You are on page 1of 9

Innovative Food Science and Emerging Technologies 38 (2016) 32–40

Contents lists available at ScienceDirect

Innovative Food Science and Emerging Technologies

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

Fitting of broken and intact cell model to supercritical fluid extraction


(SFE) of sunflower oil
Amit Rai a,⁎, Bikash Mohanty b, Ravindra Bhargava b
a
Department of Chemical Engineering, NIT Hamirpur, Hamirpur, Himachal Pradesh 177001, India
b
Department of Chemical Engineering, IIT Roorkee, Roorkee, Uttarakhand 247667, India

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

Article history: The present work is focused on the supercritical extraction and kinetic study of sunflower oil. The effect of oper-
Received 22 March 2016 ating parameters, namely pressure (20–40 MPa), temperature (333–373 K), solvent mass flow (0.83 × 10−4 to
Received in revised form 26 August 2016 2.50 × 10−4 kg/s), particle size (0.50–1.00 mm) and ethanol as co-solvent (0–10%) have been studied by a set
Accepted 30 August 2016
of experiments. The outcomes of work show, extraction yield increases with increasing pressure, solvent mass
Available online 31 August 2016
flow and co-solvent. The retrograde phenomenon is also observed between 333 and 353 K. The experimental
Keywords:
data have been used to simulate the broken and intact cell model and fitting parameters are optimized by the
Sunflower oil global optimization method. The model successfully fitted the experimental data at high pressure with lower par-
Broken and intact cell ticle size.
Model optimization Industrial relevance: The sunflower seeds are a major source of good quality, high oleic acid, vegetable oil and Vi-
Kinetics tamin E. Hence, largely used in beauty care products, production of vegetable oil based fuel oil and lubricants.
SFE There are also many other applications of rich–oleic acid oils such as food, cosmetics, pharmaceuticals, biodiesel
production, and also as a feedstock for the oleo-chemical industry. In recent years, due to rigid environmental
regulation and health concern, high pressure extraction of the vegetable oil with supercritical carbon dioxide
has received considerable attention as a promising alternative to conventional solvent extraction and mechanical
process. Hence, the present work is focused on extraction of seed oil by supercritical CO2.
Moreover, for industrial application of supercritical fluid extraction, it is very important to study the kinetic
model that gives large scale extraction curves with minimum error by using model fitting parameters with
small scale experiment. There are large numbers of models based on different approach suggested by various re-
searchers to fit experimental data of supercritical fluid extraction. Amongst them, the broken and intact cell
model proposed by Sovova's (1994) is widely used to fit the experimental data of vegetable oil extraction. The
broken and intact cell model has three fitting parameters and many authors have applied different optimization
techniques to optimize these fitting parameters. Therefore, in the present work the optimized kinetic parameters
of Sovová's model for sunflower oil extraction are obtained by applying the Box complex optimization method to
study the influence of pressure, temperature, particle size, co-solvent and solvent flow rate of extraction yield.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction However, these traditional methods have some drawbacks, like long ex-
traction time, large amounts of organic solvent and high temperature is
The sunflower seeds (Heliantus annuus L.) are a major source of good required due to which loss or degradation of active compounds occurs
quality, high oleic acid vegetable oil which is largely used for human (Mukhopadhyay, 2000). Hence, the SFE is an attractive alternative tech-
consumption in India as well as in many Western countries due high nique has several advantages, including the use of non-flammable, non-
ratio of polyunsaturated to saturated fatty acids (Salgin, Doker, & toxic, non-explosive, cost efficient, non-polar green solvent, which gives
Calimli, 2006; Shahidi, 2005). Also, sunflower oil is a major source of Vi- faster and more selective separation with low degradation of chemical
tamin E and hence, largely used in beauty care products (Nimet et al., compounds (Mukhopadhyay, 2000).
2011). Moreover, it is also used for the production of vegetable oil Moreover, for industrial application of SFE, it is very important to
based fuel oil and lubricants (Tomic, Savin, Micic, Simikic, & Furman, study the kinetic model that gives large scale extraction curves with
2014). Extraction of vegetable oils from seed material is mostly per- minimum error by using model fitting parameters with small scale ex-
formed by some traditional methods such as by organic solvent. periment (Casas, Mantell, Rodriguez, Lopez, & Ossa, 2009). There are
large numbers of models based on different approach suggested by var-
⁎ Corresponding author. ious researchers to fit experimental data of SFE. Amongst them, the bro-
E-mail address: mit.rai123456@gmail.com (A. Rai). ken and intact cell model proposed by Sovová's (1994) is widely used to

