in small concentra-
Processes tions, such as a color former in tallow or hormones in
Solute Carrier Solvent(s) animal oil.
3. A high-boiling component present in relatively small
Acetic acid Water Ethyl acetate,
Isopropyl acetate
quantities in an aqueous waste stream, as in the recovery
Aconitic acid Molasses Methyl ethyl ketone of acetic acid from cellulose acetate.
Ammonia Butenes Water 4. Recovery of temperature-sensitive materials, where ex-
Aromatics Paraffins Diethylene glycol, traction may be less expensive than vacuum distillation.
Furfural, Sulfur
5. Separation of mixtures according to chemical type
dioxide
Asphaltenes Hydrocarbon oil Furfural
rather than relative volatility.
Benzoic acid Water Benzene 6. Separation of close-melting or close-boiling liquids,
Butadiene 1-Butene aq. Cuprammonium where solubility differences can be exploited.
acetate 7. Separation of mixtures that form azeotropes.
Ethylene Methyl ethyl ketone Brine liquor
cyanohydrins The key to an effective extraction process is a suitable
Fatty acids Oil Propane solvent. In addition to being stable, non-toxic, inexpensive,
Formaldehyde Water Isopropyl ether and easily recoverable, a solvent should be relatively immis-
Formic acid Water Tetrahydrofuran cible with feed components other than the solute, and have
Glycerol Water High alcohols a different density from the feed to facilitate phase separa-
Hydrogen peroxide Anthrahydroquinone Water tion by gravity. It must have a high affinity for the solute,
Methyl ethyl ketone Water Trichloroethane
from which it can be easily separated by distillation, crystal-
Methyl borate Methanol Hydrocarbons
Naphthenes Distillate oil Nitrobenzene, Phenol
lization, or other means. Ideally, the distribution (partition)
Phenol Water Benzene, coefficient (2-19) for the solute between the liquid phases
Chlorobenzene should be greater than one, or a large solvent-to-feed ratio
Penicillin Broth Butyl acetate will be required. When the degree of solute extraction is
Sodium chloride aq. Sodium Ammonia not particularly high and/or when a large extraction factor
hydroxide (4-35) can be achieved, an extractor will not require many
Vanilla Oxidized liquors Toluene stages. This is fortunate because mass-transfer resistance in
Vitamin A Fish-liver oil Propane liquid–liquid systems is high. In this chapter, equipment for
Vitamin E Vegetable oil Propane liquid–liquid extraction is discussed and equilibrium- and
Water Methyl ethyl ketone aq. Calcium chloride rate-based calculation procedures are presented for extraction
in ternary systems. Use of graphical methods is emphasized to
gain a visual perspective. Extraction calculations with process
equipment is available, making it necessary to consider simulators are presented in Chapter 10.
alternatives before making a final selection. Unfortunately,
no generalized capacity and efficiency correlations are avail-
able for all equipment types. Often, equipment vendors and §8.1 EQUIPMENT FOR SOLVENT
pilot-plant tests must be relied on to determine equipment size. EXTRACTION
The petroleum industry represents the largest-volume
application for liquid–liquid extraction. Extraction processes Equipment similar to that used for absorption, stripping, and
are well suited to the petroleum industry because of the need distillation is sometimes used for extraction, but such devices
to separate heat-sensitive liquid feeds according to chemical are inefficient unless interfacial tension and liquid viscosities
type (e.g., aliphatic, aromatic, naphthenic) rather than by are low and differences in phase density are high. Generally,
molecular weight or vapor pressure. Table 8.1 lists some rep- mechanically agitated or centrifugal devices are preferred,
resentative industrial extraction processes. Other applications especially if many equilibrium stages are required. During
exist in the biochemical industry, including the separation of passage through extraction equipment, one phase is the dis-
antibiotics and recovery of proteins from natural substrates; in persed phase (discontinuous phase) in the form of droplets
the recovery of metals, such as copper from ammoniacal leach and, the other phase is the continuous phase. In static extrac-
liquors; in separations involving rare metals and radioac- tion columns of the spray, packed, and sieve tray type, it is
tive isotopes from spent-fuel elements; and in the inorganic preferred to disperse the phase of higher entering volumetric
chemical industry, where high-boiling constituents such as flow rate, unless the other phase has a high viscosity.
phosphoric acid, boric acid, and sodium hydroxide need to
be recovered from aqueous solutions. In general, extraction is §8.1.1 Mixer-Settlers
preferred over distillation for:
In mixer-settlers, the two liquid phases are first mixed in
1. Dissolved or complexed inorganic substances in organic a vessel (Figure 8.2) by one of several types of impellers
or aqueous solutions. (Figure 8.3) and then separated in a second vessel by
Variable-speed
drive unit Light
Slotted Tap for liquid
impingement scum out
Turbine baffle
Emulsion out
Compartment Emulsion
spacer in
Rotating Heavy
plate liquid out
Figure 8.4 Horizontal gravity-settling vessel.
The settling step is by gravity in a settler (decanter). In

Figure 8.4, a horizontal vessel, with an impingement baffle to
Feed in prevent the jet of the entering two-phase dispersion (emulsion)
from disturbing the gravity-settling process, is used. Vertical
Figure 8.2 Compartmented mixing vessel with turbine
and inclined vessels are also common. A major problem in
agitators.
settlers is emulsification in the mixing vessel, which may
occur if the agitation is so intense that the dispersed droplet
size falls below 1 to 1.5 μm (micrometers). When this hap-
pens, coalescers, separator membranes, meshes, electrostatic
forces, ultrasound, chemical treatment, or other ploys are
required to speed settling. If the phase-density difference is
small, the rate of settling can be increased by substituting
centrifugal for gravitational force. Many single stage and
(a) (b) (c)
multistage mixer-settler units are available as described by
Bailes, Hanson, and Hughes [3] and Lo, Baird, and Hanson
[4]. As shown in Figure 8.5, the mixer and settler vessels can
be combined into one unit or they can be arranged vertically,
as in the Lurgi Tower Extractor. Often the settler volume is
larger than the mixer volume.
(d) (e)
Figure 8.3 Some common types of mixing impellers:

§8.1.2 Spray Columns
(a) marine-type propeller; (b) centrifugal turbine; (c) pitched-blade The simplest and one of the oldest extraction devices is the
turbine; (d) flat-blade paddle; (e) flat-blade turbine. spray column. Either the heavy phase or the light phase can be
dispersed, as seen in Figure 8.6. The droplets of the dispersed
gravity-induced settling (Figure 8.4). Any number of phase are generated at the inlet, usually by spray nozzles.
mixer-settler units may be connected together to form a Because of the lack of column internals, combined volumetric
multistage countercurrent cascade. However, floor space
can be a major factor. During mixing, one of the liquids is
dispersed in the form of small droplets in the other liquid.
The dispersed phase may be either the heavier or the lighter
phase. The mixing occurs in an agitated vessel with sufficient
residence time that a reasonable approach to equilibrium Mixer-Settler stage
(e.g., 80 to 90%) is achieved. The vessel may be compart-
mented as in Figure 8.2. If dispersion is easily realized Aqueous out
and equilibrium rapidly approached, as with liquids of low Mixing
interfacial tension and viscosity, mixing can be achieved by section
Settling section
(1) impingement in a jet mixer; (2) turbulence in a noz-
zle mixer, orifice mixer, or other in-line mixing device; Organic out
(3) shearing action if both phases are fed simultaneously into
Aqueous and organic in
a centrifugal pump; or (4) injectors, wherein the flow of one
liquid is induced by another. Figure 8.5a Combined Mixer-Settler Unit
Impeller
Heavy phase Light phase distillation and absorption are suitable for liquid–liquid
shaft extraction; however, choice of packing material is more criti-
Light
phase cal. For best performance, the packing should be preferentially
wetted by the continuous phase. Throughput, especially with
Heavy
phase newer packings, is large. Because of backmixing, the HETS is
generally large, making packed columns suitable only when
Light
Interface few equilibrium stages are needed.
phase
Heavy §8.1.4 Sieve-Tray Columns

