Previous Page

400 Chapter 7. SOLVENT EXTRACTION

IN THE OIL INDUSTRY

Solvent I c3 I 4
‘ I 5‘ I c6 I
Solvent ratio (% vol) 60011 000 500/800 300/600 300/600
Temperature (“C) 55/90 901130 1651210 1801230
Pressure (MPa) 314 314 214 214
Number of theoretical stages 2 3 2 to 3 1 to 2 1 to 2
Typical yield (% wt) 40 65 80 85

Table

/716/ Operating conditions depending on the solvent used.

Variable Deasphalted oil Tar

Solvent ratio t Yield t Yield 1
Viscosity t Ring and ball ?’
Solvent ratio 1 Yield 1 Yield t
Viscosity L Ring and ball 1
Top temperature Yield 1 Yield ?
at constant gradient t Viscosity L Ring and ball 1
~~ ~~

Top temperature Yield t Yield 1

at constant gradient 1 Viscosity f Ring and ball t

Overall effectof temperature and solvent ratio.

LYL
7.2.5 Description of a Deasphalting Unit
Below we have provided a detailed description of an industrial deasphalting
unit designed to produce bright stock and road bitumens by fluxing the tar.
The choice of a light solvent (propane or a mixture of propane and butane) jus-
tifies conventional solvent recovery.

7.2.5.1 Description of Deasphalted Oil Circuits (Fig. 7.25 and 7.26)

The feed for the unit is pumped to battery limits by PIol(A or B). The residue
storage tank is located outside battery limits. The rotary pumps give consid-
erable differential pressure and are protected against pressure surges on the
discharge side by PIC 105 which establishes recirculation to suction. FRC 101
 Extractor

To C102A (HP flash)

' LP steam
I- MP steam
Feed storage tank
n 7Condensate

LRC

FRC
105
To Flol (asphalt furnace)

Hot El12
El 02 oil

From,,B
, via PlO3
(HP accumulating drum)
From B,, via P,,
(LP accumulating drum)

1 Figure
7.25 Deasphalting process flow scheme. Extraction section.
 Flash towers Deasphalted oil stripper

From E,,, I 4
’1 LP superheated
steam from F,,,
u

Deasphalted
oil

To El,,
9
(stripper
overhead
condenser)
From E,,,
(asphalt flash
To overhead
c101 condenser)
To
ClOl

p102 y
L
Figure
7.26 Deasphalting process flow scheme. Deasphalted oil-solvent separation section.
 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY 41 1

regulates the feed flow rate for the unit (when FRCV 101 closes, the pressure
rises on the discharge side of PlOland PIC 105 recycles the feed).
Predilution solvent is injected downstream of FRC 101 by means of FRC 102.
The solvent is injected in the feed before it goes through cooler Elol which
adjusts the processing temperature by means of TRC 101 (TRCV 101 shunts a
variable amount of feed into the bypass). Feed is prediluted in order to reduce
its viscosity and saturate it with solvent.
The feed enters the vertical baffle-type extractor (37 trays) above trays 9
to 18.
The process solvent is introduced either above tray 28 or in the lowest
compartment below tray 37. The solvent ratio is regulated by FRC 104 which
gets the solvent from Blol by means of PlO2and from BlO2by means of P103.
The solvent inlet temperature is controlled by TRCV 103 which shunts a
greater or lesser amount of solvent through exchanger ElO2.The exchanger has
a special feature: it cools or heats depending on the quality of products
required and the quality of solvent used.
Most of the solvent ends up at the top of the column with the deasphalted
oil. The temperature gradient is automatically controlled by TRC 102 acting on
the amount of steam by coils located in the upper part of extraction tower Clol.
The instrumentation related to Clol includes interface level recording and dif-
ferential pressure indication throughout the column. The pressure is con-
trolled by PRC 101 at the top of Clol.
Under the control of PRCV 101, the mixture of deasphalted oil and solvent
is sent to the recovery section. It goes through the low pressure steam evapo-
rator El03 where part of the solvent is vaporized, before entering the flash
tower Clo2,.
The vapors coming from the top of C102A are condensed in exchanger ElO4
by contact with the liquid coming from the bottom of Clo2,. The condensed
solvent flows into the high pressure accumulating drum BlO2before being
pumped up by Pi03 for reinjection in tower Clol.
The level of BlO2is controlled by LRC 104 while the pressure is maintained
by PRC 102 acting on the exit of uncondensed vapors in this stage.
From C102A, the liquid mixture of deasphalted oil and solvent already minus
part of the solvent goes to C102Bfor a second low pressure flash (by LIC 101).
The solvent continues being stripped by reboiling in Elos with medium
pressure steam. The overhead vapors are condensed in ElO7with vapors com-
ing from BlO2.
The liquid leaving reboiler El,, is sent to stripper C103 under the control of
LC 102. In C,,, the last traces of solvent are stripped by superheated low pres-
sure steam. The final pressure is maintained and controlled by PRC 103.
The stripped deasphalted oil is pumped up by P104(AorBl then cooled in EIo6
before being sent to battery limits under level control by LIC 103.
The stripping steam comes from the superheater of furnace Flol or other-
wise from the high pressure, 31 bar, expanded steam (this method is used dur-
ing start up or in the event of major disadjustment).
412 Chapter 7 SOLVENT EXTRACTION
IN THE OIL INDUSTRY

The oil must be stripped in any case before it leaves the unit. The heavy
condensate recovered in the boot of BlO2is purged through the drainage e v a p
orator drum.

7.2.5.2 Description of Asphalt Circuits (Fig. 7.27)

The asphalt-solvent mixture leaves the bottom of Clol under the control of
FRC 105, and is fed into preheater El12 (hot oil heater) and furnace F,,,. A low
flow rate alarm establishes emergency solvent circulation to furnace F,,, (from
Plod. The exit temperature of the furnace, controlled by TRC 104, is of great
importance for process stability.
At the furnace exit the asphalt-solvent mixture goes into the flash column
C104 where a large proportion of the solvent is separated and goes to El,, to be
condensed with the vapors delivered by compressor Klol.
The liquid from the bottom of C104 goes into stripper Clos, (by LlC 105),
where stripping is performed by low pressure superheated steam. The over-
head vapors go into the low pressure recovery system while the deoiled
asphalt minus the solvent is pumped up by P105(A or B), discharging through
steam generator Ello (under the control of LIC 107).
Steam is generated in Ello from treated water supplied at battery limits (by
LC 108).
Fluxed asphalt can be produced by injecting a fluxing agent with PlOsand
cooling in exchanger E,,,.

7.2.5.3 Description of Solvent Recovery and Storage Circuits

We saw earlier that the high pressure solvent vapors were condensed in E104,
that the condensate was recovered in accumulating drum Blo2 and that the
vapors coming from C102Bwere condensed in El,, before going to Blol.
In El,, we have condensation of the vapors supplied by the top of asphalt
flash C104 and the discharge side of compressor K,,,. The vapors from s t r i p
pers C103 and ClO5 go to El,, where the stripping steam is totally condensed,
whereas the residual solvent vapors are sucked into drum Blol through con-
denser E,,,.
It is drum Blol that supplies a little more than half of the solvent inventory
in circulation. Its pressure is controlled by PRC 104 which bypasses ElO7
vapors.
Circulating pumps PlO2are driven by high pressure steam turbines which
discharge into the medium pressure collector. They work under pressure con-
trol. Solvent make up is possible from drum B,,,.

7.2.5.4 Description of Compressor Circuits

The gas which is the feed for compressor Klol comes from B103 where the s t r i p
ping steam condensate is separated and drained into the oily effluent sewer
under the control of LIC 109.
 To water
treatment

pi06
Fluxing agent
From E, u
Fluxed asphalt

Asphalt to
Fuel oil LP steam bitumen section

steam

Treated water

1 Figure
7.27 Deasphalting process flow scheme. Asphalt-solvent separation section.
414 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY

The vapors then go to the surge separator B104 where the pressure is con-
trolled by PRC 103 which recycles compressor discharge vapors. In the event
of overpressure due to malfunctioning, rapid decompression at the low pres-
sure flare is possible.