http://dx.doi.org/10.1016/j.ifset.2016.08.019
1466-8564/© 2016 Elsevier Ltd. All rights reserved.
A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40 33

fit the experimental data of vegetable oil extraction. The broken and in- of easily accessible solute (P) and the inaccessible solute (K) inside the
tact cell model has three fitting parameters and many authors have ap- solid particle. Thus, the overall extraction process includes three differ-
plied different optimization techniques to optimize these fitting ent stages. First is the constant extraction rate (CER): governed by con-
parameters. Mira, Blasco, and Subirats (1996) fitting parameters by vection and extraction rate is proportional to the solubility, second is the
Excel solver, Louli, Folas, Voutsas, and Magoulas (2004) used fourth transition extraction step or falling extraction rate (FER): extraction rate
order Runga-Kutta method, Moura, Carvalho, Stefanini, Ming, and is governed by diffusion and convection, third is the slow extraction pe-
Meireles (2005) steepest descent and Newton method, Pederssetti et riod: governed by only diffusion. The analytical solution of broken and
al. (2011) used Nelder-Mead simplex method to minimize an error ob- intact cell model is described by the following equations, where the ex-
jective function but all these methods may not give optimum value and tractions curve as a function of time.
can be trapped at local minima. Similarly, Martinez and Martinez (2008) 8
used the lexicographical grid method to obtain the global optimum so- >
> Q CO y ½1− expð−ZÞt if 0≤ t ≤tCER
< Q y ½t−t 2 rexpðz −ZÞ if tCER≤ t≤tFER
lution, but this method required good programming skills. Therefore, yðt Þ ¼ CO2 r CER W  
>
> y expðWx0 Þ WQ CO2 x
: mSI x0 − r ln 1 þ −1 exp ðtCER −tÞ k if t NtFER
the objectives of the present work are to obtain the optimized kinetic W tr mSI x0
parameters of Sovová's model for sunflower oil extraction using super- ð3Þ
critical carbon dioxide by applying the Box complex optimization meth-
od to study the influence of pressure, temperature, particle size, co- where x0, xp and xk represent the total oil, easily accessible and inacces-
solvent and solvent mass flow of extraction yield. sible oil present in the solid with respect to non-extractable solid and
express by x0 = xp + xk, QCO2 is solvent mass flow, mSI is mass of non-ex-
2. Material and methods tractable solid. In the second extraction period, the depletion of easily
accessible oil in the raw material is given by a dimensionless variable
The sunflower seed had been purchased from local markets. In order zw and given by the Eq. (4).
to reduce the water content the seed have been unsealed and dried  
 
under daylight. All the seed has been crushed in a domestic kitchen Zyr x0 exp WQ CO2 ðt−t CER Þ=mSI −xk
zW ¼ ln ð4Þ
mixer grinder and sample based on three different sizes of particle has Wx0 x0 −xk
been prepared by sieved analysis as discussed in our previous paper of
sunflower and watermelon seed oil extraction (Rai, Mohanty, & Here tCER is the end of the first extraction period and the start of the
Bhargava, 2016; Rai, Mohanty, & Bhargava, 2015). The supercritical ex- second extraction period. Also tFER represent the end of the transition
traction grade CO2 is supplied by Sigma gases, India in a deep tube cyl- period with zW = 0 at the start and zW = Z of the transition period. The
inder. Analytical grade ethanol is procured from Merck Ltd. value of tFER and tCER is given by Eqs. (5) and (6) respectively.
(Darmstadt, Germany). A Soxhlet extraction with n-hexane has been
performed to determine total hexane-extractable sunflower oil content ðx0 −xk ÞmSI
t CER ¼ ð5Þ
of unsieved crushed seed as discussed in our previous paper (Rai et al., yr ZQ CO2
2015; Rai et al., 2016) and 54.87 ± 0.22% oil has been obtained. The
 
value of hexane-extractable oil is referred as total oil content in the mSI x þ ðx0 −xk Þ expðWx0 =yr Þ
t FER ¼ t CER þ ln k ð6Þ
present work and calculated by the Eq. (1). WQ CO2 x0

