J.

of Supercritical Fluids 47 (2009) 611–618

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids

journal homepage: www.elsevier.com/locate/supflu

Review

Modeling and optimization of supercritical fluid processes

M.S. Diaz, E.A. Brignole ∗
Planta Piloto de Ingeniería Química (PLAPIQUI), Universidad Nacional del Sur-CONICET, Camino La Carrindanga Km 7,
Bahía Blanca 8000, Argentina

a r t i c l e i n f o a b s t r a c t

Article history: In this work, we discuss recent advances on thermodynamic modeling, process simulation and optimiza-
Received 30 July 2008 tion of supercritical extraction and fractionation processes. The increasing availability of experimental
Received in revised form data, as well as the need for estimation of economical viability, has encouraged the development of
10 September 2008
supercritical processes computer models. This review will discuss main approaches on phase equilib-
Accepted 11 September 2008
rium thermodynamic modeling of high-pressure processes, as well as process simulation. In this case,
two main supercritical processes are considered: extraction of valuable products from solids and coun-
Keywords:
tercurrent fluid–fluid extraction and fractionation. In the latter process, again two main approaches are
Supercritical fluid extraction
Simulation model
considered: rigorous models based on equilibrium stages and mass transfer models in packed columns.
Thermodynamic modeling Finally, optimization procedures will be reviewed, both based on rigorous models and on experimental
data.
© 2008 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611
2. Phase equilibrium in natural product mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
3. Process simulation models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
3.1. Extraction of valuable products from solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
3.2. Countercurrent fluid–fluid processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
3.2.1. Models based on equilibrium stages at steady state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
3.2.2. Models based on mass transfer approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
4. Process optimization approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616
Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617

1. Introduction of experimental information on supercritical processes is nowa-

days available. At the same time the development of supercritical
The computer aided design, simulation and optimization of processes was accompanied by improved models to describe the
chemical processes has had an impressive growth during the phase equilibrium of multicomponent mixtures at high pressures,
last 50 years. During the same period supercritical fluids tech- and the prediction of physicochemical properties, viscosities, dif-
nology, mainly based on carbon dioxide, has found applications fusivities and thermal conductivities in supercritical media. The
in many fields, from separations, chemical reactions to particle combination of these methods together with rigorous mathemat-
micronization and materials processing, due to its tunable solvent ical modeling of different unit operations opens the way to the
power, good transport properties and very low interfacial ten- application of computer modeling and optimization of supercrit-
sions. From bench scale to pilot plant and industrial size a wealth ical processes, which is of great interest, considering the cost of
high-pressure pilot plant facilities and industrial scale experimen-
tation.
∗ Corresponding author. The present review will cover advances in thermodynamic mod-
E-mail addresses: [email protected], [email protected] eling, simulation and optimization of supercritical extraction and
(E.A. Brignole). fractionation processes in the last 20 years.

0896-8446/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.supflu.2008.09.006
612 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618

2. Phase equilibrium in natural product mixtures

The analysis of supercritical fluid processes requires the support

of simulation tools based on rigorous modeling of complex phase
equilibrium behavior. In these processes a new dimension, the sol-
vent density, is added to tune the operating conditions and process
scheme to the separation needs. This property is highly sensitive
to changes in pressure, temperature and supercritical solvent com-
position. The development of thermodynamic models to describe
the phase equilibrium in supercritical processes has been possible
thanks to the wealth of experimental data that have been presented
in the literature on high-pressure phase equilibria. A very extensive
review of systems studied and the applied experimental techniques
has been presented by Christov and Dohrn [1]. The experimen-
tal techniques can be classified as dynamic or flow, analytic (fixed
or variable volume cell), synthetic, etc., examples of apparatus for
different techniques are given by Richon [2].
Group contribution equations of state have been widely used
for predicting thermodynamic equilibrium in supercritical fluid
processes, as a great variety of natural products can be repre-
sented with a limited number of functional groups. The Group Fig. 1. GC-EOS predictions and experimental data for CO2 -orange peel oil.
Contribution Equation of State (GC-EOS) model was proposed by
Skjold-Jorgensen [3,4] to study gas solubilities in non-ideal mix- that the Peng–Robinson equation of state, using the one-fluid
tures at high pressures. This model was applied for the prediction van der Waals mixing rules, provided good qualitative predic-
and correlation of solubilities of solvents in supercritical fluids by tions of double retrograde vaporization behavior for the ternary
Brignole et al. [5]. Temelli et al. [6] studied the thermodynamic system CO2 + limonene + linalool. At certain conditions, this phe-
modeling of the deterpenation of citrus peel oils with supercritical nomenon can cause the decrease in solubility of the less soluble
carbon dioxide and presented phase equilibriums data at tem- component, while having no adverse effects on the solubility of
peratures from 40 to 70 ◦ C and pressures from 8.3 to 12.4 MPa. the component to be extracted. More recently, Raeissi et al. [22]
Experimental data were correlated with a simple model using vapor used the Soave–Redlich–Kwong [18] equation of state (SRK EOS)
fugacity coefficients from the GC-EOS and unsymmetric convention together with the classical one-fluid van der Waals mixing rules as
activity coefficients. The original model takes into account only the thermodynamic model for orange peel oil deterpenation with
repulsive and dispersive interactions. Gros et al. [7] extended its supercritical fluids. They tuned SRK adjustable binary interaction
capability to treat associating compounds (GCA-EOS) in mixtures parameters to a combination of different literature data sets con-
of water and alcohols with non-polar gases, like propane or CO2 . sisting of binary isothermal P-x,y, ternary constant composition P-T,
Natural oils and derivatives are complex mixtures of glycerides and ternary isothermal isobaric x–y data of representative citrus oil
with fatty acids of different chain length and degree of satura- component(s), limonene and linalool, with not only CO2 as solvent
tion, but their molecular structure can be characterized with a but also with ethane.
few functional groups. Bottini et al. [8] and Espinosa et al. [9] pro- Several approaches have been proposed for the prediction of
posed an upgraded version of the GCA-EOS, as well as revised sets thermodynamic equilibrium in mixtures of fatty acid ethyl esters
of parameters for the use of this equation of state in the simula- with carbon dioxide. Coniglio et al. [23] studied phase equilibrium
tion of natural oils and derivatives processing with near critical predictions of supercritical carbon dioxide with fish oil derivatives.
fluids. Diaz et al. [10–12] tuned the Group Contribution Equation They explored the combination of the Huron–Vidal first order
of State to predict thermodynamic properties for mixtures com- (MHV1) and second order (MHV2) mixing rules [24,25] to combine
posed of citrus peel oils key components and carbon dioxide for the Soave–Redlich–Kwong equation of state with different excess
representing supercritical deterpenation of lemon and orange peel Gibbs energy-based models such as UNIFAC/UNIQUAC and van der
oil, respectively. Fig. 1 shows a good agreement between GC-EOS Waals models. They concluded that the MHV1 mixing rules used
predictions in multicomponent vapor liquid equilibrium [12] and with modified UNIFAC [26,27] and a quadratic expression for the
experimental data [13,14] for the system orange peel oil, modeled co-volume in the equation of state with a universal constant for all
as a mixture of linalool and limonene, and carbon dioxide. Sato systems rendered very satisfactory results. Riha and Brunner [28]
et al. [15] used the Peng–Robinson equation of state (PR-EOS) [16] measured and modeled phase equilibrium of 13 different fatty acid
with two binary temperature dependent parameters to describe ethyl ester mixtures and pointed out the need for further work on
the phase equilibrium properties of limonene and linalool and they the thermodynamic modeling of these multicomponent systems,
also discussed the process selectivity. Soave [17] applied the Soave- especially if separation of a real mixture is to be rigorously simu-
Redlich-Kwong equation of state [18] with Huron–Vidal mixing lated and optimized. Espinosa et al. [9] modeled multicomponent
rules to describe the solubility of solids in supercritical solvents, vapor–liquid, liquid–liquid and vapor–liquid–liquid equilibrium of
using two adjustable parameters, vapor pressure and an interac- fish oil fatty acid ethyl esters with carbon dioxide with the extended
tion factor in the expression for its fugacity coefficient in the gas GC-EOS. They introduced new pure groups for representing esters
phase. (CH3COO and CH2COO) and estimated binary interaction param-
Raeissi and Peters [19] studied the double-domed retrograde eters with the remaining groups. Additionally, Espinosa et al. [29]
vaporization phenomenon using the original Peng–Robinson equa- proposed a strategy for the determination of hard sphere diameter
tion of state [16] for the binary systems of ethane + limonene and for long chain esters with different degree of unsaturation as a func-
ethane + linalool. Espinosa et al. [20] applied the GC-EOS model to tion of pure esters van der Waals volume and the number of CH = CH
predict the double retrograde behavior of ethane + limonene and groups in the molecule. These authors modeled phase equilibrium
ethane + linalool. In a later study, Raeissi and Peters [21] showed for multicomponent mixtures made up of thirteen fish oil FAEE
 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618 613