7.2.5.5 Description of Utilities

A positive displacement pump is installed to purge the unit during shut down
and rinse critical points that may be affected by the asphalt, should there be
any accidental emulsion in the solvent circuits, in particular:
exchangers El,,, Elos, and Elo9;
top and bottom exits of C,,,;
bottom exits of C104 and C105.
Rinsing can be carried out at a pressure of approximately 20 bar.
An ejector is installed to deaerate or degas the unit via the low pressure
system.
A blowdown drum receives the discharge from high pressure set valves as
well as their bypass. It is heated to vaporize the liquid solvent that would have
been carried over to the flare. The oil condensate can be pumped by PIo5.
At the entrance of the steam superheater of F,,, a separator with automatic
purge and a high pressure steam expansion valve ensure stripping in towers
C103 and Clos under all circumstances.
There is also a flushing circuit for the pump gear of pumps P,,, and P103.
The circuit is supplied by tapping the discharge of PlOs(fluxing agent pump)
and kept at constant pressure.

7.2.6 Solvent Recovery Techniques

These techniques have evolved a great deal under the dual impetus of the
search for energy savings and the development of deasphalting units as extra
conversion units (utilization of heavy C5 solvents).
The two basic types of recovery can be described as either “conventional”
or under “supercritical conditions”.

7.2.6.1 Recovery by the Conventional Method

This method consists in vaporizing the solvent (several flash stages with low-
ered pressure and increased temperature), usually followed by steam stripping
to eliminate remaining oil or asphalt fractions. The oldest units use one single
evaporation stage while the most recent can have up to three. The configura-
tion of this type of recovery method is presented in Figures 7.28a and b. The
use of multi- evaporation stages allows a dramatic reduction in energy costs,
achieved by recovering the energy used to vaporize the solvent in the high
pressure flash and reusing it in the following flash stages at lower pressure.
Chapter 7. SOLVENT EHRACTIONIN THE OIL INDUSTRY 415
416 Chapter 7. SOLVENT EXTRACTION
IN THE OIL INDUSTRY
 Chapter 7. SOLVENT EXTRACTION IN THE OIL /NDUSTRY 417

For example, units with two or three vaporization stages consume approx-
imately 55 and 45%respectively of the energy required by a single vaporization
stage. In any case, sudden drops in pressure must be avoided, as they may
cause foaming (this phenomenon can be limited by constantly injecting anti-
foaming agents).

7.2.6.2 Recovery under Supercritical Conditions

This type of recovery consists in separating the solvent either at a pressure
greater than its critical pressure or at a pressure greater than the critical pres-
sure of the mixture of deasphalted oil and solvent, which reduces the energy
costs of vaporization considerably.
The Kerr-McGee Residual Oil Supercritical Extraction process (Rose)
(Poddar et al., 1994; Nelson et al., 1985), the UOP Demex process (Hann et al.,
1984) and the IFP Solvahl process (PCriGs et al. 1988, 1995; Billon et al., 1994;
Morrison et al., 1994), are presented in Figures 7.29a, b and c.
The solvent extraction and recovery zones are indicated in Figure 7.29d.
The heavier the solvent, the more economically attractive recovery under
supercritical conditions is. This is why there are still a number of units with
propane recovery by conventional methods.

7.2.7 Economics
An economic assessment of investment and operating costs for a deasphalting
unit requires accurate data on its feed capacity and quality, on the
solvent/feed ratio and on the technology adopted for recovering the solvent.
However, the data supplied by the different licensors can be used to make
rough economic estimates.
For example, Asvahl provides a comparison of investment and utilities
costs for a 1 Mt/year (i.e. 20 000 BPSD) pentane deasphalting unit with Arabian
Light vacuum residue as feed for the following four cases (Table 7.18):
column extractor with conventional solvent recovery;
Solvahl extractor with conventional solvent recovery;
Solvahl extractor with “opticritical” solvent recovery (supercritical
recovery outside of the retrograde condensation zone as defined in
Figure 7.29d);
mixer-settler type extractor with supercritical solvent recovery.
For this case, the comparison shows investments 30% lower for Solvahl
than for conventional deasphalting with an extraction tower and approxi-
mately 18%lower than for mixersettler processes.
The economic advantage of supercritical recovery compared to conven-
tional one-stage recovery has been quantified by Nelson (1985) for a propane
deasphalting unit gable 7.19) with a solvent/feed ratio of around 11 to 1.
418 Chapter 7. SOLVENT EXTRACTION
IN THE OIL INDUSTRY
U
-
a,
c
m
Stripper

+(7+ L

Stripper

5
I-+
420 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY
 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY 421

Bubble and dew point

envelope for deasphalted
oil-solvent mixture
/
I--,
Retrograde
/condensation -- -----\
zone“, ..- - -
--*

Mixer-
\
operation Supercritical
zone recovery
Solvent zone

/‘
LLE
/
critical

T
-
Figure
7.29d Soluent recovery. L: Liquid. V: Vapor. E: Equilibrium.

Process cO1-
extractor 1
I

solvahl solvahl
Mixer-
settler

Solvent
recovery
Conventional

I
Conventional
T
“Opticritical” Supercritical

Investments *
(M$ US)
45.4 30.5
30.5 I 42.0
Utilities:
Electricity (kWh/m3) 11.5 11.5
I
Fuel oil (kg/m?** 25.3 25.3
MP steam (kg/m3) 4.1 4.1

* Erected cost, including engineering.

** Heating value 10 000 kcal/kg (41 860 kJ/kg) with furnace efficiency of 85%.

Economic data for the Soluahl process (1994) updated using the Nelson Cost
Index.
422 Chapter 7. SOLVENT EXTRACTION IN THE O I L INDUSTRY

Utilities costs Conventional recovery

6/m3 1 evaporation stage Supercritical recovery

Fuel oil 3.43 5.02

Electricity 0.66 0.63
Steam 5.92 0.28
Total 10.01 5.93

- I
Table
7.19 Comparison of utilities costs (1985) For a propane deasphalting unit with super-
critical recovery and conventional one-stage evaporation recovev.

The energy costs indicated by Nelson are summarized in Table 7.20 accord-
ing to the solvent recovery technique.

Recovery technique Relative cost

Conventional evaporation:
One stage 280
Two stages 170
Three stages 150
Supercritical recovery 100
(Rose process)

w Relative energy costs (Nelson, 1985).

As indicated by Foster Wheeler (1983) (Table 7.2 l), these differences can
warrant adding one or two evaporation stages: pay out times vary from 0.9 to
1.2 years.
In addition to the solvent recovery technique, investment and utilities
costs are highly dependent on the unit’s production objective: oil for bright
stock or for cracking feeds, as shown in a summary of data collected from
Hydrocarbon Processing (1992) (Table 7.22).
 ChaRter 7. EXTRACTIONIN THE OIL~NDUSTRY 423
SOLVENT

I Number of evaporation stages 1 2 3

I Extra investment required *, ** (k$) 0 696 I 1040

Yearly utilities costs (k$)

MP steam ($5.65/1 000 lbs) *** 141.1 141.1 141.1
LP steam ($4.80/1 000 Ibs) *** 1442.5 720.3 584.5
Electricity ($0.048/kWh) *** 218.4 218.4 218.4
Cooling water ($0.60/103 gal) *** 76.8 42.0 34.6
Fuel oil ($4.3/106 BTU) *** 93.9 93.9 93.9

Total yearly cost (k$) 1972.7 1215.7 1072.5

Reduction in utilities costs (k$) 0 757 900.2
Pay out time (years) - 0.9 1.2

* 1 700 BPSD (s85 000 t/year) of vacuum residue, with a deasphalted oil yield of 93% and
operation 350 days/year.
** US Gulf Coast, January 1983.
*** $5.65/1 000 Ibs = $0.0124/kg. $4.80/1 000 Ibs = $0.0106/kg. $0.048/kWh = $0.0133/MJ.
$0.60/103 gal = $0.158/m3. $4.3/106 BTU = $0.004/MJ.
-
Table
7.2 1 Evaluation of adding extra evaporation stages.