Oil Extracted The parameters Z and W are directly proportional to fluid and solid
Yield ð%Þ¼ 100 ð1Þ
Total Sample Taken phase mass transfer coefficient respectively, and given by Eqs. (7) and
(8)
Similarly, moisture and ash content of sunflower seed are obtained
as suggested by standard method 926.12 (AOAC, 1995) and ISO 749 k f a0 ρ mSI kYa ρCO2 FmSI
Z¼ ¼ ¼ ð7Þ
(ISO, 1977) and found approximately 5.31 ± 0.73% and 9.61 ± 0.59% re- Q CO2 Q CO2 ð1−ε Þρs Q CO2
ð1−ε Þρs
spectively. Moreover, the porosity, density and bulk density of the par- mSI
ticle bed are obtained by the apparatus supplied by Macro scientific
works, New Delhi by the following procedure. 50 g dry seed has been ks a0 mSI kXa SmSI Zy
W¼ ¼ ¼ ¼k r ð8Þ
taken into a graduated cylinder and set the time around 3 min to take Q CO2 Q CO2 ð1−ε Þ Q CO2 xk
ð1−ε Þ
100 stoke as per the standard requirement. After 100 St ± 5% the mSI
timer switch off the motor automatically. Read the volume of a cylinder
and calculate the bulk density by the following formula. where kXa and kYa are fluid and solid phase mass transfer coefficient re-
spectively, a0 shows the specific interfacial area, ρs and ρCO2 represents
Bulk Density¼ðweight of the seed particle Þ=ðvolume Þ ð2Þ the density of the sunflower seed material and the density of supercrit-
ical carbon dioxide respectively, ε represents bed void fraction. This
Now, to obtain the particle density, take a 100 ml of water (density model is used to fit the SFE experimental data of sunflower oil. In the
1000 kg/m3) into a measuring cylinder and pour 50 g sample of seed present investigation, the broken and intact cell model fitting parame-
particle into it. Note down the increased in volume. This is the volume ters (Z, W and xk) have been optimize by minimizing the average abso-
of the seed particles. Now calculate the particle density by dividing lute relative deviation (AARD) as objective function given by Eq. (9) the
the weight of sample by volume of seed particle. Moreover, the porosity Box complex optimization method has been used to optimize the model
is also calculated by the formula discussed in Section 4. parameters.
 
2.1. Broken and intact cell model 100 N yieldpredicted −yieldexperimental 
AARD ð%Þ ¼ ∑  ð9Þ
N i¼1  yieldexperimental 
The broken and intact cell model proposed by Sovová's has been
used to explain the behavior of the extraction process of various natural 2.2. Optimization technique
products (Pederssetti et al., 2011). This model considers that the solute
is protected by non-extractable solid cell walls inside the cell and also In the present work, a global optimization technique known as the
considered that the total extractable solute in the solid (O) is composed Box complex optimization method has been used to optimize the fitting
34 A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40

parameter of broken and intact cell model because it is a very simple given by Eq. (13).
and reliable method to optimize the problems with explicit and implicit
constraints (Erdogdu & Balaban, 2009). This method does not require Xk
1
gradient calculation and is a direct pattern search method includes gen- X¼ XL ð13Þ
k−1
eration and maintenance of search (trial) points. To obtained new L¼1
search point, all remaining search points are projected toward the cen- L≠k
ter (centroid, X) (Ravindran, Ragsdell, & Reklaitis, 2006). The flow chart
of a Box complex optimization method is shown in Fig.1 and principal Step 3: A new reflected point (XR) is now again check for feasibility. If
calculation steps are given below: f(XR) b f(XH) and satisfy all constrains the new search point replaceable
Step 1: Start with the generation of k (~2n) sequential and random and XH is replaced by XR and go to step 2. It XR is not a feasible, then a
search points by using pseudo-random variables distributed in the in- new trial point is obtained by redirecting the reflected point by half
terval of (0, 1) given as Eq. (10) the value of α.