phase
Sieve trays reduce axial mixing and promote a stagewise
Light
phase type of contact. The dispersed phase, which is analogous
to vapor bubbles in distillation, flows up the column, with
Heavy redispersion at each tray. The heavy phase is continuous,
phase flowing at each stage through a downcomer, and then across
the tray like a liquid in a distillation tower. If the heavy phase
is dispersed, upcomers are used for the light phase. Columns
Light phase Heavy phase have been built with diameters larger than 4.5 m. Holes
Figure 8.5b Lurgi Tower Extractor. from 0.32 to 0.64 cm (1∕8 to 1∕4 inches) in diameter, spaced
1.27 to 1.81 cm apart are used, and tray spacings are closer
than in distillation—10 to 15 cm for low-interfacial-tension
Light liquids. Plates are usually built without outlet weirs on the
liquid downspouts.
Light
liquid If designed and operated properly, extraction rates in sieve-
tray columns are high because the dispersed-phase droplets
Heavy Heavy
liquid liquid
coalesce and re-form on each tray. This destroys concentra-
tion gradients, which develop if a droplet passes through the
entire column undisturbed. Sieve-tray extractors are subject to
the same limitations as distillation columns: flooding, entrain-
ment, and, to a lesser extent, weeping. An additional problem
is scum formation at phase interfaces due to small amounts of
impurities.
Light Light
liquid liquid
§8.1.5 Columns with Mechanically
Assisted Agitation
Heavy Heavy
liquid liquid
If (1) interfacial tension is high, (2) density difference between
liquid phases is low, and/or (3) liquid viscosities are high, then
(a) (b)
gravitational forces are inadequate for proper phase dispersal
Figure 8.6 Spray columns: (a) light liquid dispersed, heavy liquid and turbulence creation. In that case, mechanical agitation is
continuous; (b) heavy liquid dispersed, light liquid continuous. necessary to increase interfacial area per unit volume, thus
decreasing mass-transfer resistance. For packed and plate
columns, agitation can be provided by an oscillating pulse to
throughputs can be large, depending upon phase-density the liquid, either by mechanical or pneumatic means. Pulsed,
difference and phase viscosities. As in gas absorption, axial perforated-plate columns find considerable application in
dispersion (backmixing) in the continuous phase limits the nuclear industry. The most prevalent agitated columns
these devices to applications where only one or two stages are those that employ rotating agitators driven by a shaft
are required. extending axially through the column. The agitators create
shear mixing zones, which alternate with settling zones. Nine
of the more popular mechanically-agitated devices are shown
§8.1.3 Packed Columns
in Figure 8.7a–i.
Axial dispersion in a spray column can be reduced, but not Agitation can also be induced in a column by moving
eliminated, by packing the column. This also improves mass plates back and forth in a reciprocating motion (Figure 8.7j)
transfer by breaking up large drops to increase interfacial or in a novel horizontal contactor (Figure 8.7k). These devices
area and promote mixing in drops by distorting droplet shape. answer the 1947 plea of Fenske, Carlson, and Quiggle [7] for
With the exception of Raschig rings [5], the packings used in equipment that can efficiently provide large numbers of stages
Rotating shaft
Light liquid
out Heavy Outer
liquid horizontal
in baffle
Wire-mesh
packing Inner
Feed
horizontal
(if operated
Turbine baffle
for fractional
agitator
extraction) Turbine
Wire-mesh impeller
packing
Light liquid in Tie rod
Heavy
liquid
out
(a) (b)
Motor
Light
Heavy liquid out
liquid in Light liquid out
Heavy
Feed if liquid in
operated for
fractional Perforated
extraction distributor
Baffle
Impeller
Tie rod
Compartment
baffle
Light
liquid in
Heavy liquid out Upper inlet port
Flat impeller Figure 8.7 Commercial extractors with
Flow- Perforated Light mechanically assisted agitation:
control packing Heavy liquid in (a) Scheibel column—first design;
plates liquid (b) Scheibel column—second design;
out
(c) Scheibel column—third design;
Lower inlet port Rotating shaft
(d) Oldshue–Rushton (Mixco) column;
(c) (d) (Continued)
in a device without large numbers of pumps, motors, and Scheibel [9] added outer and inner horizontal annular baffles
piping. They stated, “Despite . . . advantages of liquid–liquid (Figure 8.7b) to divert the vertical flow in the mixing zone
separational processes, the problems of accumulating 20 or and promote mixing. For systems with high interfacial tension
more theoretical stages in a small compact and relatively and viscosities, the wire mesh is removed. The first two
simple countercurrent operation have not yet been fully Scheibel designs did not permit removal of the agitator shaft
solved.” In 1946, it was considered impractical to design for inspection and maintenance. Instead, the entire internal
for more than seven stages, which represented the number assembly had to be removed. To permit removal of just the
of mixer-settler units in the only large-scale, commercial, agitator assembly shaft, especially for large-diameter columns
solvent-extraction process in use. (e.g., >1.5 m), and allow an access way through the column
Perhaps the first mechanically agitated column of impor- for inspection, cleaning, and repair, Scheibel [10] offered a
tance was the Scheibel column [8] in Figure 8.7a, in which third design, shown in Figure 8.7c. Here the agitator assembly
liquid phases are contacted at fixed intervals by unbaffled, shaft can be removed because it has a smaller diameter than
flat-bladed, turbine-type agitators (Figure 8.3) mounted on the opening in the inner baffle.
a vertical shaft. In the unbaffled separation zones, located The Oldshue–Rushton extractor [11] (Figure 8.7d) consists
between the mixing zones, knitted wire-mesh packing prevents of a column with a series of compartments separated by annu-
backmixing between mixing zones, and induces coalescence lar outer stator-ring baffles, each with four vertical baffles
and settling of drops. The mesh material must be wetted by attached to the wall. The centrally mounted vertical shaft
the dispersed phase. For larger-diameter installations (>1 m), drives a flat-bladed turbine impeller in each compartment.
Variable-speed drive
Light liquid
outlet
Heavy Settling zone
liquid
inlet Interface Contact zone
Transport zone
Stator ring Shell
Stator
Rotor disk
Light
liquid Agitator
inlet Settling zone
Heavy
liquid
outlet
(e) (f) (g)
Variable-speed drive
Light
Heavy phase out
phase in
Light N
phase in Figure 8.7 (Continued)
(i) (e) rotating-disk-contactor (RDC);
(f) asymmetric rotating-disk contactor
Heavy (ARD); (g) section of ARD contactor;
phase out
(h) Kuhni column; (i) flow pattern in
(h) Kuhni column.
A third type of column with rotating agitators that appeared drop size, the rotor speed can be continuously varied over a
about the same time as the Scheibel and Oldshue–Rushton wide range.
columns is the rotating-disk contactor (RDC) [12,13] A modification of the RDC concept is the asymmet-
(Figure 8.7e), an example of which is described at the ric rotating-disk contactor (ARD) [14], which has been in
beginning of this chapter and shown in Figure 8.1. It is an industrial use since 1965. As shown in Figure 8.7f, the contac-
extensively used device worldwide [4]. Horizontal disks, tor consists of a column, a baffled stator, and an offset multi-
mounted on a centrally located rotating shaft, are the agitation stage agitator fitted with disks. The asymmetric arrangement,
elements. The ratio of disk diameter to column diameter is shown in more detail in Figure 8.7g, provides contact and
0.6. The distance, H in m, between disks depends on column transport zones that are separated by a vertical baffle, to which
diameter, DT , in m, according to H = 0.13(DT )0.67 . Mounted is attached a series of horizontal baffles. This design retains the
at the column wall are annular stator rings with an opening efficient shearing action of the RDC, but reduces backmixing
larger than the agitator-disk diameter, typically 0.7 of DT . because of the separate mixing and settling compartments.
Thus, the agitator assembly shaft is easily removed from the Another extractor based on the Scheibel concept is the
column. Because the rotational speed of the rotor controls the Kuhni extraction column [15] in Figure 8.7h, where the
Counterweight
Connecting Variable speed
rod drive
Eccentric
shaft
Seal
Light phase
Stub outlet
shaft
Heavy phase
feed Spider plate
sparger
Center shaft
and spacers
Metal baffle
plate Tie rods
Perforated and spacers
plate (k)
Teflon baffle
Light phase plate
feed sparger
Heavy phase Figure 8.7 (Continued) ( j) Karr