Solvent recovery Conventional evaporation Supercritical

technique 2 stages recovery

Type of oil production Base stock Cracking feed

Investment US$93/bbl 7 900 1550 1085

Utilities costs
Fuel oil lo3 BTU/bbl (MJ/m3> 86 (568) 89 (587) 80-110 (528-726
Electricity kWh/bbl (kWh/m3> 2.1 (13.2) 2.0 (12.6) 2.0 (12.6)
Steam (150 psig = 10 bar rel.)
Ibs/bbl (kg/m3> 115 (328) 60 (171) 12 (34)
Cooling water
gal/bbl (m3/m3> - -

Table
424 Chapter 7. SOLVENT EXTRACTION
IN THE OIL /NOUSTRY

7.3 Aromatics Extraction from Light Oil Cuts

7.3.1 Purpose
Aromatics are extracted from an oil cut either to give added value to the cut
itself or to the aromatics that it contains.
In the first instance, dearomatization aims to produce a better quality
"raffinate", for example a kerosine with a better smoke point, a diesel oil with
a higher cetane number or a lube oil with a better viscosity index. The sepa-
rated aromatics do not have any particular application in the petrochemical
industry and their purity is not one of the process objectives.
Dearomatization processes date back a good number of years. Their pre-
cursor was the Edeleanu process (1907) based on the use of liquid sulfur
dioxide as a solvent. Other solvents, especially phenol and furfural are easier
to use and so have gradually displaced sulfur dioxide. There is a detailed
description of a furfural dearomatization process in Section 7.1.
When the aromatics in the oil cut are the target of the extraction process,
the main objective is to produce them pure. Basically benzene, toluene, xylene
isomers and ethylbenzene are involved. This group of aromatics will be
referred to from now on by the initials BTX. BTX extraction started developing
in the fifties. At that time the growing demand for benzene and xylenes, feed
stocks for polyamide and polyester synthesis, products that had previously
been produced from coal distillation gasoline, prompted an increasing return
to aromatics of petroleum origin. Distillation is incapable of purifying these
aromatics that are found mixed with paraffins and naphthenes, because the
boiling point temperatures of the components are too close and because
azeotropes are present. Solvent extraction is the best suited and most eco-
nomical separation technique.
New processes have appeared with more efficient solvents than simple
dearomatization solvents, such as diethylene- then tetraethyleneglycol,
sulfolane, N-methylpyrrolidone, dimethylsulfoxide and N-formylmorpholine.
These processes are still operational and the objective of this section is to
describe their main characteristics after having reviewed the different feeds
that can be used and the specifications that can be expected of their products.

7.3.2 Sources of BTX Aromatics

Over 90% of the BTX products in the world come from conversion units and
mainly from naphtha steam cracking and catalytic reforming. Coal distillation
gasoline no longer accounts for more than a marginal source.

7.3.2.1 Catalytic Reforming Effluents

The composition of a C,C, catalytic reforming effluent cut is generally located
within the limits indicated in Table 7.23. The total aromatics account for
80 to 90% wt of the cut, with xylenes alone (ortho-, meta- and para-xylene)
 Chapter 7. SOLVENT EXTRACTION IN THE OIL /NOUSTRY 425

I Component I Content in % wt

Benzene 1 to 8
Toluene 8 to 24
Aromatics
Xylenes 30 to 50
Cg+(including mesitylene) 0.1 to 6

c5 0.6 to 6
c6 1.5 to 11
Paraffins c7 1 to 7
c8 0.5 to 2
C9 Oto 1

Naphthenes + olefins 0.5 to 2

-
Table
7.23 Composition of a C6C8 cut from catalytic reforming.

representing over half of this figure. The major impurities in aromatics are
paraffins, with naphthenes and olefins a small minority.
If we examine in detail how impurities are distributed according to the
number of carbon atoms, we can see that their concentration decreases con-
siderably going from c6 to C,. Some modern reforming units produce effluents
from which a C8 cut can be isolated by simple distillation. The cut is made up
of xylenes which, after a clay finishing treatment to remove olefins, are of the
desired purity. In this case, if benzene and toluene are due to be purified, the
solvent extraction involves only the c&7 cut.

7.3.2.2 Steam Cracking Effluents

After a preliminary step of olefin hydrogenation and hydrodesulfurization, the
composition of a steam cracking effluent C&g cut is situated within the limits
presented in Table 7.24. Aromatics are a majority, but contrary to catalytic
cracking effluents, their concentration decreases from benzene to xylenes.
The impurities associated with these aromatics are paraffins and naphthenes
in roughly equal amounts.
Let us point out that when only benzene is the key end product, extractive
distillation of the C, cut is more advantageous than extraction.

7.3.3 Product Specifications

Each of the benzene, toluene and xylenes aromatics is marketed at different
degrees of purity defined by standards. The most commonly used on an
international scale are the ASTM (American Society for Testing and Materials)
standards.
426 ChaDter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY

Component Content in % wt

Benzene 25 to 50
Toluene 14 to 23
Aromatics
Xylenes 10 to 17
C,, (including mesitylene) 0.1 to 6

c5 0.2 to 2
6' 6 to 27
Paraffins 1 to 6
and naphthenes c7
c8 0.5 to 4
C9 Otol

1 I
7.24 Composition of a C,,-C8cut h m steam cracking (afier hydrogenation).

Table 7.25 shows the major ASTM specifications for three common quali-
ties of refined benzene, of "nitration grade" toluene and of xylene mixtures
designed for paraxylene extraction. Some specifications that are common to
all aromatics (appearance, color, acidity, corrosivity, H,S and CO, content) are
not indicated since they do not cause any particular problems in the usual
extraction processes.
The principal criteria which show the overall purity of an aromatic com-
pound are the distillation range, the specific gravity and, for benzene, the crys-
tallization temperature. The type and ratio of impurities are determined by a
number of different titrations and supplementary tests: paraffinic impurities
are determined by gas chromatography, unsaturated non-aromatic hydrocar-
bons are characterized generically by the bromine number and the acid wash
color. The latter test is particularly sensitive to cyclic diolefins: an AWC equal
to 2 for example corresponds to a content of some 20 ppm of cyclopentadiene,
100 ppm of methylpentadiene or 1 000 ppm of cyclohexene. Trace impurities
such as sulfur, nitrogen, thiophene, carbon dioxide and hydrogen sulfide are
subjected to specific titrations.
The quality required for an aromatic is dictated by its use. In some cases a
given application may have higher purity requirements than those of stan-
dardized products. Nylon synthesis for example requires benzene with a very
low sulfur content (0.2 ppm) to keep from deactivating the hydrogenation cat-
alyst too rapidly. It also needs a toluene and methylcyclohexane content lim-
ited to 200 ppm so that the nylon quality is not affected.

7.3.4 Solvent Properties

Solvents used in industrial processes are either glycols: di-, tri- and tetra-
ethyleneglycols (abbreviated to DEG, TEG and TETRA respectively); amides:
 Chapter 7. SOLVENT E ~ ~ R ~ C TINI O N OIL INDUSTRY
THE 427

Benzene485 * Benzene535 Benzene545 Toluene

ASTM D 835 ASTM D 2359 ASTM D 4734 LSTM D 841 'ylenes2

Benzene min.
(% wt)
- - 1 99.9

Acid wash color 2 1 1

(max.1
Bromine number - - 20
(ma.)
Sulfur max. - - 1
(mg/b)
Thiophene max. - 1 0.6
(mg/kg)
Density 0.8780 to 0.8780 to 0.8780 to
(g/cm3) 0.882 0 0.886 0 0.8820
Distill. range 13 13 13
("C)
Cryst. temp. 4.85 5.35 5.45
min.("C)(anhy-
drous benzene)
Non-arom. max. -
(% wt)
Toluene max. -
("A wt)
Paraxylene min. -
(% Wt)
Ethylbenzene -
max. (% wt)
Arom. C9, m a . -
(% wt)
Nitrogen m a . -
(nwkg)
1. Nitration grade. 4. Including 110.6"C at 101.3 Wa.