Step 4: If at any step, reflected point XR obtained in step (3) not fol-
ðlÞ
xi; j ¼ xi þ ri; j xui −xli ; i ¼ 1; 2; :::::; n j ¼ 2; 3; :::::::; k ð10Þ low any of the constrain, it must move toward the centroid until it be-
comes feasible as given in Eq. (14)
where xi,j is the ith component of point Xj and ri,j is the random var-
iable. It is noted that, each generated point must satisfy all m constraints 1
ðX R Þnew ¼ XR þ X ð14Þ
and if not, trial point moved the centroid (X) of remaining accepted 2
point. The centroid of accepted point is given by Eq. (11)
This search procedure continues until the vertices of the simplex
1 X j¼1 made by the value of objective function at trial points shrink so that
X¼ L¼1 L
X ð11Þ
j−1 the generations of new search point are not possible by reflection or
redirection.
This process is repeated until all feasible search points are found and Step 5: In the last step, the search point is checked for convergence
must follow all m constrains. by satisfying the following criteria
Step 2: The objective function (f) is evaluated at each trial points
(vertices of simplex) and the search point corresponds highest ( )1=2
(worst) value of the objective function is reflected toward the center 1X k  2
f X −f X J ≤ε ð15Þ
and the new search point (XR) is obtained as k J¼1

X R ¼ ð1 þ α ÞX 0 −αX H ð12Þ
where, X is the center of all vertices and ε ≤10−6. The above procedure
where, XR is the new reflected point, XH represent the worst point, α≥ 1 leads to find the global optimum since the initial complex (set of the ini-
(normally taken as 1.2) and X is the centroid of all vertices except XH and tial trial points) is randomly scattered through the feasible space.

Find out the centriod of all search points except


Start

Set the value Constant & variable


(n, α, m, x0, ε, XL, XU ) Reflection of worst search point

Variable Initialization
(X1=W, Z, Xk)
Yes

Generate the k=2n search points randomly


No
Yes

Constrain violation No

No

Find out the centriod of accepted search points


Yes

Evaluate the objective function at each search points.


, ,…………….,

Find the worst point with the highest function value Find the second highest point with
Find the favorable point with the lowest function value the highest function value

Stop

Fig. 1. Flow chart of Box complex optimization method.


A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40 35

30
have a bulk density of 506.12, 510.30, 509.14 kg/m3 and corresponding
Sim. (333 K) particle density of 960.03, 1090.93, 1081.6 kg/m3 respectively is obtain-
Cumulative Extraction Yield x 10 -3 (kg)

Exp. (333 K) ed. Furthermore, the values of bed porosity are 0.52, 0.47, 0.45 and the
25
Sim. (353 K) values of corresponding particle porosity (1 − (ρb / ρp)) are 0.47, 0.53,
Exp. (353 K) 0.47 for particle sizes of 1.00, 0.75 and 0.50 mm respectively has been
20 Sim. (373 K) obtained. The amount of initial oil content (xo) in the sunflower seed
Exp. (373 K) is 1.2153 g of oil per g of non-extractable matter. In the present work,
the values of sunflower oil solubility (0.009330 g/g of CO2 at 333 K
15 and 30 MPa) are found larger than reported by Salgin et al. (2006) as
0.00669 g/g of CO2 at 333 K and 30 MPa. The higher value of solubility
is due to fact that, in the present study, co-solvent (5% and 10% of super-
10 critical carbon dioxide mass flow) is used with supercritical carbon di-
Pressure : 30 M Pa oxide, which increases the polarity of supercritical carbon dioxide. As
Particle Size: 0.75 mm a result, interactions between the molecules of solute and supercritical
5
CO2 flow rate: 1.67 x 10-4 kg/s carbon dioxide are increased and hence, solubility increases. The total
Co-solvent (EtOH) : 5% of CO2 amounts of seed kernel used in the work of Salgin et al. (2006) are 4 g
and the supercritical carbon dioxide mass flow was about 3 g/min at
0
30 MPa and 353 K. Hence, the space time (τ) of supercritical carbon di-
0 50 100 150 200 250
Time/60 (s) oxide is approximately [4 g seed/3(g/min) CO2 ≈ 1.33 min g seed g CO ].
2