representing fish oils esters and carbon dioxide. A comparison with mathematical modeling. Regarding solids processing, they reported
experimental data reported by Staby et al. [30] and Borch-Jensen more than 35 applications for the extraction of seed oils, essen-
et al. [31] showed good agreement with GC-EOS predicted values. tial oils or volatile compounds and to the adsorption/desorption
An alternative approach to equations of state to describe solute of terpenic model mixtures, based on the three mentioned mod-
solubilities in supercritical fluids is the use of empirical mod- eling approaches. Meireles [46] performed an exhaustive review
els. Chrastil [32] proposed a linear log–log functionality between on papers providing information on process parameters required
solute concentration in the supercritical fluid phase and its den- for economical analysis of supercritical fluid extraction from solids,
sity. He derived this expression applying chemical equilibrium and which is required to decide whether SFE can be considered among
entropy concepts. He reported agreement between predicted and the feasible alternatives to produce high added value natural
experimental solubilities in supercritical carbon dioxide for several extracts. Experimental data were classified into two types, accord-
compounds, including stearic acid, oleic acid, behenic acid, tribu- ing to the extractor vessel size. For capacities greater than 50 ml,
tyrin, tripalmitin, triolein, trilinolein, palmityl behenate, behenyl the SFE units were considered as process units, and for capacities
behenate, a-tocopherol, cholesterol, water, cafestol, naphthalene, lower than 50 ml, as analytical ones. She reviewed almost 40 appli-
anthraquinone and p-chloroiodobenzene. Adachi and Lu [33] pro- cations for process vessels and more than 15 for analytical ones.
posed a modification to Chrastil’s equation in which the association Del Valle and De La Fuente [47] presented an exhaustive review
number was calculated as function of solvent density. In a recent on mechanisms of oil transport within the solid matrix, including
paper, del Valle and Aguilera [34] suggested a further modification desorption from the solid, formation of a shrinking core of con-
by accounting for variations in the solute’s heat of vaporization with densed oil in a non-adsorbing porous matrix, and diffusion in a
temperature. They validated the model with experimental data for homogenous medium. They discussed on several simplifications of
several vegetable oils in a wide range of pressure and tempera- a general mass transfer as well as more complex two-stage models.
ture. They also analyzed trends for the external mass transfer coefficient
and effective diffusivity in the solid matrix from studies on super-
3. Process simulation models critical carbon dioxide extraction of oil from vegetable substrates.
Furthermore, they analyzed the high-pressure vegetable oil-carbon
3.1. Extraction of valuable products from solids dioxide phase equilibrium in connection with its influence on the
mass transfer process.
An important application of supercritical fluids is the extraction More recently, and based on the model developed by Cocero
of one or more components from natural solid matrices. In most and Garcia [48] considering the mass balance of solute into the
cases, this technology constitutes a cleaner alternative to the tradi- SC phase and linear equilibrium, Lucas et al. [42] proposed a
tional industrial techniques, which are based on hexane extraction. physicochemical model for the description of supercritical extrac-
Mass transfer processes take place when a solute is extracted from tion, adsorption and impregnation processes. The model considers
a fixed bed of a disintegrated vegetable material, with a solvent two parameters: convective flow transport between particles and
in supercritical or near-critical state. The pressurized solvent flows bulk phase (supercritical solvent) modeling with an external mass
continuously through the bed of particles. After extraction, pres- transfer coefficient (kf), and linear equilibrium on the matrix
sure is reduced and, consequently, solubility decreases and the surface (h). In supercritical fluid extraction and adsorption pro-
solute precipitates in a separator. The process is characterized by cesses parameter h considers two mass transfer steps (internal
an extraction curve, the plot of the cumulated extract versus the diffusion and adsorption/desorption equilibrium). In impregna-
extraction time or the amount of solvent required in the extractor. tion processes, this parameter represents the partition coefficient.
The different approaches to mathematical modeling of extraction These authors propose an analytical solution to model equations
from solids range from models based on experimental data [35] for simultaneously determining temporal and spatial profiles for
or based on heat and mass transfer analogy [36,37] to the for- solute concentration, both in the fluid and in the solid phase.
mulation of differential mass balances [38–42]. In the differential Zizovic et al. [49,50] introduced micro-scale mathematical mod-
mass balances formulation, time dependent concentration profiles els of essential oil supercritical fluid extraction process based on the
are obtained for fluid and solid phase through integration and the hypothesis that essential oil extraction processes should be depen-
extraction curve is later calculated from the solute concentration at dent on the type of secretory structure. Their objective was to verify
the extractor outlet stream. Sovova [41] presented a simple model the phenomena on the micro-scale and to optimize the supercritical
which divides the extraction curve into three main sections. The fluid extraction process according to the behavior of specific secre-
first part, where extraction rate is constant, is controlled by the tory structure during the extraction. Supercritical carbon dioxide
thermodynamic equilibrium between solute and solvent and exter- extraction confirmed previously reported phenomenon that parti-
nal mass transfer. The last section considers the internal diffusion cle size does not affect the evolution of extraction yield of some
controls. The midsection is the only nonlinear section and repre- Asteraceae and Apiaceae family species.
sents the period where both internal and external mass transfer
controls the extraction process. Martinez and Martinez [43] pro- 3.2. Countercurrent fluid–fluid processes
posed global optimization strategies for parameter estimation in
Sovova’s model. Lu et al. [44] adjusted parameters of Sovova’s model 3.2.1. Models based on equilibrium stages at steady state
for the extraction of borage seed oil by compressed CO2 , based on In a typical supercritical extraction/fractionation scheme, the
their experimental data. They also studied the effects of the most mixture to be fractionated is fed to an extraction column in coun-
important operating parameters (pressure, temperature, flow rate tercurrent with carbon dioxide at high pressure. The extract is
and bed length) on the rate of extraction. expanded and sent to a separator tank, where the condensed solute
Reverchon and De Marco [45] presented a comprehensive and is obtained. The recovered solvent is recycled to the extraction unit
critical overview of the analytical, processing and modeling aspects after compression or pumping. Cygnarowicz and Seider [51] pro-
on supercritical fluid extraction and fractionation. They covered posed a model for the dehydration of acetone with supercritical
around 250 applications that range from extraction from solids carbon dioxide. Both the extractor and the distillation column were
to liquid feed processing and antisolvent extraction, focusing on modeled as equilibrium stages and thermodynamic properties have
the selection of operating parameters, examples of application and been calculated with the Group Contribution Equation of State [3,4].
614 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618