--
Table
7.25 ASTM aromatics specifications.
U

N-methylpyrrolidone (NMP), N-formylmorpholine (NFM); or oxygenated

derivatives of sulfur-bearing molecules such as dimethylsulfoxide (DMSO) or
tetramethylenesulfone (sulfolane). They of course have the general properties
of industrial solvents: thermal and chemical stability, low toxicity and corro-
sivity, easy availability and moderate cost (between 5 and 35 FF/kg). They have
become popular among the large number of other solvents on the market
because the totality of their physical and solvent properties provide the best
compromise solution for aromatics extraction under well defined and opti-
mized conditions of implementation for each of them.
428 Chapter 7 SOLVENT EXTRACTION IN THE OIL /NOUSTRY

7.3.4.1 Physical Properties

According to the list of physical properties in Table 7.26, it can be seen in par-
ticular that all the solvents have:
A sufficiently low crystallization temperature so that they can be imple-
mented without elaborate equipment. At the most, solvents such as
sulfolane, DMSO and NFM, which have the highest crystallization tem-
peratures, require steam tracing on some storage tanks and lines.
A boiling temperature much higher than that of xylene (= 140°C) which
is generally the least volatile of the aromatics to be extracted. The sol-
vent can therefore be regenerated from the extract by easy, and conse-
quently economical, distillation.
A specific gravity close to or greater than 1.1, thereby ensuring a gravity
differential with the hydrocarbons in the feed (density at 20°C between
0.660 and 0.880 g/cm3) which promotes proper phase settling and extrac-
tor operation.
A viscosity that may be high at ordinary temperature, especially for
glycols, but which is always lower than 2.5 mPa.s at the operating tem-
perature in the extractor (see the second viscosity values indicated in
Table 7.26) and consequently favorable to rapid mass transfer kinetics.

7.3.4.2 Solvent Properties: Capacity and Selectivity

The solvents have a molecular structure made up of a radical or a relatively
short hydrocarbon ring, and a polar group (see the second column in
Table 7.26). This structure gives them two properties: that of being miscible
with each other and with water, and that of selectivity toward hydrocarbons.
With a given hydrocarbon, they exhibit a solubility which depends on the
chemical nature of the hydrocarbon and, to a lesser degree, its molecular size.
More specifically, if we consider the different chemical families of hydro-
carbons, there is always decreasing solubility in the order indicated below for
components with the same number of carbon atoms: aromatics > diolefins >
olefins > naphthenes > paraffins. In the same chemical family, solubility
decreases moderately when the molecular weight of the hydrocarbon
increases.
Different solvents are generally placed on a selectivitycapacity diagram in
order to compare their solvent properties. This type of diagram is represented
in Figure 7.30 for solvent-benzene-hexane ternary systems at 25°C. On the
abscissas, solvent capacity (or solvent power) is expressed by the distribution
coefficient of benzene at the origin in volume fractions. On the ordinates,
solvent selectivity is expressed by the ratio of the distribution coefficients of
benzene and hexane. The diagram shows a fairly large dispersion of points,
with however a clear-cut tendency to follow a law of inverse variation between
selectivity and capacity. Two solvents are in a prime position: dimethylsulfox-
ide and sulfolane. A mixture of two solvents, N-methylpyrrolidone and
ethyleneglycol for example, has properties somewhere in between those of the
 Chapter 7. SOLVENT EXTRACTION IN THE OIL lNDUSTRY 429

Tb P
Pure solvent Developed formula :760 mmHg: W m 9
("C) :at 20°C)

Diethyleneglycol 245 1116 35.7120

(Dm 0.971140

Triethyleneglycol H(OCH,CH&OH 150.2 -7 288 1123 49/20

CTEG) 0.701140
-
Tetraethylene- 194.2 -4 29 1 1 125 61.9/20
glycol (TETRA) (300 mmHg; 1.8/150

Sulfolane H2C- CHZ 120.2 27.6 287 1266 10.3130

:at 30°C) 2.51100
I 1
H2c\S /CH2
o// +o
N-met hylpyrro- H2C- CH2 99.1 -24.4 202 1027 1.65125
lidone (NMP) 1.3/30
I I
H2C\ C,"
N
I
CH3
-
N-formylmorpho- 115.1 20 244 1 150 8.13120
line (NFM) /O\ 0.651150
HzC CH2
I I
H2C\N /CH2

H-C=O
I
Dimethylsulfo- CH3- S--CH, 78.1 18.5 189 1080 1.99/25
xide (DMSO)
0
II 1.6135

Table

two components. Additionally, the arrows on the diagram indicate that solvent
properties can be modified:
By adding water: water would be represented on the diagram by a point
with coordinates (lo-", 135). Adding water to a solvent therefore
decreases its capacity and increases its selectivity.
By varying the temperature: raising the temperature improves solvent
capacity at the expense of its selectivity.
430 ChaDter 7. SOLVENT EXTRACTION IN THE OIL lNDUSTRY

100

50

20

10

1
0.1 0.2 0.5 1 2 5 10 Capacity
Figure
7.30 Selectivity and capacity For industrial aromatic hydrocarbon solvents at 25°C.
EG :ethylene glycol DEG: diethylene glycol
DMSO: dirnethylsulfoxide NMP: N-methylpyrrolidone.

In an extraction process, the temperature and water content of the solvent

are two determining parameters for the extraction step as such, but also have
an impact on downstream units. Accordingly, they are optimized for each pro-
cess taking into account all the units that compose it. Average conditions of
use for the principal solvents in extractors are presented in Table 7.27.

7.3.5 General Process Flow Scheme

An industrial BTX aromatics extraction process always operates on a continu-
ous basis. It includes a series of operations which have been represented
schematically in Figure 7.31. Three main operations can be differentiated:
extraction as such,
purification of aromatics,
solvent regeneration or BTX/solvent separation,
and three ancillary operations involving smaller streams and equipment:
raffinate washing,
 Chapter 7. SOLVENT EXTRACTION
IN THE OIL /NOUSTRY 431

Anti-solvent Operating
Process name License Base solvent temperature (“C)
(% wt)

Udex UOP-DOW DEG Water - 8 140

Sulfolane Shell-UOP Sulfolane Water - 1 100
Arosolvan Lurgi NMP Water - 12 30
Arosolvan Lurgi NMP Glycol - 50 60
Tetra Union Carbide TETRA Water - 4 150
Morphylex KruppKoppers NFM Water - 4 180
DMSO IFP DMSO Water - 5 35

w Conditions of use for extraction solvents.

olvent-water Solvent
separation

Recycled solvent

+
BTX -Unproces- BTX Purified BTWsolvent
(make up) BTX
extraction sed extracy purification extract * separation
__+

-
Figure
7.3 1 BIX extraction process simplified flow diagram.

solventlwashing water separation,

solvent purification.
The purpose of aromatics extraction is to ensure the process yield, i.e. to
deliver a raffinate practically free of aromatics. It is a simple feed and solvent
countercurrent extraction. The resulting raw extract can therefore not be
totally free of paraffinic and naphthenic impurities and needs t o undergo fur-
ther purification.
The purpose of purifying the aromatics is t o make them comply with their
final purity specifications with respect to non-aromatic hydrocarbons. It is per-
formed either by reextraction or by extractive distillation of the impurities
contained in the raw extract.
432 Chapter 7. SOLVENT EXTRACTION
IN THE OIL kRUSTRY

In processes based on reextracting impurities, the raw extract is sent to an

extractor in countercurrent flow with a secondary solvent, for example a paraf-
finic hydrocarbon, which can readily be separated by distillation from the
hydrocarbons in the feed. This operation is often termed "back wash". The
flow rate of paraffinic hydrocarbon needs to be adjusted so as to re-extract the
non-aromatic impurities selectively. The extraction and back wash steps can
be performed in two adjacent zones of the same extractor, which is fed with
the extraction solvent at the top and with back wash paraffin at the bottom
and the feed at an intermediate level.
Processes based on purifying the raw extract by extractive distillation take
advantage of the fact that a solvent that is selective in liquid-liquid conditions
is also selective in vapor-liquid conditions. By distilling part of the hydrocar-
bons in the extract, the purity of the residual aromatics is therefore enhanced.
The distillate, made up of a mixture of the most volatile aromatics and non-aro-
matics, is sent back to the extraction step.
Solvent regeneration is the operation that separates the aromatics from
the solvent. It generally consists of a simple distillation or water stripping
operation, in a partial vacuum when the solvent's thermal stability so requires.
Exceptionally, re-extraction by a hydrocarbon different from those in the feed
may be preferred.
As for the ancillary operations, it is always necessary to wash the raw
extraction raffinate with water to extract the solvent carried over by its low
but non-zero solubility in paraffins and naphthenes. For obvious economic
reasons the solvent recovered in this way is recycled into the process and this
means a solvent-waterseparation operation by distillation. In addition, heavy
products may be formed during the different operations (by polymerization of
unsaturated compounds, breakdown of the solvent or other unwanted chemi-
cal reactions). To keep them from accumulating, most aromatics extraction
processes include a small solvent purification unit. It is usually a vacuum dis-
tillation unit that treats approximately 1% of the solvent in circulation on a
continuous or batch basis.
Lastly, the mixture of pure BTX must be fractionated into benzene, toluene
and xylenes. Fractionation causes no particular difficulties and is carried out
by two conventional distillation processes, or sometimes three if the xylenes
need their tail ends cut off.