outlet
reciprocating-plate column (RPC);
(k) Graesser raining-bucket (RTL)
(j) extractor.
column is compartmented by a series of stator disks made Another novel device for providing agitation is the
of perforated plates. The distance, H in m, between stator Graesser raining-bucket contactor (RTL), developed in the
disks depends on column diameter, DT in m, according late 1950s [4] primarily for processes involving liquids of
to 0.2 < H < 0.3 (DT )0.6 . A centrally positioned shaft has small density difference, low interfacial tension, SX and a
double-entry, radial-flow, shrouded-turbine mixers, which tendency to form emulsions. Figure 8.7k shows a series of
promote, in each compartment, the circulation action shown disks mounted inside a shell on a horizontal, rotating shaft
in Figure 8.7i. The ratio of turbine diameter-to-column diam- with horizontal, C-shaped buckets fitted between and around
eter ranges from 0.33 to 0.6. For columns of diameter greater the periphery of the disks. An annular gap between the disks
than 3 m, three turbine-mixer shafts on parallel axes are and the inside shell periphery allows countercurrent, longi-
normally provided to preserve scale-up. tudinal flow of the phases. Dispersing action is very gentle,
Rather than providing agitation by impellers on a vertical with each phase cascading through the other in opposite
shaft or by pulsing, Karr [16, 17] devised a reciprocating, directions toward the two-phase interface, which is close to
perforated-plate extractor column in which plates move up and the center.
down approximately 27 times per second with a 6.5–25 mm When fast extraction rates are needed as in bioseparations,
stroke. This uses less energy than pulsing the entire volume high-speed centrifugal extractors are favored, such as the Pod-
of liquid. Also, the close spacing of the plates (25–50 mm) bielniak (POD) extractor, used for penicillin extraction [19].
promotes high turbulence and minimizes axial mixing, thus They provide residence times as short as 10 s, and small liquid
giving high mass-transfer rates and low HETS. The annular holdups. PODs can separate liquid phases of density differ-
Teflon baffle plates in Figure 8.7j are placed in the plate stack ences as small as 0.01 g∕cm3 without emulsification.
to minimize axial mixing. The perforated plates use large
holes (typically 9/16-inch diameter) and a high hole area
§8.1.6 Comparison of Industrial
(typically 58%). The central shaft, which supports both sets of
plates, is reciprocated by a drive at the top of the column. Karr
Extraction Columns
columns are particularly useful for bioseparations because Maximum loadings and sizes for industrial extraction columns,
residence time is reduced, and they can handle systems that as given by Reissinger and Schroeter [5, 20] and Lo et al. [4],
tend to emulsify and feeds that contain particulates. are listed in Table 8.2. As seen, the Lurgi tower, RDC, and
A modification of the Karr column is the vibrating-plate Graesser extractors have been built in very large sizes. Com-
extractor (VPE) of Prochazka et al. [18], which uses per- bined volumetric throughputs per unit cross-sectional area
forated plates of smaller hole size and smaller percent hole are highest for the Karr extractor and lowest for the Graesser
area. The small holes provide passage for the dispersed phase, extractor. Table 8.3 lists the advantages and disadvantages of
while one or more large holes on each plate provide passage the various types of extractors, and Figure 8.8 shows a selec-
for the continuous phase. Some VPE columns have uniform tion scheme for commercial extractors. For example, if only a
motion of all plates; others have two shafts for countermotion small number of stages is required, a set of mixer-settler units
of alternate plates. might be selected. If more than five theoretical stages, a high
§8.2 General Design Considerations 239
Table 8.2 Maximum Size and Loading for Commercial §8.2 GENERAL DESIGN CONSIDERATIONS
Liquid–Liquid Extraction Columns
Liquid–liquid extraction involves more design variables than
Maximum Combined Maximum distillation. To determine stages, one of the three cascade
Liquid Throughout, Column arrangements in Figure 8.9, or an even more complex arrange-
Column Type m3 ∕m2 -h Diameter, m
ment, must be selected. Packed-column configurations are
Lurgi tower 30 8 shown in Figure 8.9, but other extraction equipment may be
Pulsed packed 20 3 preferred. The single-section cascade of Figure 8.9a, which
Pulsed sieve tray 25 3 is similar to that used for absorption and stripping, will
Scheibel 25 3 transfer solute in the feed to the solvent. The two-section
RDC 40 8 cascade of Figure 8.9b is similar to that used for distillation.
ARD 25 5 Solvent enters at one end and reflux, derived from the extract,
Kuhni 40 3 enters at the other end. The feed enters in between. With two
Karr 40 3 sections, depending on solubilities, it is sometimes possible to
Graesser 10 7 achieve a separation between two feed components; if not, a
Above data apply to systems of:
dual-solvent arrangement with two sections, as in Figure 8.9c,
with or without reflux at the ends, may be advantageous. For
1. High interfacial surface tension (30 to 40 dyne∕cm).
configurations 8.9b and 8.9c, calculations should be made
2. Viscosity of approximately 1 cP.
by a process simulator, as discussed in Chapter 10. For the
3. Volumetric phase ratio of 1:1. configuration of Figure 8.9a, it is useful and instructive to
4. Phase-density difference of approximately 0.6 g∕cm3 . make the graphical calculations described in this chapter.
Operative factors are:
throughput, and a large load range (m3 ∕m2 -h) are needed, and 1. Entering feed flow rate, composition, temperature, and
floor space is limited, an RDC or ARD contactor should be pressure
considered. 2. Type of stage configuration (one- or two-section)
Table 8.3 Advantages and Disadvantages of Different Extraction Equipment

Class of Equipment Advantages Disadvantages
Mixer-settlers Good contacting Large holdup
Handles high viscosity and High power costs
interfacial tension High investment
Handles wide flow ratio Large floor space
Low headroom Interstage pumping may be
High stage efficiency required
Many stages available
Reliable scale-up
Continuous, static counterflow Low initial cost Cannot handle high flow ratio
contactors Low operating cost High headroom
(no mechanical drive) Simplest construction Low stage efficiency
High throughput except for Difficult scale-up
small density difference
Continuous counterflow Good dispersion Limited throughput with small
contactors with mechanical Reasonable cost density difference
agitation Many stages possible Cannot handle emulsifying
Relatively easy scale-up systems
Cannot handle high flow ratio
Centrifugal extractors Handles low-density High initial costs
difference between phases High operating cost
Low holdup volume High maintenance cost
Short holdup time
Low space requirements
Small inventory of solvent
Avoids emulsification
Process
Yes
Small volume Separators, centrifugal extractors
No
Emulsion Yes
formation, poor Separators
separation
With Centrifugal extractors,
No reserva- Graesser, RDC, ARD
tions
Small number
of theoretical
stages Yes
required (<5)
Yes Mixer-settler battery,
Low height centrifugal extractors
No No
Yes All column types,
Small floor
area centrifugal extractors
Large number
of theoretical Yes
stages
required (>5)
Yes Mixer-settler battery,
Low height
Graesser
No
Small floor Yes

area
High Pulsating sieve
Yes Small load Yes
throughput tray column,
range
(>50m3/h) Kuhni, Lurgi
Tower Extractor
No
No Yes
Large load
RDC, ARD
range
Low Yes Small load Yes Pulsating

throughput range packed column,
(<50m3/h) Karr column
No
Large load Yes Scheibel,

range Oldshue–
Rushton
Figure 8.8 Scheme for selecting extractors.

[Reproduced from [5] with permission of Institution of Chemical Engineers.]
3. Desired degree of recovery of one or more solutes for 13. Maximum residence time to avoid degradation
one-section cascades 14. Type of extractor
4. Degree of feed separation for two-section cascades 15. Extractor cost and horsepower requirement
5. Choice of solvent(s)
The ideal solvent has:
6. Operating temperature (often ambient)
7. Operating pressure (greater than the bubble point of 1. High selectivity for the solute relative to the carrier
both phases) to minimize the need to recover carrier from the
solvent
8. Minimum-solvent flow rate and actual-solvent flow
rate as a multiple of the minimum rate for one-section 2. High capacity for dissolving the solute to minimize
cascades or reflux rate and minimum reflux ratio for solvent-to-feed ratio
two-section cascades 3. Minimal solubility in the carrier
9. Number of equilibrium stages 4. A volatility sufficiently different from the solute that
10. Emulsification and scum-formation tendency recovery of the solvent can be achieved by distillation,
but not so high that a high extractor pressure is needed,
11. Interfacial tension or so low that a high temperature is needed if the sol-
12. Phase-density difference vent is recovered by distillation
§8.2 General Design Considerations 241
Solvent-free Solvent-free
extract Solvent product A Solvent S
SR SR
Reflux
Solvent T
Extract
Feed
Feed
Feed
A, B
Solvent S
Reflux Figure 8.9 Common liquid–liquid extraction