Another application of supercritical fluids in countercurrent

processes is the selective entrainment in the fractionation of fish
oil fatty acid ethyl esters (FAEE). Numerous experimental studies
have been reported and, therefore, several authors have proposed
mathematical models of different complexity. Krukonis et al. [62]
proposed a two-column scheme and presented simplified simula-
tion results for the continuous recovery of eicosapentaenoic acid
(EPA) and docohexaenoic acid (DHA) esters from urea-adducted
menhaden oil ethyl esters. Espinosa et al. [29] extended their
rigorous simulation model [61], integrated to GC-EOS [8] to rep-
resent a fractionation scheme for a 13 component FAAE mixture,
consisting of three consecutive supercritical columns with reflux
for the production of highly concentrated streams of EPA and
DHA esters from both natural and urea-adducted fish oil FAEE.
More recently, Espinosa et al. [63] proposed a single column
scheme for the production of pharmaceutical grade concentrates
of eicosapentaenoic acid and docohexaenoic acid esters, within
an optimization environment. They compared ethane and carbon
dioxide as supercritical solvents, concluding that ethane shows a
better performance, with higher selectivity to EPA and DHA among
the esters with the same number of carbon atoms, together with
higher solubility of these esters in ethane.
Fig. 2. Hexane extraction with CO2 (R: solvent/feed weight ratio). Supercritical fluid fractionation by extraction with carbon diox-
ide is also an attractive alternative for the deterpenation of citrus
peel oils, as it combines low operating temperatures that prevent
The model was developed within program SEPSIM [52] and the products from thermal degradation, and a benign solvent. In cit-
nonlinear program has been solved with a Successive Quadratic rus deterpenation, the main issue is to produce desirable flavors
Programming algorithm [53]. for processed orange peel oil and to minimize undesirable fea-
Several authors [5,54,55] have proposed the use of light hydro- tures such as bitterness. Budich et al. [13] and Budich and Brunner
carbons at near critical conditions for the recovery of dehydrated [64] carried out a thorough study on the deterpenation of orange
oxychemicals from azeotropic or dilute aqueous mixtures. The peel oil with supercritical carbon dioxide, reporting phase equilib-
process is based on a cycle in which the aqueous mixture is rium data, countercurrent column experiments and flooding point
fed to an extractor and the solvent and the extracted product measurements. They also formulated a simplified model, based on
are separated in a solvent recovery column. This is the simplest vapor liquid equilibrium data, to perform stage calculations and
two-column (extraction + distillation) separation process, typical of a preliminary scale-up of the deterpenation process. Citrus peel
liquid or supercritical extraction. A main feature of this process is oil extraction optimization has been addressed in our group using
that the distillation column not only performs solvent recovery, different fractionation schemes and alternative supercritical fluids
but also completely removes water from the bottom product. In [11,12,65]. A discussion on optimization models and results is given
this way, dehydrated products can be obtained without additional in Section 4.
columns. The hydrocarbon solvent shows good selectivity for the Castier and co-workers have developed the simulation soft-
oxychemical recovery and also exhibits a water entrainment effect ware program Thermat, based on Mathematica® , and integrated
(water-solvent relative volatility greater than one in the solvent to Fortran routines for the calculation of thermodynamic prop-
recovery column). Moreover, no azeotrope formation between the erties. Process models were developed for the simulation of
solvent and the extracted product is required. A thorough study of isothermal extraction columns [66], dynamic simulations of flash
process alternatives for the recovery of ethanol from water using drums, parameter fitting interval arithmetic [67] and the simu-
propane as supercritical solvent was presented by Gros et al. [56] lation of simple and reactive distillation columns [68]. Nogueira
through the formulation of simulation models based on equilib- et al. [69] extended Thermath for the simulation of isothermal
rium stages for the countercurrent extractor and the distillation and non-isothermal supercritical extraction columns. As an exam-
column, as well as thermodynamic predictions with a Group Con- ple, they modeled the separation of ethyl acetate, isoamyl alcohol,
tribution with Association Equation of State [57]. The simulation ethanol and isoamyl acetate, which are present in an enzymatic
program included rigorous models for a high-pressure multistage transesterification reaction, with carbon dioxide as SC solvent.
extractor [58], low and high-pressure distillation columns [59], and The resulting algebraic equation system was solved with the
a multiphase flash [60]. Newton–Raphson method and phase equilibrium was modeled
Espinosa et al. [61] developed a rigorous simulation model based with the Soave–Redlich–Kwong (SRK) equation of state (EOS) with
on the above mentioned unit models [58–60] to study the use the Huron–Vidal mixing rule.
of CO2 for the removal of liposoluble pollutants or valuable sub- Several authors have addressed dynamic simulation of counter-
stances from natural oils. Both CO2 and propane were considered current supercritical extraction processes. Cygnarowicz and Seider
as SC solvents; the latter one with greater solvent power than CO2 [70] performed a dynamic analysis of ␤-carotene recovery from fer-
for organic solutes, but with the disadvantage of being flammable. mentation broths by SC CO2 , studying the influence of feed load
Fig. 2 shows concentration of oil in the extract as a function of pres- and vapor flow rate-induced disturbances on product separator
sure and solvent to feed ratio (R) required for achieving treated oil pressure. Sato et al. [71] reported the semi-batch deterpenation
specifications (5 ppm for hexane in raffinate). It can be seen that of orange peel oil with supercritical carbon dioxide; they studied
R is almost constant (2.7), when operating above 15 MPa. How- the effect of internal reflux induced by a temperature gradient in
ever, when pressure increases from 15 to 20 MPa there is a drastic the column, on process selectivity and product recovery, using the
increase in oil concentration in the extract. Peng–Robinson [16] equation of state to estimate the ternary phase
 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618 615