7.3.6 Industrial Processes

Industrial processes can be classified into two types, according to whether
they operate with a single solvent, which can be a homogeneous mixture of
solvents, or a pair of solvents that are immiscible with each other.

7.3.6.1 Single Solvent Processes

These processes generally follow the same sequencing for the main opera-
tions: aromatics extraction, purification by extractive distillation and solvent
 ChaDter 7. SOLVENT IN THE OILINDUSTRY 433
EXWACTION

regeneration by conventional distillation. They can be differentiated from one

another by the type of solvent, the equipment and specific implementation fea-
tures that are the result of solvent properties.

a. The Udex Process

The Udex process was developed jointly by UOP and Dow Chemical Co. with a
diethyleneglycol P E G ) based solvent and has been applied since 1952. For
some ten years it was the only one to produce aromatics of petroleum origin
with a purity of over 99.8%and an average yield in excess of 95%.
It is no longer considered to be a competitive process today. DEG in fact
has a relatively weak capacity for aromatics, entailing the use of high extrac-
tion solvent ratios, from 6 to 20, depending on the feed. More efficient solvents
have replaced DEG.

b. The Shell/UOP Process

Royal Dutch Shell studied ways of replacing glycol with a better solvent and
finally settled on tetramethylenesulfone or sulfolane, which they began to mar-
ket in 1961 jointly with UOP. It quickly became the leading process. Compared
to glycol, sulfolane exhibits a number of advantages, in particular a greater
capacity for aromatics, a higher boiling temperature, better thermal stability
and less sensitivity to oxidation.
A simplified diagram of the Shell/UOP process is shown in Figure 7.32. A
description will give a better understanding of all of the processes of this type.
The feed is let in the lower part of an extractor which also receives sul-
folane with approximately 1%wt of water at the top of the column, and at the
bottom the recycled hydrocarbons. The extractor is a rotating disk contactor
(RDC, initially patented by Shell), or a sieve tray column, with an overall aver-
age efficiency equivalent to around ten theoretical stages. It operates at a tem-
perature of approximately lOO"C, under a pressure of 0.2 to 0.4 MPa. The
solvent ratio is adjusted so as to meet aromatics yield specifications and gen-
erally ranges between 3/1 and 5/1 by weight.
The raw extract coming from the extractor is sent to the top of the extrac-
tive distillation column, which has some thirty trays and operates at atmo-
spheric pressure, with a reboil ratio regulated so that the aromatics at the
bottom of the column comply with non-aromatics content specifications. The
overhead vapors include non-aromatics, a fraction of the lightest aromatics
and water carried over by azeotropic phenomena with the hydrocarbons.
After condensation, the hydrocarbon phase is left to settle out and is then sent
back to the extractor.
The purified extract is sent to the aromatics stripper, which also has some
thirty trays and is supplied with steam at the bottom of the column. It oper-
ates under reduced pressure (300-400 mmHg) so that the bottom temperature
remains compatible with the solvent's thermal stability, i.e. approximately
180°C. The column is regulated so as to produce aromatics complying with
sulfolane specifications at the top and the regenerated solvent at the bottom,
which is recycled to the extractor.
 P
0
Extractor Extractive distillation Aromatics P
column stripper

t3
Steam

Raffinate

7.32 Simplified flow diagram of the Shell/UOP process. S: solvent (sulfolane). A: aromatics. NA: non-aromatics.
 Chapter 7. SOLVENT EXTRACTION
IN THE OIL INDUSTRY 435

The water recovered at the top of the aromatics stripper is used in wash-
ing the raw raffinate to extract the sulfolane, then reinjected into the same
stripper, at least partially in the form of steam after partial topping (not rep-
resented in the figure) to remove the traces of solubilized non-aromatic hydro-
carbons and thereby avoid polluting the aromatics.

c. The Arosolvan Process by Lurgi

This process uses N-methylpyrrolidone as a base solvent (NMP). In its first
version industrialized in 1962, water (12-14%) was added to the NMP to opti-
mize its extraction efficiency. Later on it was discovered that a mixed NMP-
glycol solvent gave much greater energy savings, since it avoided excessive
vaporization of water with the hydrocarbons during extract purification by
extractive distillation. The proportion of glycol depends on the aromatics con-
tent in the feed and generally ranges between 40 and 50%.A particular feature
of the process is the Lurgi extractor, built on the principle of a stack of mixer-
settlers, whose advantage is that it can be easily extrapolated up to a diameter
of eight meters. The solvent properties of NMP mean operating the extraction
process at close to ambient temperature, i.e. some 50°C.

d. The Tetra Process by Union Carbide

This process is named after the solvent it uses: tetraethyleneglycol with 4%
water added. It was designed in 1968 to debottleneck Udex installations.
Replacing diethyleneglycol by tetraethyleneglycol, which has a greater solvent
power, allows the solvent ratio to be halved and the treatment throughput to
be increased by 80%using the same flow scheme and the same extraction tem-
perature (approximately 150°C). For grass roots plants, Union Carbide pro-
poses specific column internals: sieve trays with multiple upcomers (MU) for
the extractor, allowing higher flow rates than conventional trays, and multiple
downcomer (MD) trays for the non-aromatics stripping column. This column
requires internals adapted to high vapor/liquid flow rate ratios in order to be
effective. The washing water circuit for the aromatics and the raffinate, along
with the solvent purification by adsorption are all specific to the Tetra
process.
e. The Morphylex Process by KruppKoppers
This process utilizes N-formylmorpholine (NFM) with 4% water. Extraction is
carried out in a conventional sieve tray column at a high temperature, approx-
imately 180°C. The general flow scheme exhibits two original features com-
pared to the previously described processes. One is the non-aromatics stripper
which is designed like a complete extractive distillation column, i.e. with pure
solvent introduced at the top of the column and consequently a stripping sec-
tion and a rectifying section. The advantage of this configuration is to limit aro-
matics moving upward in the distillate, which for one thing also limits the
amount of water vaporized with the aromatics by azeotropic phenomena, and
for another curbs hydrocarbon recycling to the extractor. The second original
feature resides in the hydrocarbon washing water circuit: the water coming
from the top of the two strippers is used to wash the raffinate. Then the water
436 Chapter 7. SOLVENT EXTRACTION IN THE O I L INDUSTRY

joins the regenerated solvent coming from the bottom of the aromatics s t r i p
per in order to rehydrate it before it is recycled to the extractor.