Raffinate Solvent Raffinate Solvent TR cascade configurations: (a) single-section cascade;
Solvent T Solvent-free (b) two-section cascade; (c) dual solvent with
Product B
SR: Solvent S recovery, TR: Solvent T recovery two-section cascade. SR is solvent S recovery and
(a) (b) (c) TR is solvent T recovery.
5. Stability to maximize the solvent life and minimize the From (2-21), the relative selectivity, β, of the solute with
solvent makeup requirement respect to the carrier is obtained by taking the ratio of (8-1) to
6. Inertness to permit use of common materials of (8-2), giving
construction (KA )D (x )II ∕(x )I (γ )I ∕(γ )II
βAC = = A II A I = A I A II (8-4)
7. Low viscosity to promote phase separation, minimize (KC )D (xC ) ∕(xC ) (γC ) ∕(γC )
pressure drop, and provide a high-solute mass-transfer
For high selectivity, βAC should be high, so at equilibrium there
rate
is a high concentration of A and a low concentration of C in
8. Non-toxic and non-flammable characteristics to facili- the solvent. A first estimate of βAC is made from predictions of
tate its safe use activity coefficients (γA )I , (γA )II , and (γC )II at infinite dilution
9. Availability at a relatively low cost where (γC )I = 1, or by using equilibrium data for the lowest tie
10. Moderate interfacial tension to balance the ease of dis- line on a triangular diagram of the type discussed in Chapter 4.
persion and the promotion of phase separation If A and C form a nearly ideal solution, the value in (8-4) of
(γA )I ≈ 1.
11. Large difference in density relative to the carrier to
For high solvent capacity, (KA )D in (8-4) should be high.
achieve a high capacity in the extractor
From (8-2) it is seen that this is difficult to achieve if A and
12. Compatibility with the solute and carrier to avoid con- C form nearly ideal solutions such that (γA )I = 1.0, unless A
tamination and S have a great affinity for each other, which would result
13. Lack of tendency to form a stable rag or scum layer at in a negative deviation from Raoult’s law to give (γA )II < 1.
the phase interface Unfortunately, such systems are rare.
14. Desirable wetting characteristics with respect to For solvent recovery, (KS )D should be large and (KC )D as
extractor internals small as possible to minimize solvent in the raffinate and car-
rier in the extract. This will be the case if activity coefficients
Solvent selection is a compromise among all these proper- (γS )I and (γC )II at infinite dilution are large.
ties. However, first consideration is usually given to selectivity If a water-rich feed is to be separated, it is common to
and environmental concerns, and second to capacity and cost. select an organic solvent; for an organic-rich feed, an aqueous
From (2-19) in Chapter 2, the distribution (partition) coeffi- solvent is preferred. In either case, the solvent chosen should
cient, KD , for solute A between solvent S and carrier C can be lower the activity coefficient of the solute. Consideration
applied. Combining (2-19) with (2-15) and (2-17) gives of molecule group interactions can help narrow the search
(KA )D = (xA )II ∕(xA )I = (γA )I ∕(γA )II (8-1) before activity coefficients are estimated or equilibrium data
are sought. Organic group interactions for solvent-screening
where II is the extract phase, rich in S, and I is the raffinate purposes, based on 900 binary systems, as given by Robbins
phase, rich in C. Similarly, for the carrier and the solvent, [22], are shown in Table 8.4, where the solute and solvent
respectively, each belong to any of 12 different chemical groups. In this
(KC )D = (xC )II ∕(xC )I = (γC )I ∕(γC )II (8-2) table, a minus (−) sign for a given solute-solvent pair means
the solvent will desirably lower the activity coefficient of the
(KS )D = (xS ) ∕(xS ) = (γS ) ∕(γS )
II I I II
(8-3) solute relative to its value in the feed.
Table 8.4 Group Interactions for Solvent Selection
Solute Solvent Class

Class H-Donor Groups 1 2 3 4 5 6 7 8 9 10 11 12
1 Phenol 0 0 − 0 − − − − − − + +
2 Acid, Thiol 0 0 − 0 − − 0 0 0 0 + +
3 Alchol, Water − − 0 + + 0 − − + + + +
4 Active-H on Multihalo Paraffin 0 0 + 0 − − − − − − 0 +
H-Acceptor Groups
5 Ketone, Amide with no H on N Sulfone, − − + − 0 + + + + + + +
Phosphine, Oxide
6 Tert Amine − − 0 − + 0 + + 0 + 0 0
7 Sec Amine − 0 − − + + 0 0 0 0 0 +
8 Pri Amine, Ammonia, Amide with 2H on N − 0 − − + + 0 0 + + + +
9 Ether, Oxide, Sulfoxide − 0 + − + 0 0 + 0 + 0 +
10 Ester, Aldehyde, Carbonate, Phosphate, − 0 + − + + 0 + + 0 + +
Nitrate, Nitrite, Nitrile, Intra Molecular
Bonding, e.g., O-nitro Phenol
11 Aromatic, Olefin, Halogen Aromatic, + + + 0 + 0 0 + 0 + 0 0
Multihalo Paraffin without Active-H,
Monohalo Paraffin
Non H-Bonding Groups
12 Paraffin, Carbon Disulfide + + + + + 0 + + + + 0 0
(+) Plus sign means that compounds in the column group tend to raise activity coefficients of compounds in the row group.
(−) Minus sign means a lowering of activity coefficients of compounds in the row group.
(0) Zero means little or no effect on activity coefficients of compounds in the row group.
Choose a solvent that lowers the activity coefficient.
[Reproduced from [22] with permission of the American Institute of Chemical Engineers.]
Suppose it is desired to extract acetone from water. In using tables of partition coefficients and Godfrey Miscibility
Table 8.4, Group 3 applies for this solute, and desirable Numbers in Perry’s Chemical Engineers Handbook [2].
solvents are given in Groups 1 and 6. In particular, trichloro-
ethane, a Group 6 compound, is a selective solvent with high
§8.2.1 Representation of Ternary Data
capacity for acetone. If the compound is environmentally
objectionable, it is rejected. Chapter 4 introduced ternary diagrams for representing
A sophisticated solvent-selection method, based on the liquid–liquid equilibrium data at constant temperature.
UNIFAC group-contribution method for estimating activity Such diagrams are available for a large number of systems
coefficients and utilizing a computer-aided constrained opti- (Humphrey et al. [6]). For ternary systems, the most common
mization approach, has been developed by Naser and Fournier diagram is Type I in Figure 8.10a; much less common are
[23]. A preliminary solvent selection can also be made Type II systems, in Figure 8.10b, which include (1) n-heptane/
Solute
A Solute
A
Plait Equilibrium
point curve
se
Extract
ha
phase
p
ate
Raffinate Tie
phase lin
ffin
e
Ra
e
Tie lin
ph tract
e
as
Ex
Carrier Solvent Carrier Solvent

C S C S
(a) (b)
Figure 8.10 Common classes of ternary systems: (a) Type I, one immiscible pair; (b) Type II, two immiscible pairs.
§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 243
aniline/methylcyclohexane, (2) styrene/ethylbenzene/dieth- Except in the near-critical region, pressure has little effect
ylene glycol, and (3) chlorobenzene/water/methyl ethyl ketone. on liquid–liquid equilibrium because it has little effect on
For Type I, the solute and solvent are miscible in all propor- liquid-phase activity coefficients. The operating pressure need
tions; in Type II, they are not. For Type I systems, the larger only be greater than the bubble-point pressure. Most extractors
the two-phase region on line CS, the greater the immiscibility operate at near-ambient temperature, with the process being
of carrier and solvent. The closer the top of the two-phase nearly isothermal if the heat of mixing is small.
region is to apex A, the greater the range of feed composi- Laboratory or pilot-plant work, using actual plant feed and
tion along line AC that can be separated with solvent S. In solvent, is normally necessary to ascertain dispersion and coa-
Figure 8.11, only feed solutions in the composition range lescence properties. Although rapid coalescence of drops is
from C to F can be separated because, regardless of how desirable, this reduces interfacial area and leads to reduced
much solvent is added, two liquid phases are not formed in the mass-transfer rates. Thus, compromises are necessary. Coa-
feed composition range of FA (i.e., FS does not pass through lescence is enhanced when the solvent phase is continuous and
the two-phase region). Figure 8.11a has a wider range of feed mass transfer of solute is from the droplets. This phenomenon,
composition than Figure 8.11b. For Type II systems, a high the Marangoni effect, is due to a lowering of interfacial ten-
degree of insolubility of S in C and C in S will produce a sion by a significant solute presence in the interfacial film. If
desirable high relative selectivity, but at the expense of solvent solvent is the dispersed phase, the interfacial film is depleted of
capacity. Solvents that result in Type I systems are thus more solute, causing an increase in interfacial tension and inhibition
desirable. of coalescence.
Whether a system is of Type I or Type II often depends For a given (1) feed liquid, (2) degree of solute extraction,
on temperature. Figure 8.12 shows the data of Darwent and (3) operating pressure and temperature, and (4) solvent for a
Winkler [24] for the ternary system n-hexane (H)–methyl- single-section cascade, there is a minimum solvent-to-feed
cyclopentane (M)–aniline (A) for temperatures of 25, 34.5, flow-rate ratio that corresponds to an infinite number of
and 45∘ C. At the lowest temperature, 25∘ C, the system is countercurrent equilibrium contacts. As with absorption and
Type II because both H and M are only partially miscible in stripping, a trade-off exists between the number of equilibrium
the aniline solvent. As temperature increases, the solubility of stages and the solvent-to-feed ratio. For a Type I system, this
M in aniline increases more rapidly than the solubility of H ratio is best determined by a graphical method described next.
in aniline, until 34.5∘ C, the critical solution temperature for
M in aniline, where the system is at the border of Type II and
Type I. At 45∘ C, the system is clearly of Type I, with aniline §8.3 HUNTER–NASH GRAPHICAL
more selective for M than H. Type I systems have a plait EQUILIBRIUM-STAGE METHOD
point (see Figure 8.10a); Type II systems do not. For preliminary studies, stagewise extraction calculations for
Type I ternary systems are conveniently carried out with equi-
A
A librium diagrams. In this section, procedures are developed
F using the Hunter-Nash method [26] with equilateral-triangle
and right-triangle diagrams. Free triangular graph paper can
be downloaded from several websites, including http://www.
waterproofpaper.com/graph-paper/ternary-diagram-triangular
-graph-paper.shtml.
F
Figure 8.13 shows a schematic diagram for a countercur-
rent, continuous-flow, N-equilibrium-stage extraction cascade
C S C S operating isothermally at steady state. Each stage includes
(a) (b) mixing and phase separation. Stage numbering begins at the
feed end, opposite to the end at which the solvent enters.
Figure 8.11 Effect of solubility on range of feed composition that
Typically, mass, rather than molar flow rates are used. The
can be extracted.
final raffinate feed rate is RN and the final extract feed rate
M M M
Figure 8.12 Effect of temperature on