equilibria. These authors also reported laboratory-scale data for the Runge–Kutta 4–5 formula, implemented in Matlab [82]. At each
continuous countercurrent extraction of terpenes from orange peel integration step, an algebraic equation system was solved for mass
oil. Benvenuti et al. [72] modeled a semi-continuous single-stage and energy balances to the interphase at each column height and
process for deterpenation of natural lemon oil, where equilibrium physical properties calculation. The model was used to simulate
conditions were determined with the Peng–Robinson equation of the separation of a binary mixture of squalene and methyl oleate,
state [16], assuming that the vapor phase leaving the apparatus reported by Ruivo et al. [78], and the fractionation of fish oil fatty
was always in equilibrium with the liquid stationary phase inside acid ethyl esters [83]. In this case, long chain fatty acid ethyl
the vessel. These authors implemented the differential algebraic esters were lumped into two pseudo components: one represent-
system representing the SCF process in gPROMS [73] and their pre- ing low molecular weight components (LMC, with carbon numbers
dictions are in good agreement with experimental data they have from C14 to C18) and a high molecular weight component (HMC,
obtained in their laboratory. Additionally, they report simulation C20–C22). In both cases, carbon dioxide was used as supercritical
results for a steady state countercurrent process, simulated with solvent.
the commercial simulator Aspen Plus. In a recent paper, Varona et al. [84] have addressed modeling and
experimental determinations for the separation of linalool and lina-
3.2.2. Models based on mass transfer approaches lyl acetate, main components of lavandin oil, using the previously
Several authors have proposed mass transfer models for the sim- described model for packed extraction column and flash condenser.
ulation of extraction columns at high pressure. De Haan and de
Graaw [74] developed a mass transfer model for the fractionation of 4. Process optimization approaches
hydrocarbons with supercritical carbon dioxide, in which physical
properties were represented by correlations. They reported numer- As high-pressure processes render important energy consump-
ical and experimental results for the separation of hexadecane and tion, they must be highly integrated to be economically feasible.
2-methylnaphthalene with carbon dioxide at high pressure and Special attention must be devoted to solvent cycle scheme [85].
found good agreement between measured and calculated height There are different sequences of operations to change the solvent
equivalent of a theoretical plate (HETP). The model was based on state conditions and composition; they depend on operating condi-
a previous one proposed by Bravo et al. [75] for structured pack- tions of the main process, the nature of the solvent and the scale of
ing distillation columns. These authors, Rocha et al. [76], extended the process unit. In supercritical fluid processes, the solvent can be
the model for packed distillation columns. Ruivo et al. [77] com- driven either by a compressor or by a pump and it can be recycled
pared the performance of several available mass transfer models in either in supercritical or in liquid state. If there is a pump cycle,
packed columns with experimental results obtained at their labo- lower fixed costs are associated to pumps as compared to com-
ratory in a countercurrent supercritical extraction packed column pressors and energy consumption is lower than the compressor
of 2 m height and 2.4 cm internal diameter for the fractionation of process for pressures higher than 300 bar. As a disadvantage, the
a squalene/methyl oleate mixture by supercritical carbon dioxide. pump cycle requires several heat exchangers and condensers and
They concluded that the model proposed by Rocha et al. [76] was additional heat energy at low extraction pressures. The compres-
appropriate for modeling high-pressure extraction systems with sor cycle needs only one heat exchanger and it shows low heat
corrugated gauze packing columns. energy consumption but it has higher fixed costs for the com-
More recently, Ruivo et al. [78] formulated a dynamic model pressor, higher electrical energy consumption and lower energy
for the simulation of a SCF extraction column with structured consumption compared to the pump cycle for pressures lower
packing for the fractionation of a binary mixture of squalene and than 300 bar. Solvent cycle determination constitutes an important
methyl oleate, with SC carbon dioxide. The model comprises a design problem in supercritical fluid processes.
set of partial differential equations (PDE) for the differential mass Cygnarowicz and Seider [51] formulated a nonlinear program-
balances on the packed column, algebraic equations for mass trans- ming model to minimize either the utility or the annualized cost
fer, hydrodynamics of the two-phase flow through the packings for the dehydration of acetone with supercritical carbon dioxide.
and thermodynamic equilibrium. They transformed the PDE into Based on the rigorous simulation models proposed by Gros et al.
an ordinary differential equations system by performing a spatial [56] and within a mixed integer nonlinear programming frame-
discretization with finite volumes. The ternary phase equilibrium work, Diaz et al. [86] formulated models for the optimization of
was calculated with a simplified empirical correlation, as function the dehydration of oxychemicals with supercritical light hydrocar-
of pressure, temperature and squalene composition in the liquid bons, associating binary variables to potential process units. The
phase. Column temperature was assumed constant. The model was objective function was process energy consumption minimization.
implemented in gPROMS [73] and the authors report a good agree- They studied several aqueous mixtures with oxychemicals such as
ment between measured and predicted composition profiles of the ethanol, 2-propanol and acetone at different concentrations and
outlet streams. Fernandes et al. [79] extended the model to account the use of several dual effect solvents, such as propane and isobu-
for non-isothermal conditions in the packed column. tane.
Martin and Coccero [80] developed a rigorous model of a SCF Diaz et al. [10] formulated a mixed integer nonlinear pro-
fractionation column, which includes differential mass and energy gramming model for the simultaneous determination of operating
balances in the column, and coupled mass and heat transfer in conditions and process and solvent cycle scheme, including invest-
the interface between the two fluid phases. Mass transfer is rep- ment and operating costs for the entire process in the deterpenation
resented by means of the generalized Maxwell–Stefan relations of lemon peel oil with supercritical carbon dioxide. The objective
for multicomponent systems and thermodynamic properties are was the maximization of net profit, while fulfilling product recovery
calculated with the Stryjek–Vera Peng–Robinson equation of state and purity specifications. Discrete decisions, associated to binary
[81]. Transport properties (diffusivity, viscosity and interfacial ten- variables (y) were related to the potential existence of units in the
sion) were calculated with correlations. The entire model for the solvent recovery system (mainly represented by a compressor or
extractor was represented with a differential algebraic equation a pump cycle). Fig. 3 shows a deterpenation process superstruc-
(DAE) system that was solved in an iterative way to include the ture. Rigorous simulation models [58–60] were integrated to the
calculation of the reflux flash. To solve the DAE, these authors extended GC-EOS [9], while heating, cooling and compression loads
used an ordinary differential equation solver based on an explicit in the solvent recovery cycle were calculated with rigorous models
616 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618