7.3.6.2 Dual Solvent Processes

Processes using two solvents are different from those described in the pre-
ceding section because the aromatics purification and solvent regeneration
steps are not performed by extractive distillation and stripping, but by re-
extraction with a secondary solvent. The secondary solvent must of course
not be too miscible with the main solvent and it also has to be readily sepa-
rated from feed hydrocarbons by distillation.
Different processes have been proposed, particularly with tetraethyleneg-
lycol (Union Carbide), N-formylmorpholine (Snam Progetti), or dimethylsul-
foxide (IFP) as an extraction solvent, and a paraffin that may be either more or
less volatile than the feed hydrocarbons as a reextraction solvent. The
sequencing of steps is similar from one process to another. It will be illustrated
by taking the IFP process as an example in its most commonly used version,
i.e. with the dimethylsulfoxide @MSO)-butane solvent pair.
The DMSO process, shown with a simplified flow diagram in Figure 7.33,
was developed by the Institut Frangais du Petrole in the early sixties and
industrialized in 1967.
The feed enters the midsection of the extractor, a Kuhni column, whose
internals are sized taking the properties of the phases into account at each
level. DMSO with 4 to 6% water added enters the top and butane, called back
wash butane, enters the bottom. Extraction takes place at an average temper-
ature of 30°C under a pressure of 4 to 6 bar abs., in order to keep the butane in
the liquid phase. The solvent ratio (DMSO/feed) is controlled so as to ensure
the aromatics yield and depending on the feed composition, it ranges between
3/1 and 6/1. The back wash ratio (butane-feed) is adjusted according to the aro-
matics purity specifications, it is generally something in the region of 0.15/1.
The raw raffinate from the extractor is washed with water to eliminate the
solubilized DMSO, then distilled to separate the butane, which is recycled as a
back wash. The extract is sent to the reextractor, which is also a Kuhni col-
umn, where it is contacted countercurrently with butane in the same condi-
tions of temperature and pressure as those of the extractor. The ratio of
butane/extract flow rates, on the average 0.35/1 by weight, is defined so as to
reextract the aromatics totally.
The regenerated DMSO is recycled directly to the extractor. The butane,
loaded with aromatics, is washed with water to remove the saturation DMSO,
then distilled and recycled to the reextractor. The hydrocarbon washing
water, loaded with DMSO, is distilled under a partial vacuum (200 mmHg) and
recycled to the washing contactors. The recovered DMSO is reincorporated
into the solvent circuit.
The advantage of this system is to exploit the excellent extractive proper-
ties of DMSO without any risk of degrading it. More generally speaking,
another advantage that is sometimes pointed out is its aptitude to extract the
 Extractor Re-extractor Washing DMSO recovery Extract Raffinate
column debutanizer debutanirer
I I

L I
L

-
Feed
J

1
Water

L -
Raffinate

-
Aromatics
I Re-extraction butane
Back-wash butane I 1

L
Figure
7.33 Aromatics extraction by DMSO. IFP process.
438 Chapter 7. SOLVENT EXTRACTION IN THE O I L INDUSTRY

aromatics from a wide cut, since the problem of separating the aromatics from
the solvent by distillation no longer exists. However, it should be noted that
the wider the cut, the higher the flow rates of extraction and re-extraction sol-
vents per unit of aromatics extracted. There is thus an economic limit to the
treatment of wide cuts and no industrial application is known any wider than
BTX extraction.

7.3.7 Recovery Ratio and Aromatics Purity

All the processes presented are capable of producing aromatics with the most
stringent purity specifications, sometimes after a clay finishing treatment to
eliminate unsaturated hydrocarbons, which are the most difficult to separate
from aromatics.
The data presented by way of example in Table 7.28 for different operating
cases with the Udex, Sulfolane, Arosolvan and DMSO processes show that the
recovery ratio of each aromatic is comparable from one process to the other.
In contrast, for all processes, the recovery ratio drops quite dramatically from
benzene (> 99.5%) to xylenes (9596%). It becomes low (70% on the average)
for aromatics with nine carbon atoms.

Process
-
Udex 1 Sulfolane Arosolvan DMSO

a
-
b a b 7
Benzene 7.5 99.8 34.0 99.8 34.0 99.5
Toluene 21.2 99.0 19.0 99.0 19.0 99.0
C, aromatics 20.0 33.5 95.0 17.0 96.0 17.0 95.0
C, aromatics 0.9 0.1 60.0 0.2 70.0 0.2 -
Non-aromatics 50.4 36.2 29.8 29.8
Total 100.0 100.0 100.0 100.0

k?l Composition of the feed (a) and recovery ratio (b) (% wt).

7.3.8 Economics
A rigorous economic comparison of the processes is impossible to establish
because the information published never involves the same base case. We will
confine our discussion therefore to a reproduction in Table 7.29 of investment
and consumption data which might be useful during a preliminary capital bud-
geting project stage (Chauvel, 1985; data updated in mid-1992). The data con-
cern the applications presented in Table 7.28. The only obvious indication that
can be deduced is that the Udex process is no longer competitive, as men-
tioned earlier.
 Chapter 7. SOLVENT EXTRACTION
IN THE OIL INDUSTRY 439

Company

Process

Battery limits
UOP-DOW :;h

Udex

91
S 1 ,"
Sulfolane * Arosolvan DMSO *

72
IFP

investments (MF) **
Initial solvent inventory
and miscellaneous (MF)

Consumption per ton

4.0
2-6 I 2.5
3.2

of extracted aromatics:
Feedstock feed (t) 2.00 1.65 1.45 1.45
By-products: raffinate (t) 1.oo 0.65 0.45 0.45
Utilities:
Steam (t) 1.5 0.8 0.9 1.o
Electricity (kWh) 30 11 18 8
Cooling water (m3> 40 35 3 17
Solvent and chemicals (FF) 5.0 3.0 0.8 1.2

Manpower
(number of shift operators) 2 2
I 2
* New generation.
** Investments do not include BTX fractionation.
.
Aromatics extraction. Economics. Production capacity: 200 000 t/y of extracted
aromatics (Western European conditions, mid-1992).

In conclusion, producing benzene and toluene or benzene, toluene and

xylenes from a c& or c& cut from steam cracking or catalytic reforming
effluents requires industrially tried and true processes that are able to deliver
aromatics that meet the most stringent specifications and have an excellent
yield. These processes differ from one another as to the nature of the extrac-
tion solvent, which determines a good proportion of the different implemen-
tations and operating conditions. It is, however, difficult to differentiate
between them from an economic standpoint in an initial analysis. Selecting a
process for a particular application must therefore be based on a thorough
study, taking all local parameters in particular into consideration.

7.4 LiquideLiquid Extraction Equipment (Internals)

7.4.1 General Introduction
7.4.1.1 Geometrical Characteristics of an Ideal Extraction System
The principle itself of liquid-liquid extraction dictates the functions that the
extractors need to fulfill.
440 Chaoter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY

In order to reach high efficiency it is necessary to have:

the largest contact area between phases (interfacial area) to promote
transfer between them;
the shallowest depth of liquid to avoid the detrimental effects of diffusion
phenomena in the mass of this phase;
the least dispersion of flow velocity for each phase so that contact times
are as homogeneous as possible, thereby preventing any fraction of the
phase from getting through the system without being sufficiently treated.
Compliance with these constraints directly conditions the choice of equip-
ment. For a given type of equipment, extraction generally increases with resi-
dence time.
For a given geometry, operating conditions - particularly temperature -
will have an influence on extraction efficiency either directly (via partition
coefficients) or indirectly (specific gravity, viscosity, etc.).
The choice of a technology and the design of internals for a liquid-liquid
extraction system are guided by several criteria, particularly:
total flow rate;
required efficiency (number of theoretical stages);
difference in specific gravity between the two phases;
ratio of phases (amount of light phase/amount of heavy phase);
dispersion behavior (stability, formation of a stable emulsion, etc.);
presence of solids (degradation products, sediments, etc.);
operating conditions (especially pressure);
fouling tendency;
residence time limited by the risk of one of the products becoming
degraded;
corrosivity of one of the products;
flammability of one of the products;
area and height available for the extractor;
specific maintenance and safety constraints, etc.
Other criteria, often based on the operator's own experience, may be
involved, such as resistance to fouling or corrosion, absence of "no-flow"
zones.
The price of the complete system (extractor + auxiliaries) is of course one
of the determining factors in selection (Chauvel et al., 1976).