H A H A H A solubility for the system n-hexane (H)/
T = 25°C T = 34.5°C T = 45°C methylcyclopentane (M)/aniline (A).
Extract Solvent S
E1 E2 E3 En En + 1 EN –1 EN
1 2 n N–1 N
R1 R2 Rn–1 Rn RN–2 RN –1 Raffinate

F
RN
Feed
Figure 8.13 Countercurrent-flow, N-equilibrium-stage, liquid–liquid extraction cascade.
is E1 . Phase equilibrium is assumed at each stage, so for Some systems, such as isopropanol–water–benzene, exhibit
any intermediate stage n, the components in extract En and a phenomenon called solutropy, wherein moving from the
raffinate Rn are in equilibrium. Mass transfer of all species plait point into the two-phase region of the diagram, the tie
occurs at each stage. The feed, F, contains the carrier, C, lines first slope in one direction, but then the slope diminishes
and the solute, A. It can also contain solvent, S, up to the until an intermediate tie line becomes horizontal. Below that tie
solubility limit. Entering solvent flow rate, S, can contain C line, the remaining tie lines slope in the other direction. Some-
and A up to the solubility limits, but preferably contains little times the solutropy phenomenon disappears if mole-fraction
of either. Because most liquid–liquid equilibrium data are coordinates, rather than mass-fraction coordinates, are used.
listed and plotted in mass rather than mole concentrations,
let: (yi )n = mass fraction of species i in extract leaving stage §8.3.1 Determining the Number of
n and (xi )n = mass fraction of species i in raffinate leaving Equilibrium Stages
stage n.
Although Figure 8.13 implies that the extract is the light A degrees-of-freedom analysis for the ternary system cascade
phase, either phase can be the light phase. Assume the ternary of Figure 8.13 shows there are six variables that must be
system is at temperature, T, such that liquid–liquid equilib- specified for a ternary system. Four of the variables are F,
rium data are as shown in the equilateral-triangle diagram of (xi )F , (yi )S , and T. The remaining two variables are selected
Figure 8.14, whose coordinates were described in §4.4. The by choosing one of the following sets.
bold line is the equilibrium curve (also called the binodal
curve because the plait point separates the curve into an Set 1. S and (xi )RN Set 4. N and (xi )RN
extract to the left and a raffinate to the right). The dashed lines Set 2. S and (yi )E1 Set 5. N and (yi )E1
are tie lines connecting compositions of equilibrium phases Set 3. (xi )RN and (yi )E1 Set 6. N and S
lying on the binodal curve. Because the tie lines slope upward
from the C side toward the S side, at equilibrium, the solute, where (xi )RN and (yi )E1 and all equilibrium phases lie on the
A, has a concentration higher in S than in C. This makes S an equilibrium curve.
effective solvent for extracting A from C. If the tie lines slope Calculations for sets 1 to 3 involve determination of N and
downward, S is not an effective solvent. can be made directly using a graphical procedure on a trian-
(Solute)
A
90 10
Tie line
80 20 Operating line
Equilibrium curve
70 30
60 40
Plait
50 point 50
40 E1 60
30 70
F
20 M 80
Mmax 10 90
RN Figure 8.14 Construction 1: Location of
(Solvent) S C (Carrier)
90 80 70 60 50 40 30 20 10 product points.
gular diagram. Process simulators do not allow these three The combined flow rates and compositions of entering
specification sets because N must be specified. Sets 4 to 6 streams, F and S, must also equal the combined flow rates and
involve a specified N and require a tedious, iterative graphical compositions of the exiting streams, RN and E1 . With point M
procedure. Set 6 is often used with process simulators. First, located, the composition of exiting extract E1 is determined
consider Set 1, with the procedures for Sets 2 and 3 being from overall total and component material balances for the
minor modifications. The technique, called the Hunter–Nash cascade:
method [26], involves three construction steps on the triangu-
lar diagram. They are more difficult than the McCabe–Thiele M = RN + E1 = 350 kg
staircase-step method for distillation. Although the procedure (xA )M M = 60 = (xA )RN RN + (xA )E1 E1
is illustrated here only for a Type I system, parallel principles (xC )M M = 190 = (xC )RN RN + (xC )E1 E1
apply to a Type II system. The constructions are shown in (xS )M M = 100 = (xS )RN RN + (xS )E1 E1
Figures 8.14, 8.15, and 8.17, where A is the solute, C the
carrier, and S the solvent. On the equilibrium curve, extract Specify (xA )RN = 0.025. Because it must lie on the equilib-
compositions lie to the left of the plait point, while all raffinate rium curve, RN can be located and the values of (xC )RN and
compositions lie to the right. Determination of N is as follows, (xS )RN can be read from Figure 8.14. A straight line drawn
given feed, F, and solvent, S, flow rates and compositions, from RN through M locates E1 at the equilibrium-curve inter-
and the final raffinate composition: section, from which the composition of E1 can be read. Values
of the flow rates RN and E1 can then be determined from the
Construction 1 (Product Composition Points) overall material balances above, or from Figure 8.14 by the
inverse-lever-arm rule:
On Figure 8.14, locate mixing point M, which represents the
combined flow rate and composition of feed F plus entering E1 ∕M = MRN ∕E1 RN
solvent S. Assume the following feed and solvent specifica- RN ∕M = ME1 ∕E1 RN
tions, as plotted in Figure 8.14:
with M = 350 kg. By either method, the results are:
Feed Solvent
Raffinate Product Extract Product
F = 250 kg S = 100 kg
(xA )F = 0.24 (xA )S = 0.00 RN = 198.6 kg E1 = 151.4 kg
(xC )F = 0.76 (xC )S = 0.00 (xA ) = 0.025 (xA ) = 0.364
(xS )F = 0.00 (xS )S = 1.00 (xC ) = 0.900 (xC ) = 0.075
(xS ) = 0.075 (xS ) = 0.561
By total and component material balances, the composition
of M = F + S = 350 kg is determined as follows: Again, the composition mass fractions for the two products are
seen to sum to 1.000.
(xC )M M = (xC )F F + (xC )S S Also included in Figure 8.14 is point Mmax , which lies on
= 0.76(250) + 0(100) = 190 kg the equilibrium curve along the straight line connecting F to S.
Mmax corresponds to the maximum possible solvent addition
(xC )M = 190∕350 = 0.543 if two liquid phases are to exist. By the inverse-lever-arm rule,
it is seen that a very large ratio of S to F exists.
(xS )M M = (xS )F F + (xS )S S
= 0(250) + 1(100) = 100 kg Construction 2 (Operating Point and Operating Lines)
(xS )M = 100∕350 = 0.286
For vapor–liquid cascades, §6.3 and §7.2 describe an operating
Note that the component mass fractions for M sum to 1.000. line that is the locus of passing streams in a cascade. Referring
From these two (xi )M values, point M is located, as shown to Figure 8.13, material balances around groups of stages from
in Figure 8.14. Based on properties of the triangular diagram the feed end can be written as differences:
discussed in §4.4, point M must be located on the straight line
F − E1 = · · · = Rn−1 − En = · · · = RN − S = P (8-5)
connecting F and S. Therefore, M can be located knowing just
one value of (xi )M , say, (xS )M . Also, the ratio S∕F is given by Because the passing streams are differenced, P defines a
the inverse-lever-arm rule as difference point, not a mixing point, M. From the same geo-
metric considerations that apply to a mixing point, a difference
S∕F = MF∕MS = 100∕250 = 0.400
point also lies on a line through the points involved. However,
or whereas M lies inside the diagram and between the two end
S∕M = MF∕SF = 100∕350 = 0.286 points, P usually lies outside the triangular diagram along
an extrapolation of the line through a pair of passing-stream
Thus, point M can be located by two composition points or by points forming differences, such as F and E1 , RN and S,
measurement, employing either of these ratios. and so on.
Tie line
90 10
Operating line
80 20 Equilibrium curve
70 30
60 40
Plait
point
50 50
40 E1 60
30 70
F
20 80 Opera
ting lin
e Operating
10 90 point
P Figure 8.15 Construction 2: Location of
RN
S C Operating line
90 80 70 60 50 40 30 20 10 operating point.
To locate the difference point, P, two straight lines C