carbon dioxide [29] and ethane [63]. More recently, Raeissi et al.
[22] proposed nonlinear models for the minimization of solvent
flowrate in a deterpenation plant, using the Soave–Redlich–Kwong
equation of state [18]. While achieving a fixed level of oil fold-
ing, they studied the performance of ethane as supercritical
solvent.
Only few authors have addressed process optimization based
on rigorous mass transfer models. Fernandes et al. [79] devel-
oped a dynamic optimization model for an extraction column,
also including energy and momentum balances in the packed
bed. They performed process optimization to maximize squalene
recovery from a binary mixture composed of squalene and methyl
oleate, with supercritical carbon dioxide within program gPROMS
[73].
Alternative approaches include process optimization based on
experimental data. L’etisse et al. [90] developed a quadratic correla-
tion for the maximization of yield in EPA and DHA production from
sardine oil, using carbon dioxide as SC solvent. They performed
Fig. 3. Supercritical deterpenation scheme with solvent cycle superstructure. optimal design of experiments in their laboratory with the Doehlert
uniform shell design [91], with five test points evenly distributed
over the experimental domain. Independent variables were extrac-
based on SRK [18] predictions (for a single component, carbon diox- tion temperature, solvent flow rate and operating time.
ide). The mixed integer nonlinear programming model was solved
with the Outer Approximations algorithm [87], in which nonlinear 5. Conclusions
problems were solved with a Successive Quadratic Programming
algorithm [88] and mixed integer linear problems were solved with The great advances in the last two decades in phase equilibrium
LINDO [89] within a Fortran 90 environment. Simulation results modeling, rigorous simulation and optimization of supercritical
were also checked against pilot plant data. processing of natural products is opening the way to computer
As the main alternatives for the solvent recovery system can aided optimum design of supercritical plants for the extraction
be lumped into a pump and a compressor cycle, Diaz et al. [11,12] and fractionation of natural products. Phase equilibrium modeling
performed the optimal design of process and solvent cycle for the of supercritical processes has been based in several approaches.
deterpenation of orange peel oil by formulating and solving two Extensive use has been made of cubic equations of state in the
nonlinear programming problems representing the process and extraction or fractionation of low to medium molecular weight
solvent cycle for those two cycles, respectively. To perform column solutes. When the mixture non-ideality is high, cubic equations
sizing, they used correlations for the estimation of the minimum of state have been combined with local composition models like
inner diameter derived by Budich and Brunner [64], based on their UNIFAC. However, in highly asymmetric systems like propane or
experimental flooding results for the extraction of orange peel oil carbon dioxide with vegetable oils, the use of a more realistic repul-
with carbon dioxide over a wide range of pressure and temperature sive term in the equation of state has been introduced to describe
[64]. A comparison between optimal profit and costs was per- liquid–liquid and liquid–supercritical solvent phase equilibria. This
formed for the two alternatives, as well as a scale up of the process, approach has been used in the GC-EOS and extended for associat-
as it is shown in Fig. 4 for a compressor cycle process. Nonlinear pro- ing systems in the GCA-EOS. When modeling phase equilibrium
gramming models based on rigorous models were also developed in systems involving natural products, the use of group contri-
for fish oil fatty acid ethyl esters fractionation with supercritical bution methods has been a very efficient procedure to deal with
multicomponent natural mixtures that can be described with a
reduced set of groups. The use of specific correlations, Chrastil’s
equation for example, that take into account the effect of tem-
perature and solvent density on solute solubility has found many
applications in modeling the extraction from solid matrices. In the
extraction of solutes from solids the use of semi-empirical mod-
els like Sovova’s has been found in many applications. In recent
work, rigorous modeling of mass and heat transfer in packed beds
has shown promising results for the design and simulation of
solid–fluid supercritical extraction, based on first principles. Stage
by stage rigorous simulation of countercurrent processes com-
bined with equations of state models has been successful in the
synthesis, simulation and optimization of supercritical extraction
and fractionation processes. Recently a rigorous approach, based
on differential mass, heat and momentum balances equations,
more suitable for packed columns, has been applied. Associated
to advances in simulation and optimization software, increas-
ingly detailed models can now be formulated to guide the
development of industrial applications of supercritical fluid extrac-
tion and fractionation processes, especially in the field of high
Fig. 4. Plant capacity optimization results for 99.9 wt% limonene purity in top prod- added value compounds, such as pharmaceutical and nutraceutical
uct: compressor mode costs and profit for 5-fold raffinate product. products.
 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618 617