7.4.1.2 Backmixing
As the reader may notice while studying the equipment presented below,
designers have often attempted to reduce backmixing.
 Chapter 7. SOLVENT ExrRaCTlON IN THE O I L INDUSTRY 441

ixing of the light phase

-No backmixing
- - -With backmixing

B I
/
. \
\
I

.....
\
I \
I \

I
I
I
-_ -
Residence time

Residence time

Figure
7.34 A: Backmiwing stream. B. Influence of backmiwing on residence time.
C. Consequence of residence time dispersion.
442 Chapter 7 SOLVENT EXTRACTION IN THE OIL INDUSTRY

Backmixing is an axial mixing phenomenon (Fig. 7.34A) which lengthens

residence times of the relevant phase (Fig. 7.34B). In effect, when the phe-
nomenon occurs, part of the relevant phase tends to turn round and round in
the extractor, and this prolongs its residence time, while another portion will
have to transit through faster since its space is occupied by the part that is
backmixed.
Backmixing has two negative effects on extractor performance:
lower efficiency, by accelerating the transfer of part of one or of both
phases present (zone @ of Fig. 7.34C);
lower apparent extractor capacity, since part of the stream is going
around in a loop, thereby needlessly overloading the equipment.
The increased residence time of part of a sensitive product may cause it to
become degraded (zone @ in Fig. 7.34C).

7.4.1.3 Different Contacting Systems between W o Phases

(Wuithier et al., 1972; Trambouze et al., 1988; Blass et al., 1994)

The simplest liquid-liquid extraction column is the spray column where the
dispersed phase is simply sprayed inside the continuous phase. The advan-
tage of this type of column is its large capacity, but its efficiency is low because
of the high degree of backmixing.
Extractors have been equipped with internals designed to improve effi-
ciency, mainly by lengthening residence times for the dispersed phase and by
bettering dispersion quality.
Better dispersion generally means a dissipation of energy. Energy can be
provided by the liquid (pressure drop and static liquid head in static mixers,
packed columns and fixed tray columns) or by agitation produced by column
internals, either a rotary movement (mixer-settlers, RDC, Kiihni, Scheibel
columns) or oscillation (pulsed columns, Karr alternating movement trays in
columns).
The experience gained with specific equipment for a particular application
also has great influence, as it weighs in the compromise between potential
risks/improvement. This restricts to a not inconsiderable degree the penetra-
tion of new technologies for well known applications.

7.4.2 Static Systems

7.4.2.1 In Line Mixers
In line mixers, also called static mixers (Fig. 7.35), allow phases to mix in a co-
current system. They can be used as mixers in mixer-settler systems.
They allow quite good dispersion with reduced pressure drop and make
the flow velocity in pipes more uniform, thereby making residence time dis-
tribution narrower. This is particularly significant for laminar flow (viscous
systems).
 Chapter 7. SOLVENT EXTRACTION
IN THE OIL INDUSTRY 443

Sulzer S W
disperser.

Except in multiple mixer-settler systems, this type of equipment is usually

reserved for easy extraction where the transfer occurs quickly.

7.4.2.2 Spray C o l u m n s (Fig. 7.36)

This is the simplest system imaginable, where the continuous phase flows
freely and is placed in contact with the dispersed phase which goes through it
in countercurrent flow.
Mass transfer between the phases is limited by two phenomena:
the dispersed phase is not remixed;
backmixing disturbs the plug flow conditions and therefore makes resi-
dence time distribution wider.
Simple spray systems (with no device to redistribute the dispersed phase)
are restricted to the easiest extraction operations, because their efficiency is
low.
The types of equipment described below all have a common feature: a sig-
nificant decrease in backmixing, the phenomenon that reduces spray column
efficiency.

7.4.2.3 Baffle C o l u m n s (Fig. 7.37)

In order to improve the efficiency of spray columns, baffles (discs and rings or
segments of discs) have been installed between the diffuser and the interface.
The baffles achieve intermediate remixing of the dispersed phase. The contact
between the heavy phase and the light one occurs in crosscurrent flow.
444 Chapter 7. SOLVENT EXTRACTION
IN THE O I L INDUSTRY

t +
$dp Light
phase

Figure
Spray column. 7.36

Figure
7.37 Baffle column.

' Continuous
heavy phase
1
 Chapter 7. SOLVENT IN THE OILINDUSTRY 445
EXTRACTION

The efficiency of this type of system is low (0.05 to 0.2 theoretical stage per
baffle), but the large through cross sections and simple construction can
prompt its use in fouling or corrosive systems.

7.4.2.4 Columns with Sieve Rays (Fig. 7.38)

The trays of these extractors collect and remix, then redisperse the dispersed
phase. Extractor efficiency is significantly improved (about 0.7 theoretical
stage per tray) and, despite their greater sensitivity to fouling and corrosion,
they are usually preferred to baffles.
This type of column is utilized for example in aromatics extraction pro-
cesses (Udex, Formex, Morphylan, etc.).

7.4.2.5 Columns with Packing (Fig. 7.39)

Depending on the relative affinity of the phases with the packing construction
material, two operating modes can be observed:
One is where the dispersed phase wets the packing. While the three
extractors presented before worked by dispersing one phase in the other,
packed columns work mainly by a film effect of the “discontinuous”
phase. The phase is redispersed on the surface of the packing each time
there is a break in the packing.
The second is where the continuous phase wets the packing and the
other phase is redispersed each time there is a break in the packing.
All types of packing are used (rings, saddles, structured packing).
First, ceramic saddles were widely used, but they have now been phased
out except in special cases when the advantages of their corrosion resistance
or surface phenomena seem more important than their limitations:
severe pressure drop and reduced capacity (low void fraction);
brittleness during operation (breakage after a packing bed expansion)
and during installation, which often must be carried out in a column full
of liquid.
Ceramic saddles and rings have generally been replaced by metal ones
(lMTP by Norton, Pall rings, CMR by Glitsch, etc.), which are easier to handle
and allow column sizes to be smaller.
The presentday trend is to attempt to use structured packing to improve
equipment performance even more, but the gain seems relatively small com-
pared to the extra costs and added constraints:
lower corrosion resistance (structured packing is generally thinner than
saddles and rings);
difficult to dismantle for cleaning (note that fouling is sometimes not too
severe, however).
446 Chapter 7. SOLVENT EXTRACTION
IN THE OIL lNDUSTRV

I
, Heavy phase
(continuous)
Figure
Extraction column with 7.38
sieve trays (light phase
dispersed).

Interface
Heavy phase distributor

Packing bed

-
Light phase distributingtray Figure

Redistributor 7.39 Extractor with packing

(light phase dispersed).

Packing bed

Light phase distributing tray

Light phase inlet
 Chapter 7, SOLVENT EXJRACTION
IN THE OIL INDUSTRY 447

The smaller the packing (tight mesh or small diameter rings), the higher
the efficiency. At the same time, it should be noted that the capacity is smaller
and the sensitivity to fouling increases.
As a rough estimate in order to have an homogeneous flow, it is considered
that the extraction column diameter should be at least 8 to 10 times greater
than the diameter of a ring or than the interval of structured packing.
Usually, efficiency (number of theoretical stages per meter) improves with
the increase in specific flow rate. Of course, this improvement is limited by
extractor flooding or the impossibility of keeping a stable interface zone.
Generally speaking, the efficiency of a packed column is 0.5 to 4 theoretical
stages per meter of packing height.
Packed columns are used for example in the following cases:
furfural treatment of lube oil cuts;
elimination of H,S from petroleum cuts:
extraction of aromatics from C,-C, cuts (Duosol process);
extraction of phenols from refinery and coking effluents (Phensolvan pro-
cess).

7.4.3 Agitated Systems

7.4.3.1 Mixer-Settlers (Fig. 7.40)
In these systems dispersion is effected by the input of outside energy in one
section of the unit, then coalescence and homogenization of the dispersed
phase take place in another section. Efficiency can go up to 90% per stage.
Conventional mixer-settlers, where each stage is separate, are simple to
operate but take up a lot of space on the ground. They are especially found in
installations with large throughput flow rates (manufacture of phosphoric acid
for example).

Heavy
phase

Light
phase

Figure
7.40 Group of mixer-settlers working countercurrently.
448 Chapter 7. SOLVENT EXTRACTION
IN THE OIL INDUSTRY

Lurgi uses a special design for its Arosolvan aromatics extraction process,
where mixer-settlers are stacked up one on top of the other in the same col-
umn (Mehner extractor, Fig. 7.41).

1
.................. ....................,..
____________..___.__
1.41 Stacked Lurgi rnixer-
_I settlers.
Figure

L
I

Mixer-settlers are appreciated because of

their high efficiency per stage,
their great flexibility,
the extrapolation reliability.
Their drawbacks are:
high cost of multistage systems;
sometimes large site coverage.