are drawn through the passing-stream point pairs (E1 , F) and
(S, RN ) established by Construction 2, as shown in Figure 8.15. Plait
These lines are extrapolated until they intersect at difference point
E
(operating) point P. Figure 8.15 shows the intersection at
point P. From (8-5), straight lines drawn through points for D
any other pair of passing streams, such as (En , Rn-1 ), must C
also pass through point P. Thus, the difference point becomes
A H B Plait
an operating point, and lines drawn through pairs of points for point
F G
passing streams and extrapolated to point P are operating lines.
The difference point has properties similar to a mixing E
point. If F − E1 = P is rewritten as F = E1 + P, F can be
interpreted as the mixing point for P and E1 . Therefore, by H
D
the inverse-lever-arm rule, the length of line E1 P relative to
the length of line FP is J
A J B
E1 P E1 + P F (a) (b)
= = (8-6)
FP E1 E1 Figure 8.16 Use of conjugate curves to interpolate tie lines:
(a) Method 1. (b) Method 2.
Thus, point P can be located with a ruler using either pair of
feed-product passing streams.
In Figure 8.15, the operating point, P, lies on the feed or raf-
either of two interpolation methods illustrated in Figure 8.16.
finate side of the diagram. Depending on the relative amounts
In Figure 8.16a, the conjugate line from the plait point to J
of feed and solvent and the slope of the tie lines, point P may
is determined from four tie lines and the plait point. From tie
lie on the solvent or feed side of the diagram, and inside or
line DE, lines DG and EF are drawn parallel to triangle sides
outside of the diagram.
CB and AC, respectively. The intersection at point H gives a
second point on the conjugate curve. Subsequent intersections
Construction 3 (Tie Lines and Equilibrium Lines) establish additional points from which the conjugate curve is
The third type of construction involves the tie lines that define drawn. Then, using the curve, additional tie lines are drawn
the equilibrium curve, which is divided into the two sides (raf- by reversing the procedure. If it is desired to keep the conju-
finate and extract) by the plait point. For Type I diagrams, the gate curve inside the two-liquid-phase region of the triangular
plait point is the composition of two equilibrium phases that diagram, the method illustrated in Figure 8.16b is used, where
become one phase. A material balance around stage n for any lines are drawn parallel to triangle sides AB and AC.
of the three components is
Stepping Off Stages
(xi )n−1 Rn−1 + (yi )n+1 En+1 = (xi )n Rn + (yi )n En (8-7)
In Figure 8.17, equilibrium stages are stepped off by alternate
Because Rn and En are in equilibrium, their composition points use of tie lines and operating lines. Constructions 1 and 2 have
are at the two ends of a tie line. Typically, a diagram will not been employed to locate points F, E, S, R1 , and P. Starting at
contain all tie lines needed; however, more tie lines may be the feed end of the cascade, Construction 3 is used to draw
added by centering them between existing tie lines, or by using a tie line from point E1 to equilibrium phase R1 . Because R1
90 10
Tie line
80 20
Operating line
70 30 Equilibrium curve
60 40
Plait
point
50 50
40 E1 60
30 70
F
20 E2 80
R1 90
10 E3
R2 P
Figure 8.17 Determination of the
S RN C
90 80 70 60 50 40 30 20 10 number of equilibrium stages.
passes E2 , Construction 2 requires that E2 lie at the intersec- then be located on the triangular diagram so as to pass through
tion of a straight operating line drawn through points R1 and P. Details of this procedure are given by Treybal [25].
P, and back to the extract side of the equilibrium curve. R2 is
located with a tie line from E2 by Construction 3. From R2 , E3 §8.3.2 Minimum and Maximum Solvent-to-Feed
is located with an operating line through P by Construction 2.
Flow-Rate Ratios
Continuing in this fashion by alternating between equilibrium
tie lines and operating lines, the specified point RN is reached The procedure of §8.3.1 for determining the number of
or passed. If the latter, a fraction of the last stage is taken. In equilibrium stages to achieve a desired solute extraction for
Figure 8.17, 2.8 equilibrium stages are required, where equi- a given solvent-to-feed ratio presupposes that this ratio is
librium stages are counted by the number of equilibrium tie greater than the minimum ratio, which corresponds to an
lines used. infinite number of stages, and less than the maximum ratio
Procedures for specification Sets 2 and 3 are similar to that that would prevent the formation of the required second liquid
for Set 1. Sets 4 and 5 can be handled by iteration on assumed phase. In practice, the minimum ratio is determined before
values for S and following the procedure for Set 1. Set 6 can solving specification Sets 1 or 2. This is done by solving
also use the procedure of Set 1 by iterating on E1 . Set 4 with N = ∞, where, as in distillation, absorption, and
From (8-6), it is seen that if the ratio F∕E1 approaches a stripping, the infinity of stages occurs at an equilibrium-curve
value of 1, operating point P will be located at a large distance and operating-line pinch point. In ternary systems, the pinch
from the triangular diagram. In that case, by using an arbitrary point occurs when a tie line coincides with an operating line.
rectangular-coordinate system superimposed over the triangu- Thus, the pinch point becomes a pinch line. The procedure is
lar diagram, the coordinates of P can be calculated from (8-6) involved because the location of the pinch line is not known.
using the equations for the two straight lines established in Figure 8.18 shows the previous A–C–S system. The
Construction 2. Operating lines for intermediate stages can composition points F, S, and RN are specified, but E1 is not
90 10 Tie line
Operating line
80 20
Equilibrium curve
70 30
60 Plait 40
Extract point Raffinate
side 50 50 side
40 60
30 70
F
20 80
10 90 L
O RN
S P2 P3 P1 = Pmin
90 80 70 60 50 40 30 20 10 C
Figure 8.18 Determining minimum solvent-to-feed ratio.

because the solvent rate is not known. The operating line OL is as the other extreme of infinite stages, it is necessary to select a
drawn through the points S and RN and extended to the left and solvent ratio, S∕F, such that (S∕F)min < (S∕F) < (S∕F)max . In
right. This line is the locus of all possible material balances Figure 8.14, the mixing point Mmax is located as shown, from
determined by adding S to RN . Each tie line is then assumed which (S∕F)max is determined to be about 16.
to be a pinch line by extending each tie line until it intersects
the line OL. In this manner, a sequence of intersections P1 , P2 , EXAMPLE 8.1 Equilibrium Stages for Extraction.
etc., is found. If these points lie on the raffinate side of the
diagram, as in Figure 8.18, the pinch line corresponds to the Acetone is to be extracted from a 30 wt% acetone (A) and 70 wt%
line extending to Pmin located at the greatest distance from RN . ethyl acetate (C) feed at 30∘ C, using pure water (S) as the solvent,
in the cascade shown at the bottom of Figure 8.19, which estab-
If the triangular diagram does not have a sufficient number
lishes the nomenclature for this example. The final raffinate is to
of tie lines to determine that point accurately, additional tie contain 5 wt% acetone on a water-free basis. Determine the minimum
lines are introduced by the method illustrated in Figure 8.16. and maximum solvent-to-feed ratios and the number of equilibrium
If it is assumed in Figure 8.18 that no other tie line gives a stages required for two intermediate S∕F ratios. The equilibrium data
point Pi that is farther away from RN than P1 , then P1 = Pmin . plotted in Figure 8.19 are from Venkataratnam and Rao [28] and
With Pmin known, the point for the composition of E1 is at the correspond to a Type I system, but with tie lines sloping downward
extract-side end of the tie line coinciding with the pinch line. toward the extract side. Thus, although water is a convenient solvent,
In Figure 8.18, the composition at that point is 51% A, 13.5% it does not have a high capacity, relative to ethyl acetate, for dissolv-
C, and 35.5% S. An operating line can be drawn through point ing acetone. Also determine, for the feed, the maximum wt % acetone
F and extended to E1 at an intersection with the extract side that can enter the extractor. This example and Example 8.2 are taken
of the equilibrium curve. largely from an analysis by Sawistowski and Smith [29].
From the compositions of the four points S, RN , F, and E1 ,
Solution
the mixing point M can be found and material balances used
to solve for Smin ∕F: Point B represents the solvent-free raffinate. By drawing a straight
line from B to S, the intersection with the equilibrium curve on the
F + Smin = RN + E1 = M (8-8) raffinate side, B′ , is the raffinate composition leaving stage N.
(xA )F F + (xA )S Smin = (xA )M M (8-9) Minimum S/F. Because the tie lines slope downward toward the
from which extract side, the extrapolated tie line that intersects the extrapolated
Smin (x ) − (xA )M line SF closest to point S is sought. This tie line, leading to Pmin ,
= AF (8-10)
F (xA )M − (xA ) S is shown in Figure 8.20. The intersection is not shown because it
occurs far to the left of the diagram. Because this tie line is at the feed
A solvent flow rate greater than Smin is required for extrac- end of the extractor, location of extract composition D′min is found as
tion in a finite number of stages. In Figure 8.18, such a solvent shown in Figure 8.20. The mixing point, Mmin , for (S∕F)min is the
rate results in an operating point P to the right of Pmin , which intersection of lines B′ D′min and SF. By the inverse-lever-arm rule,
is a location farther away from RN . A reasonable value for (S∕F)min = FMmin ∕SMmin .
S might be 1.5 Smin . From Figure 8.18, the simultaneous
solution of (8-8) and (8-9) gives (xA )M = 0.185, from which, Maximum S/F. If M in Figure 8.20 is moved along line FS toward S,
the intersection for (S∕F)max occurs at the point shown on the extract
by (8-10), Smin ∕F = 0.30. In Figure 8.17, S∕F = 0.40, giving
side of the binodal curve. By the inverse-lever-arm rule, using line
S∕Smin = 1.33.
FS, (S∕F)max = FM max ∕SM max = 12.
In Figure 8.18, the tie lines slope downward toward the raf-
finate side. If the tie lines slope downward toward the extract Equilibrium stages for other S/F ratios. First consider S∕F = 1.75.
side of the diagram, the above procedure for finding Smin ∕F is In Figure 8.19, the composition of the saturated extract D′ is obtained
modified. The sequence of points P1 , P2 , etc., now occur on from an extractor material balance,
the other side of the triangular diagram. However, the pinch
S + F = D′ + B′ = M
line now corresponds to the point, Pmin , that is closest to point
S, so an operating point, P, must be chosen between points For S∕F = 1.75, point M is located such that FM∕MS = 1.75. A
Pmin and S. For a system that exhibits solutropy, intersections straight line must pass through D′ , B′ , and M. Therefore, D′ is
P1 , P2 , etc., are found on both sides of the diagram. Those on located by extending B′ M to the extract envelope.
the extract side determine the minimum solvent-to-feed ratio. The difference point P is located to the left of the diagram. There-
In Figure 8.14, mixing point M lies in the two-phase region. fore, P = S − B′ = D′ − F. It is located at the intersection of exten-
As this point moves along the line SF toward S, the ratio S∕F sions of lines FD′ and B′ S.
increases according to the inverse-lever-arm rule. In the limit, a Step off stages, starting at D′ , by following a tie line to L1 . Then
maximum S∕F ratio is reached when M = Mmax arrives at the V2 is located by noting the intersection of the operating line L1 P with
equilibrium curve on the extract side. Now all of the feed is the phase envelope. Additional stages are stepped off by alternating
between the tie lines and operating lines. Only the first stage is shown;
dissolved in the solvent, no raffinate is obtained, and only one
four are required.
stage is required. To avoid this impractical condition, as well
Acetone
A
10 90
20 80
Tie line
Operating line 30 70
W
Equilibrium curve
ter
t%
40 60
wa
ac
%
eto
Wt
50 50
ne
Plait point
Raf
60 env finate 40
elo
ct pe F
tra pe
70 Ex elo 30
v
en
80 Tie line L1 20
M
V2 D′ L2
10
B
B′
P S 10 20 30 40 50 60 70 80 90 C
Water Wt% acetate Ethyl acetate
V2 VN
D′ S, solvent
Solvent
D 1 2 N
removal Solvent
F B
L1 LN – 1 B′ removal
(30% A) (5% A)
Figure 8.19 Determination of stages for Example 8.1 with S/F = 1.75.
Acetone
A
10 90
Tie line
Operating line 20 80
Equilibrium curve
30 70
W
G
ter
t%
40 60
wa
ac
%
eto
Wt
50 50
ne
Extract Raffinate
side side
60 G 40
L1
70 D′min Tie line
To Pmin F
80 Mmin 20
90 10
Water S B
C
10 20 30 40 50 60 70 80 90 Acetate
Extrapolated Mmax
line SB D′max
Wt% acetate
B′
Pmax
Figure 8.20 Minimum and maximum S/F for Example 8.1.