Acknowledgement [31] C. Borch-Jensen, A. Staby, J. Mollerup, Phase equilibria of urea-fractionated fish

oil fatty acid ethyl esters and supercritical carbon dioxide, Ind. Eng. Chem. Res.
33 (1994) 1574.
The authors gratefully acknowledge financial support from [32] J. Chrastil, Solubility of solids and liquids in supercritical gases, J. Phys. Chem.
CONICET, ANPCYT, Universidad Nacional del Sur, Argentina. 86 (1982) 3016.
[33] Y. Adachi, B.C.-Y Lu, Supercritical fluid extraction with carbon dioxide and ethy-
lene, Fluid Phase Equilib. 14 (1983) 147.
References [34] J.M. del Valle, J.M. Aguilera, An improved equation for predicting the solubility
of vegetable oils in supercritical CO2 , Ind. Eng. Chem. Res. 27 (1988) 1551.
[1] M. Christov, R. Dohrn, High-pressure fluid phase equilibria experimental meth- [35] K. Nguyen, P. Barton, J. Spencer, Supercritical carbon dioxide of vanilla, J. Super-
ods and systems investigated (1944–1999), Fluid Phase Equilib. 202 (2002) crit. Fluids 4 (1991) 40.
153. [36] E. Reverchon, G. Donsı, L. Sesti Osseo, Modeling of supercritical fluid extraction
[2] D. Richon, New experimental developments for phase equilibrium measure- from herbaceous matrices, Ind. Eng. Chem. Res. 32 (1993) 2721.
ments, Fluid Phase Equilib. 116 (1996) 421. [37] E. Reverchon, Mathematical modelling of supercritical extraction of sage oil,
[3] S. Skjold-Jørgensen, Gas solubility calculations. II. Application of a new group- AIChE J. 42 (1996) 1765.
contribution equation of state, Fluid Phase Equilib. 16 (1984) 317. [38] H. Sovova, R. Komers, J. Kucera, J. Jez, Supercritical carbon dioxide extraction of
[4] S. Skjold-Jørgensen, Group contribution equation of state (GC-EOS): a predictive caraway essential oil, Chem. Eng. Sci. 49 (1994) 2499.
method for phase equilibrium computations over wide ranges of temperatures [39] B.C. Roy, M. Goto, T. Hirose, Extraction of ginger oil with supercritical carbon
and pressures up to 30 MPa, Ind. Eng. Chem. Res. 27 (1988) 110. dioxide: experiments and modeling, Ind. Eng. Chem. Res. 35 (1996) 607.
[5] E.A. Brignole, P.M. Andersen, Aa. Fredenslund, Supercritical fluid extraction of [40] E. Reverchon, C. Marrone, Modeling and simulation of the supercritical CO2
alcohols from water, Ind. Eng. Chem. Res. 26 (1987) 254. extraction of vegetable oils, J. Supercrit. Fluids 19 (2001) 161.
[6] F. Temelli, J. O’Connell, C. Chen, R. Braddock, Thermodynamic analysis of super- [41] H. Sovova, Mathematical model for supercritical fluid extraction of natural
critical carbon dioxide extraction of terpenes from cold-pressed orange oil, Ind. products and extraction curve evaluation, J. Supercrit. Fluids 33 (2005) 35.
Eng. Chem. Res. 29 (1990) 618. [42] S. Lucas, M.P. Calvo, J. Garcia-Serna, C. Palencia, M.J. Cocero, Two-parameter
[7] H.P. Gros, S. Bottini, E.A. Brignole, High pressure phase equilibrium modeling model for mass transfer processes between solid matrixes and supercritical
of mixtures containing associating compounds and gases, Fluid Phase Equilib. fluids: analytical solution, J. Supercrit. Fluids 41 (2007) 257.
139 (1997) 75. [43] J. Martinez, J.M. Martinez, Fitting the Sovova’s supercritical fluid extraction
[8] S. Bottini, T. Fornari, E.A. Brignole, Phase equilibrium modeling of triglycerides model by means of a global optimization tool, Comput. Chem. Eng. 32 (2008)
with near critical solvents, Fluid Phase Equilib. 158 (1999) 211. 1735.
[9] S. Espinosa, A. Bermúdez, G. Foco, T. Fornari, Revision and extension of the group [44] T. Lu, F. Gaspar, R. Marriott, S. Mellor, C. Watkinson, B. Al-Duri, J. Seville, R.
contribution equation of state to new solvent groups and higher molecular Santos, Extraction of borage seed oil by compressed CO2 : effect of extraction
weight alkanes, Fluid Phase Equilib. 172 (2000) 129. parameters and modelling, J. Supercrit. Fluids 41 (2007) 68.
[10] M.S. Diaz, S. Espinosa, E.A. Brignole, Optimal solvent cycle design in supercritical [45] E. Reverchon, I. De Marco, Supercritical fluid extraction and fractionation of
fluid processes, Latin Am. Appl. Res. 33 (2003) 161. natural matter, J. Supercrit. Fluids 38 (2006) 146.
[11] M.S. Diaz, S. Espinosa, E.A. Brignole, Citrus peel oil deterpenation with super- [46] M.A.A. Meireles, Supercritical extraction from solid: process design data
critical fluids optimal process and solvent cycle design, J. Supercrit. Fluids 35 (2001–2003), Curr. Opin. Solid State Mater. Sci. 7 (2003) 321.
(2005) 49. [47] J.M. Del Valle, J.C. De La Fuente, Supercritical CO extraction of oilseeds: review
[12] M.S. Diaz, S. Espinosa, E.A. Brignole, Process optimization for the concentration of kinetic and equilibrium models, Crit. Rev. Food Sci. Nutr. 46 (2006) 131.
of orange peel oil, Latin Am. Appl. Res. 35 (2005) 321. [48] M.J. Cocero, S. Garcia, Mathematical model of supercritical extraction applied to
[13] M. Budich, S. Heilig, T. Wesse, V. Leibküchler, G. Brunner, Countercurrent deter- oil seed extraction by CO2 + saturated alcohol. I. Desorption model, J. Supercrit.
penation of citrus oils with supercritical CO2 , J. Supercrit. Fluids 14 (1999) Fluids 20 (2001) 229.
105. [49] I. Zizovic, M. Stamenic, A. Orlovic, D. Skala, Supercritical carbon dioxide essen-
[14] C. Dariva, G.R. Ribeiro, J.V. Oliveira, Equilibrio de Fases a Altas Pressões para o tial oil extraction of Lamiaceae family species-mathematical modeling on the