7.4.3.2 Rotating Systems

a. RDC and ARDC (Rotating Disc Contactor
and Asymmetric Rotating Disc Contactor)
In an RDC (Fig. 7.42)meman and Olney, 1955), the active zone consists of com-
partments formed by a series of fixed rings, between which there are discs
made to rotate by a central shaft.
The upper and lower zones (product exits and interface) can be separated
from the contact zones by a metal mesh or a grid, designed to limit agitation
in the settling zones.
 Chapter 7, SOLVENT ExTFlACTlON IN THE OIL INDUSTRY 449

(stator)

- -
- -
Figure
Rotating disc contac- 7.42 - -
tors (ROC). -
- -
- -
- -

The ARDC is a later generation RDC where the shaft is off center so as to
provide a partial settling zone for the phases at each stage.
The light fraction enters the RDC in the lower zone and goes through the
column in countercurrent flow with the heavy phase which enters the upper
zone.
One of the phases is dispersed in each compartment included between two
stationary rings and the size of droplets is controlled by the disc’s rotation
speed.
Except for the end zones of an RDC, there are no dispersion and coales-
cence sequences. The liquid’s movement consists of two component parts: a
rotation and a slower vertical movement, guided by the discs and rings.
The operating (and design) parameters are as follows:
Rotation speed. Capacity decreases and efficiency increases with a
higher rotation speed. Efficiency can, however, reach a maximum before
becoming counter-productive due to a backmixing phenomenon in the
extractor.
Diameter of rotating discs. Capacity decreases and efficiency increases
with a larger rotor diameter.
Diameter of central opening in stator rings. Capacity increases and effi-
ciency decreases with a larger opening in the rings.
Vertical spacing between rings. Capacity increases with wider spacing
between rings. Efficiency is relatively constant for close spacing, then
decreases with wider spacing.
450 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY

Column diameter. If proportions are maintained, the column diameter

has little effect on efficiency and on specific capacity.
Feed flow rate. Generally speaking, efficiency reaches a maximum when
the flow rate increases and then the effects of backmixing become
counter-productive.
Ratio of phases. Efficiency increases when the dispersed phase/continu-
ous phase ratio increases.
Today many RDCs are used in a static position. This may seem surprising,
but the decision to stop rotation can be made for several reasons, the most
important of which are listed below:
Maximum increase in capacity
Maintenance of rotor mechanical drive gear
More economical operation.
In cases where extraction efficiency has proved to be insufficient, opera-
tors have sometimes preferred to add an extra column, often with packing,
along with the static RDC.
b. Other Rotating Extractors
Other extractors have been designed in an attempt to enhance efficiency by:
reducing backmixing;
improving remixing of the dispersed phase between each stage.
The more conventional models are presented in Table 7.30.

Means of Settling Separation

Type of column agitation

ARDC Disc Horizontal 7.43a

partitions

Scheibel Blade turbine Yes Horizontal 7.43b

partitions

Oldshue-Rushton Blade turbine + No Rings 7.43c

vertical partitions

Kuhni Turbine No Sieve trays 7.43d

PI Rotating extractor models.

Rotating extractors are used in several important applications:

RDC extractor: aromatics extraction (Shell Sulfolane process);
SEb’L
I I
452 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY

Light
-)phase out

-
Figure
7.43c

Heavy
-phase out

Oldshue-Rushtonextractor

Kijhni extractor Hofrnann-Laroche extractor

Figure
7.43d
 ChaDter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY 453

Kuhni extractor (Dongaonkar et al., 1991): aromatics extraction (IFP

DMSO process).
The other extractors and the Kuhni column are also used in the chemicals
and pharmaceutical industries, in treating effluents, etc.

7.4.3.3 Alternating Extractors

In alternating extractors, dispersion is performed by the relative agitation of
the mass of liquid and sieve trays.
There are two types of equipment, according to the moving part:
extractors with moving trays;
extractors with stationary internals and the mass of liquid agitated by
regular pulsing.
a. Extractors with Moving Trays
The Karr extractor (Holmes et al., 1987; Cusak et al., 1991) (Fig. 7.44) consists
of a series of perforated trays (55 to 60% open cross section, perforations from
7 to 14 mm in diameter) connected together to a shaft that has an alternating
vertical movement.

t, 7support
Vertical bar

- PTFE ring

- Sieve trays

Figure - Anti-backmixing
baffle
Karr column. 7.44
454 Chapter 7. SOLVENT EXTRACTION IN THE OIL INDUSTRY

At regular intervals there is a solid ring designed to reduce backmixing. To

guide the moving parts and prevent metal on metal contact between them and
the column, there are rings made of PTFE with a larger outside diameter than
the metal trays. They are placed at the ends of the column and sometimes mid-
way between.
There are other columns with moving trays, but the principle is identical.
The main differences lie in the details of the perforated trays (open cross sec-
tion, etc.) and in the settling zones between the moving trays.
Extractors with moving trays are used for example in:
aromatics extraction (Union Carbide TEG process);
phenol extraction from refinery water (Chempro process).

b. Pulsed Columns
In pulsed column extractors, internals are stationary and the complete mass
of liquid in the column is agitated by regular pulsations. This configuration
offers a great many advantages since it reduces the number of moving
rnechanica parts and attenuates problems of transmission of movement inside
the column.
Pulsed columns are differentiated mainly by:
the internal equipment in the column which can consist of:
- stationary trays (perforated trays occupying the whole cross section of
the column or discs and rings, etc.) (Fig. 7.45a),
- packing (Fig. 7.45b);

Gas overhead

Heavy
phase
inlet Figure
Pulsed column with
trays (mechanical
pulsation).

Bellows
system
Light
phase
inlet
 Chapter 7. SOLVENT EXTRACTION IN THE O I L INDUSTRY 455

the means of causing the liquid mass to pulsate:

- mechanical (for example with a bellows system at the foot of the col-
umn) (Fig. 7.45a);
- pneumatic (Fig. 7.45b).

Figure
Pulsed column with 7.45b
m
packing (pneu- C
I
P
matic system). 2 _ _ _I
a
Compressed gas

Whereas in a static column with trays only one phase passes through the
tray perforations with the other one passing through downcomers, in a pulse
column the two phases pass alternatively through the same holes. This keeps
one phase dispersed, and the continuous phase agitated, i.e. homogenized,
between two stages.
Packed pulse columns work like static columns but should be used only
with systems where the continuous phase wets the packing preferentially so
that the dispersed phase can progress regularly.
Efficiency is better than with a static packed column but remains relatively
low. It is, however, sufficient for a number of applications in refineries (treat-
ment of effluents, for example).
Pulse columns were first developed for the nuclear power industry where
it is very important to confine the medium that is going to be extracted. They
are now used more and more in widely different applications such as treat-
ment of effluents, pharmaceuticals, etc.

7.4.3.4 Centrifugal Extractors

To make it easier to separate the phases some designers and equipment man-
ufacturers (SGN/Robatel, Westfalia, Podbielnak, Alfa-Laval, etc.) have devel-
oped extractors that create an artificially accelerated field of gravity.
456 Chapter 7. SOLVENT EXTRACTION IN THE OIL /NOUSTRY

These columns offer several advantages:

It is possible to handle phases with similar specific gravities.
Residence times are shortened (application to unstable products).
Equipment size is reduced.
They are only seldom found in the oil and gas industry because of the fol-
lowing drawbacks:
high speed mechanical systems (reliability, maintenance);
high prices;
usually limited flow rate.

7.4.4 Conclusion
Generally speaking, the choice of an extractor is greatly influenced by the
body of experience for each application. This is particularly perceptible when
a grass roots unit is being built with a tried and true process or when equip-
ment in an existing unit is replaced.
Developing a new process usually goes through pilot plant steps in the lab-
oratory and then on a semi-industrial scale. The equipment that will be cho-
sen on the industrial scale must be selected during the pilot phase, since the
rules of extraction equipment extrapolation are normally reliable. The opera-
tor will tend to base his choice on the experience acquired in the pilot phase
and will only seldom take the risk of trying a new type of extractor on an indus-
trial unit.