For example, point M represents an overall liquid mixture

For S∕F = 5(S∕F)min = 3.0, M is determined and the stages are of (xA = 0.43, xS = 0.28). By calculation, xC , which is not
stepped off in a similar manner to give two equilibrium stages. In included in the diagram, is 1 − 0.43 − 0.28 = 0.29. Although
summary, for the countercurrent cascade:
lines of constant xC are included in Figure 8.21, they are
usually omitted because they clutter the diagram. As with the
S∕F (solvent∕feed ratio) 0.60 1.75 3 12 equilateral-triangle diagram, the binodal curve is shown with
N (equilibrium stages) ∞ 4 2 1 extract and raffinate sides, tie lines connecting equilibrium
xD (wt% acetone, solvent free) 64 62 50 30 phases, and the plait point.
Because point M falls within the phase envelope, the
If the wt% acetone in the feed mixture is increased from the base mixture separates into two liquid phases, given by points A′
value of 30%, a feed composition will be reached that cannot be and A′′ at the ends of the tie line passing through M. The
extracted because two liquid phases will not form (no phase split- extract at A′′ is richer in the solute (A) and the solvent (S)
ting). This feed composition is determined by extending a line from than the raffinate at A′ . Point M might be the result of mixing
S, tangent to the equilibrium curve, until it intersects AC at point G in a feed, point F, consisting of 26,100 kg∕h of 60 wt% A in
Figure 8.20. The feed composition is 64 wt% acetone. Feed mixtures C (xA = 0.6, xS = 0), with 10,000 kg∕h of pure solvent, point
with a higher acetone content cannot be extracted with water. S. The mixture splits into phases A′ and A′′ . The location
of M and the amounts of extract and raffinate are given by
the same mixing rule and inverse-lever-arm rule used for
§8.3.3 Use of Right-Triangle Diagrams equilateral-triangle diagrams. The mixture separates sponta-
neously into 11,600 kg∕h of raffinate (xS = 0.08, xA = 0.32)
Ternary countercurrent extraction calculations can also be and 24,500 kg∕h of extract (xS = 0.375, xA = 0.48).
made on a right-triangle diagram, using ordinary graph paper, Figure 8.22 represents the portion of an n-stage, counter-
as shown by Kinney [27]. No new principles are involved. current-flow cascade, where x and y are weight fractions of
The disadvantage is that mass-percent compositions of only solute A in the raffinate and extract, respectively, and L and V
two of the components are plotted; the third is determined, are total amounts of raffinate and extract. The feed to stage N
when needed, by the difference from 100%. An advantage of is LN +1 = 180 kg of 35 wt% of solute A in a saturated mix-
right-triangle diagrams is that either one of the coordinates can ture with carrier C and solvent S (i.e., xN+1 = 0.35), and the
be expanded, if necessary, to increase construction accuracy. solvent to stage 1 is VW = 100 kg of pure S (yW = 0.0). Thus,
Figure 8.21 shows a right-triangle diagram for the A–C–S the solvent-to-feed ratio is 100∕180 = 0.556. These points are
system as in Figure 8.14. The compositions of S (the sol- shown in Figure 8.23. The mixing point for LN +1 and VW is
vent) and A (the solute) are plotted in mass fractions, xi . M1 , as determined by the inverse-lever-arm rule.
Suppose the final raffinate, LW , leaving stage 1 is to contain
S xW = 0.05. By an overall balance,
1.0
M1 = VW + LN+1 = VN + LW (8-11)
0.9 Phase envelope
Because VW , LN +1 , and M1 lie on a straight line, the mixing
0.8 rule requires that VN , LW , and M1 also lie on a straight line.
Furthermore, because VN leaves stage N at equilibrium and LW
0.7 0
C
i on
ct
1
0.6
Ex
ra
Wt fraction S
tf
tra
Tie line
1 W VN – 1 VN – 2
xC
ct x
yN – 1 yN – 2 VW = 100
0.5
=
VN
0
C
V W , yW = 0
=
yN = ?
xC
0.
1
N–2 (Solvent)
=
0.4 N–1
0.
xC
A″ N
0
2
=
xC
0.
LN + 1 = 180
3
=
0.3 LW, xW = 0.05

xC
0.
M xN + 1 = 0.35 LN LN – 1
4
=
xC
0.
(Feed) xN xN – 1
5
=
0.2
xC
0.
(a)
6
=
xC
0.
Plait
7
=
0.1 point VN2 VW = 100 VN3 VW = 100

0.
xC
Raffinate A′ yN2 yW = 0 yN3 yW = 0

=
F
0.
A LN + 1 = 55 LW LN + 1 = 600 LW
C 00
9
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 xW = 0.05 xN + 1 = 0.35 xW = 0.05
xN + 1 = 0.35
0 1
Wt fraction A (b) (c)
Figure 8.21 Right-triangle diagram for system of Figure 8.14. Figure 8.22 Multistage countercurrent contactors.
P2 In Figure 8.23, P2 is at the top of the diagram, where LW VW

and LN+1 VN2 intersect. For case 3, difference point P3 lies at
the bottom, where LW VW and LN+1 VN3 intersect.
Equilibrium stages for Figure 8.23 are stepped off as before
by alternating the use of equilibrium tie lines and operating
S y ,V
lines that pass through the difference point. Considering case 2,
1.0 W W with a solvent-to-feed ratio of 1.818, and stepping off stages
from stage N, a tie line from the point yN2 gives xN = 0.04. But
0.9
this is less than the specified xW = 0.05, so only a fraction of
0.8 yN2 VN2 a stage is required.
If stages are stepped off for case 3, starting from yN3 , the
Tie lines
0.7
Ex
M2 yN, VN
tie line and operating line coincide, giving a pinch point for a
tr
ac
solvent-to-feed ratio of 0.167. Thus, this ratio is the minimum

t
1
0.6
Wt fraction S
corresponding to ∞ equilibrium stages.