Sistema CO2 —Óleo de Laranja, in: 5th International Conference EQUIFASE99, micro-scale and process optimization, Chem. Eng. Sci. 60 (2005) 6747.
vol. 1, Vigo-España, 1999, p. 139. [50] I. Zizovic, M. Stamenic, A. Orlovic, D. Skala, Supercritical carbon dioxide extrac-
[15] M. Sato, M. Goto, T. Hirose, Fractional extraction with supercritical carbon diox- tion of essential oils from plants with secretory ducts: mathematical modelling
ide for the removal of terpenes from citrus oil, Ind. Eng. Chem. Res. 34 (1995) on the micro-scale, J. Supercrit. Fluids 39 (2007) 338.
3941. [51] M.L. Cygnarowicz, W.D. Seider, Effect of retrograde solubility on the design
[16] D.Y. Peng, D.B. Robinson, A new two constant equation of state, Ind. Eng. Chem. optimization of supercritical extraction process, Ind. Eng. Chem. Res. 28 (1989)
Fundam. 15 (1976) 59. 1497.
[17] G. Soave, A simple model for the supercritical extraction of solids, J. Supercrit. [52] P.M. Andersen, Aa. Fredenslund, Process simulation with advanced thermody-
Fluids 19 (2000) 19. namic models, Proc. Chem. Eng. Fund., XVIII Cong., Sicily, 1987.
[18] G. Soave, Equilibrium constants from a modified Redlich–Kwong equation of [53] M.J.D. Powell, A Fast Algorithm for Nonlinearly Constrained Optimization Cal-
state, Chem. Eng. Sci. 27 (1972) 1197. culations, AERE, Harwell, England, 1977.
[19] S. Raeissi, C.J. Peters, Simulation of double retrograde vaporization using [54] M. Zabaloy, G. Mabe, S. Bottini, E.A. Brignole, The application of high water
the Peng–Robinson equation of state, J. Chem. Thermodyn. 35 (2003) volatilities over some liquified near-critical solvents as a means of dehydrating
573. oxychemicals, J. Supercrit. Fluids 5 (1992) 186.
[20] S. Espinosa, S. Raeissi, E.A. Brignole, C.J. Peters, Prediction of double retrograde [55] H. Horizoe, T. Tanimoto, I. Yamamoto, Y. Kano, Phase equilibrium study for
vaporization: transitions in binary mixtures of near critical fluids with compo- the separation of ethanol–water solution using subcritical and supercritical
nents of homologous series, J. Supercrit. Fluids 32 (2004) 63. hydrocarbon solvent extraction, Fluid Phase Equilib. 84 (1993) 297.
[21] S. Raeissi, C.J. Peters, Application of double retrograde vaporization as an opti- [56] H. Gros, S. Diaz, E.A. Brignole, Process synthesis and optimization of near critical
mizing factor in supercritical fluid separations, J. Supercrit. Fluids 33 (2005) separations of aqueous azeotropic mixtures, J. Supercrit. Fluids 12 (1998) 69.
115. [57] H.P. Gros, S. Bottini, E.A. Brignole, A group contribution equation of state for
[22] S. Raeissi, S. Diaz, S. Espinosa, C.J. Peters, E.A. Brignole, Ethane as an alternative associating mixtures, Fluid Phase Equilib. 116 (1996) 537.
solvent for supercritical extraction of orange peel oils, J. Supercrit. Fluids 45 [58] E. Kehat, B. Ghitis, Simulation of an extraction column, Comput. Chem. Eng. 5
(2008) 306. (1981) 171.
[23] L. Coniglio, K. Knudsen, R. Gani, Model prediction of supercritical fluid–liquid [59] L. Naphtali, D. Sandholm, Multicomponent separation calculations by lineariza-
equilibria for carbon dioxide and fish oil related compounds, Ind. Eng. Chem. tion, AIChE J. 17 (1971) 148.
Res. 34 (1995) 2473. [60] M. Michelsen, The isothermal flash problem. Part II. Phase split calculations,
[24] M.L. Michelsen, A modified Huron–Vidal mixing rule for cubic equations of Fluid Phase Equilib. 9 (1982) 21.
state, Fluid Phase Equilib. 60 (1990) 213. [61] S. Espinosa, S.B. Bottini, E.A. Brignole, Process analysis and phase equilibria for
[25] S. Dahl, M.L. Michelsen, High-pressure vapor–liquid equilibrium with a UNIFAC- the removal of chemicals from fatty oils using near-critical solvents, Ind. Eng.
based equation of state, AIChE J. 68 (1990) 441. Chem. Res. 39 (2000) 3024.
[26] Aa. Fredenslund, J. Gmehling, P. Rasmussen, Vapor–Liquid Equilibria Using UNI- [62] V.J. Krukonis, J. Vivian, C. Bambara, W. Nilsson, R. Martin, Concentration of
FAC, Elsevier, Amsterdam, 1977. eicosapentaenoic acid by supercritical fluid extraction: A design study of a con-
[27] B.L. Larsen, P. Rasmussen, Aa. Fredenslund, A modified UNIFAC group- tinuous production process, in: G. Flick, R. Martin (Eds.), Advances in Seafood
contribution model for prediction of phase equilibria and heats of mixing, Ind. Biochemistry, Composition and Quality. Papers from the ACS Annual Meeting,
Eng. Chem. Res. 26 (1987) 2274. vol. 169, New Orleans, Louisiana, 1992.
[28] V. Riha, G. Brunner, Phase equilibrium of fish oil ethyl esters with supercritical [63] S. Espinosa, M.S. Diaz, E.A. Brignole, Food additives obtained by supercritical
carbon dioxide, J. Supercrit. Fluids 15 (1999) 33. extraction from natural sources, J. Supercrit. Fluids 45 (2008) 213.
[29] S. Espinosa, M.S. Diaz, E. Brignole, Thermodynamic modeling and process opti- [64] M. Budich, G. Brunner, Vapor–liquid equilibrium data and flooding point mea-
mization of supercritical fluid fractionation of fish oil fatty acid ethyl esters, surements of the mixture carbon-dioxide + orange peel oil, Fluid Phase Equilib.
Ind. Eng. Chem. Res. 41 (2002) 1516. 158 (1999) 759.
[30] A. Staby, T. Forskov, J. Mollerup, Phase equilibria of fish oil fatty acid ethyl esters [65] S. Diaz, S. Espinosa, E.A. Brignole, Modeling simulation tools for supercritical
and sub- and supercritical CO2 , Fluid Phase Equilib. 87 (1993) 309. fluid processes, Comput. Aided Chem. Eng. 8 (2000) 319.
618 M.S. Diaz, E.A. Brignole / J. of Supercritical Fluids 47 (2009) 611–618

[66] M.F. Alfradique, R.O. Esposito, M. Castier, Automatic generation of procedures [78] R. Ruivo, A. Paiva, J.P.B. Mota, P. Simoes, Dynamic model of a countercurrent
for the simulation of multistage separators using computer algebra, Chem. Eng. packed column operating at high pressure conditions, J. Supercrit. Fluids 32
Commun. 189 (2002) 658. (2004) 183.
[67] A. Dominguez, J. Tojo, M. Castier, Automatic implementation of thermodynamic [79] J. Fernandes, R. Ruivo, J.P.B. Mota, P. Simoes, Non-isothermal dynamic model of
models for reliable parameter estimation using computer algebra, Comput. a supercritical fluid extraction packed column, J. Supercrit. Fluids 41 (2007) 20.
Chem. Eng. 26 (2002) 1473. [80] A. Martin, M.J. Cocero, Mathematical modeling of the fractionation of liquids
[68] M.F. Alfradique, Implementacao Automatica de Procedimentos para Simulacao with supercritical CO2 in a countercurrent packed column, J. Supercrit. Fluids
de Colunas de Destilacao Convencionale Reativa Usando Computacao Algebrica, 39 (2007) 304.
M.Sc. thesis, Escola de Quımica, UFRJ, Brazil, 2003. [81] R. Stryjek, J.H. Vera, An improved Peng–Robinson equation of state for pure
[69] B.M.M. Nogueira, M.F. Alfradique, M. Castier, Modeling and simulation of super- components and for mixtures, Can. J. Chem. Eng. 64 (1986) 820.
critical extraction columns using computer algebra, J. Supercrit. Fluids 34 [82] Matlab. The Mathworks, 2004.
(2005) 203. [83] V. Riha, G. Brunner, Separation of fish oil ethyl esters with supercritical carbon
[70] M.L. Cygnarowicz, W.D. Seider, Design control of a process to extract ␤-carotene dioxide, J. Supercrit. Fluids 17 (2000) 55.
with supercritical carbon dioxide, Biotechnol. Prog. 6 (1990) 82. [84] S. Varona, A. Martin, M.J. Cocero, T. Gamse, Supercritical carbon dioxide frac-
[71] M. Sato, M. Goto, T. Hirose, Supercritical fluid extraction on semibatch mode tionation of Lavandin essential oil: experiments and modeling, J. Supercrit.
for the removal of terpene in citrus oil, Ind. Eng. Chem. Res. 35 (1996) 1906. Fluids 45 (2008) 181.
[72] F. Benvenuti, F. Gironi, L. Lamberti, Supercritical deterpenation of lemon essen- [85] G. Brunner, Gas extraction, in: An Introduction to Fundamentals of Supercritical
tial oil, experimental data and simulation of the semicontinuous extraction Fluids and the Applications to Separation Processes, Springer, Berlin, 1994.
process, J. Supercrit. Fluids 20 (2001) 29. [86] M.S. Diaz, H. Gros, E.A. Brignole, Thermodynamic Modeling Synthesis and Opti-
[73] Process Systems Enterprise, gPROMS v2. 0 Introductory User Guide, Process mization of Extraction-Dehydration Processes, Comput. Chem. Eng. 24 (2000)
Systems Enterprise Ltd., London, 2001. 2069.
[74] A.B. De Haan, J. de Graaw, Mass transfer in supercritical extraction columns [87] G. Kocis, I.E. Grossmann, Relaxation strategy for the structural optimization of
with structured packings for hydrocarbon processing, Ind. Eng. Chem. Res. 30 process flow sheets, Ind. Eng. Chem. Res. 26 (1987) 1869.
(1991) 2463. [88] L.T. Biegler, J.E. Cuthrell, Improved infeasible path optimization for sequen-
[75] J.L. Bravo, J.A. Rocha, J.R. Fair, Mass transfer in gauze packings, Hydrocarbon tial modular simulators. II. The optimization algorithm, Comput. Chem. Eng.
Process. 64 (1985) 91. 9 (1985) 257.
[76] J.A. Rocha, J.L. Bravo, J.R. Fair, Distillation columns containing structured pack- [89] L. Schrage, Linear Integer and Quadratic Programming With LINDO, The Scien-
ings: a comprehensive model for their performance 2. Mass transfer model, tific Press, Palo Alto, 1987.
Ind. Eng. Chem. Res. 35 (1996) 1660. [90] M. Lı̌etisse, M. Rozieres, A. Hiol, M. Sergent, L. Comeau, Enrichment of EPA and
[77] R. Ruivo, M.J. Cebola, P.C. Simoes, M. Nunes da Ponte, Fractionation of edible oil DHA from sardine by supercritical fluid extraction without organic modifier. I.
model mixtures by supercritical carbon dioxide in a packed column 2. A mass Optimization of extraction conditions., J. Supercrit. Fluids 38 (2006) 27.
transfer study, Ind. Eng. Chem. Res. 41 (2002) 2305. [91] D.H. Doehlert, Uniform shell designs, Appl. Stat. 19 (1970) 231.