0.5 For case 1, where the solvent-to-feed ratio is between that
of cases 2 and 3, the required number of stages lies between 1
0.4
and ∞. The difference point and the steps for this case are not
0
M1 yN3, VN3
0.3 shown in Figure 8.23, but the difference point is located at a
very large distance from the triangle because lines LW VW and
0.2 LN+1 VN2 are nearly parallel. When the stages are stepped off,
M3
0.1
using operating lines parallel to LW VW between one and two
xN +1, LN +1 stages are required.
Raffinate
LW, xW
C A
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0 1
Wt fraction A
§8.3.4 Extract Reflux
A two-section extraction cascade can be refluxed with extract,
as in Figure 8.24. Stages are numbered from the solvent end of
the process. L is used for raffinate flows and V for extract flows.
P3
Extract reflux, LR , is provided by sending the extract, VN , to a
Figure 8.23 Right-triangle diagram constructions for cascade in solvent-recovery step, which removes most of the solvent and
Figure 8.22. gives a solute-rich solution, LR + D, divided into product D
and extract reflux LR , which is returned to stage N. Raffinate
reflux can also be provided, but its use has been judged to be
of little benefit by Skelland [30].
leaves stage 1 at equilibrium, these streams lie on the extract
A degrees-of-freedom analysis for a two-section cascade
and raffinate sides, respectively, of the equilibrium curve. The
with extract reflux is conveniently carried out by using, as ele-
resulting points are shown in Figure 8.23, where it is seen that
ments, two countercurrent cascades, a feed stage, a splitter, and
the solute in the extract is yN = 0.34.
a divider from Table 5.3. The result is ND = 2N + 3C +13. All
Figures 8.22b and 8.22c include two additional cases of
but four of the specifications usually are:
solvent-to-feed ratio, each with the same compositions for the
solvent and feed, and the same value for xW :
Variable Specification Number of Variables
Pressure at each stage N
Feed Solvent, Solvent-to- Extract Mixing Temperature for each stage N
Case LN +1 , kg VW , kg Feed Ratio Designation Point Feed stream flow rate, composition, C+2
temperature, and pressure
1 180 100 0.556 VN M1
Solvent composition, temperature, and C+1
2 55 100 1.818 VN2 M2
pressure
3 600 100 0.167 VN3 M3
Split of each component in the splitter C
(solvent-removal step)
Temperature and pressure of the two 4
For case 2, a difference point, P2 , is defined in terms of
streams leaving the splitter
passing streams, as Pressure and temperature of the divider 2
P2 = VN2 − LN+1 = VW − LW (8-12) 2N + 3C + 9
Extract stripping, recovery of just one component in the feed is

D Stream Solvent Solvent
specified.
divider removal SD
A two-section cascade with extract reflux is particularly
Extract
reflux VN desirable for Type II ternary systems, as shown in Figure 8.10b.
LR Without a plait point, the two-phase region extends across
N
the entire solute composition. Thus, while solvent-free solute
LN VN–1
concentrations in the extract are limited for a Type I system,
no such limit exists for a Type II system. It is thus possi-
Enriching
N–1 ble, with extract reflux, to achieve as sharp a separation as
section
LN–1 VN–2 desired between solute (A) and carrier (C). When reflux is used,
many stages may be required and triangular diagrams become
LF+1 VF cumbersome. In that situation, it is preferred to carry out
Feed
F
extraction calculations with a process simulator as discussed in
F Chapter 10.
LF VF–1
§8.4 THEORY AND SCALE-UP OF
F–1
EXTRACTOR PERFORMANCE
LF–1 VF–2
Industrial extraction equipment can be selected using the
L3 V2 Stripping scheme of Figure 8.8. Often in the chemical industry, the
section
choice is between a cascade of mixer-settlers and a multi-
2
compartment column extractor with mechanical agitation, the
L2 V1 main considerations being stages required, floor space and
headroom available, and capital and operating costs. Methods
1
for estimating size and power requirements for mixer-settler
systems are described next. Size and performance of other
Raffinate
type extraction equipment is considered only briefly because
B
Solvent it is preferable to rely on the experience of vendors, who
SB have pilot-plant equipment. Detailed theory of liquid–liquid
Figure 8.24 Liquid–liquid extraction with extract reflux. extractors is found in Strigle [31], Thornton [32], and
Rydberg [33].
§8.4.1 Mixer-Settler Units

The four additional specifications can be one of the follow-
ing three sets: Preliminary Sizing
Final mixer-settler unit design is done most accurately by scale-
up from laboratory or pilot-plant equipment. However, prelim-
Set 1 Set 2 Set 3 inary sizing can be done using rules-of-thumb and available
Solvent rate Reflux ratio Solvent rate theory formulated in terms of correlations. Experimental
Solute concentration Solute concentration Reflux ratio data of Flynn and Treybal [34] show that when viscosities
in extract in extract are less than 5 cP and the specific-gravity difference is
(solvent-free) (solvent-free) greater than about 0.10, the average residence time required
Solute concentration Solute concentration Number of in the mixing vessel to achieve at least 90% stage effici-
in raffinate in raffinate stages ency may be as low as 30 s and is usually not more than 5 min-
(solvent-free) (solvent-free) utes, when an agitator-power input of 1,000 ft-lbf ∕min-ft𝟑
Optimal feed-stage Optimal feed-stage Feed-stage
(𝟒 hp∕𝟏,𝟎𝟎𝟎 gal) is used.
location location location
As reported by Ryan, Daley, and Lowrie [35], the capacity
of a settler vessel can be expressed in terms of C gpm of com-
bined extract and raffinate per ft2 of phase-disengaging area.
Sets 1 and 2 are of interest in the design of a new extractor For a horizontal, cylindrical vessel of length L and diameter
because two specifications deal with products of designated DT , the economic ratio of L to DT is approximately 4. Thus,
purities. Set 2 is analogous to the design of a binary distillation, if the phase interface is located at the middle of the vessel, the
where purities of the distillate and bottoms, reflux ratio, and disengaging area is DT L or 4D2T . A typical value of C given
optimal feed-stage location are specified. For a single-section by Happel and Jordan [36] is about 5. Frequently, the settling
cascade, it is not feasible to specify the split of the feed with vessel will be larger than the mixing vessel, as is the case in
respect to two components. Instead, as in absorption and the following example.

Liquid-Liquid Extraction With Ternary Systems: 8.0 Instructional Objectives

Uploaded by

Copyright:

Available Formats

Liquid-Liquid Extraction With Ternary Systems: 8.0 Instructional Objectives

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Liquid-Liquid Extraction With Ternary Systems: 8.0 Instructional Objectives

Uploaded by

Copyright:

Available Formats

Trim Size: 8.5in x 11in Seader c08.tex V2 - 10/16/2015 10:49 A.M.

Liquid–Liquid Extraction with Ternary Systems

After completing this chapter, you should be able to:

232 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Makeup solvent Ethyl acetate-rich

Reﬂux Two liquid

Note: All ﬂow rates are in lb/h Wastewater

Figure 8.1 Liquid–liquid extraction process for recovering acetic acid.

§8.1 Equipment for Solvent Extraction 233

234 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Figure 8.4 Horizontal gravity-settling vessel.

The settling step is by gravity in a settler (decanter). In

Figure 8.3 Some common types of mixing impellers:

§8.1 Equipment for Solvent Extraction 235

Heavy §8.1.4 Sieve-Tray Columns

236 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Light liquid in Tie rod

§8.1 Equipment for Solvent Extraction 237

Stator ring Shell

(e) (f) (g)

238 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Heavy phase Figure 8.7 (Continued) ( j) Karr

§8.2 General Design Considerations 239

Table 8.3 Advantages and Disadvantages of Different Extraction Equipment

240 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Small ﬂoor Yes

Low Yes Small load Yes Pulsating

Large load Yes Scheibel,

Figure 8.8 Scheme for selecting extractors.

§8.2 General Design Considerations 241

Reﬂux Figure 8.9 Common liquid–liquid extraction

242 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Table 8.4 Group Interactions for Solvent Selection

Solute Solvent Class

Carrier Solvent Carrier Solvent

Figure 8.12 Effect of temperature on

244 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

R1 R2 Rn–1 Rn RN–2 RN –1 Rafﬁnate

Figure 8.13 Countercurrent-flow, N-equilibrium-stage, liquid–liquid extraction cascade.

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 245

246 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

To locate the difference point, P, two straight lines C

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 247

Figure 8.18 Determining minimum solvent-to-feed ratio.

248 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 249

Figure 8.20 Minimum and maximum S/F for Example 8.1.

250 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

For example, point M represents an overall liquid mixture

0.3 LW, xW = 0.05

0.1 point VN2 VW = 100 VN3 VW = 100

Rafﬁnate A′ yN2 yW = 0 yN3 yW = 0

§8.3 Hunter–Nash Graphical Equilibrium-Stage Method 251

P2 In Figure 8.23, P2 is at the top of the diagram, where LW VW

solvent-to-feed ratio of 0.167. Thus, this ratio is the minimum

corresponding to ∞ equilibrium stages.

252 Chapter 8 Liquid–Liquid Extraction with Ternary Systems

Extract stripping, recovery of just one component in the feed is

§8.4.1 Mixer-Settler Units