Whereas, in the present case 50 g of seed kernel are used in each exper-
Fig. 2. Effect of temperature on cumulative extraction yield of sunflower seed oil. iment with 5, 10 and 15 g/min supercritical carbon dioxide mass flow.
Hence, the minimum space time are 10, 5 and 3.33 min for 5, 10 and
15 g/min supercritical carbon dioxide mass flow respectively, which is
3. Supercritical fluid extraction (SFE) minimum three times larger than used by Salgin et al. (2006), which
provide large contact time between seed kernel and supercritical carbon
The supercritical CO2 extraction experiments have been performed dioxide. As a result, value of cumulative extraction yield (g oil/g kernel)
on an extraction system supplied by Thar Technologies Inc. Pittsburgh. increases. The model equation for second and third extraction periods
The process flow diagram and working principal are discussed in detail proposed by Sovová (1994) suggests that, they strongly depend on ini-
in our previous paper of watermelon seed oil extraction (Rai et al., tial oil content and the solubility of oil in supercritical carbon dioxide. In
2015). To perform the extraction of sunflower oil from seed particles the present case, for sunflower oil, the value of initial oil content is very
of different size, a cylindrical basket of 7.5 cm diameter and 9 cm height high (xo = 1.21533) and at low pressure, the value of solubility is very
has been used. The 50 g seed particles of particular diameter, glass low which leads to very high value of tCER and tFER. As the value of tCER is
beads, glass wool are used in each experiment and detail procedure very high, the model equation which should cover the first extraction
adopted for preparing the bed of seed particle are given in the author's period only, now covers all data points on the cumulative extraction
previous paper (Rai et al., 2015). The amount of extracting oil with co- curve. The model equation for first extraction period is a straight line
solvent is collected in a sample vial in every five minutes and ethanol with the slope of QCO2yr(1 − exp (‐Z)). Hence, the model does not fit
is removed by vacuum evaporator. The amount of extracted oil is ob- well for seeds with the extraction at lower pressure with high initial
tained gravimetrically and cumulative extraction yield is obtained by oil content.
the sum of all samples.

4. Result and discussion 4.1. Effect of operating parameters on extraction yield

The solubility (yr) of sunflower oil in supercritical CO2 and supercrit- 4.1.1. Effect of temperature
ical CO2 modified with co-solvent (ethanol) has been obtained from The Fig. 2 demonstrate the effect of temperature on the extraction
first and linear part of the extraction curve, which shows that solubility yield at a constant pressure, mean particle size, supercritical CO2 mass
is directly proportional to the extraction rate. In the implementation of flow and co-solvent mass flow. In the critical region, temperature
Box complex optimization method, the fitting parameters are subjected shows the nonlinear and complex effect on extraction yield. As temper-
to four constraints: 0 b Z b 10, 0 b W b 10, 0 b xk b xo and xp b xo. Where, ature increases the cumulative extraction yield of sunflower oil also in-
xo is the highest possible value of xp. The value of mass transfer coeffi- creases with time. The probable reason for the increasing cumulative
cients in solid phase (kXa) and in the fluid phase (kYa), is calculated extraction yield with increasing temperature is that, the solubility of
with the help of the optimized values of Z, W and xk by the Eqs. (7) fatty acid present in the sunflower oil increases with the increase in
and (8). Ferreira and Meireles (2002) suggested that k-value is directly temperature. This can also be observed from the first and linear part
proportional to the ratio of solid-phase and the fluid phase mass trans- of extraction curve. The solubility of the solute is a function of tempera-
fer coefficients, i.e. k =kXaρsxk/kYaρCO2yr and is a measure of the relative ture as well as the density of the supercritical CO2, which intern is a
importance of intraparticle diffusion (diffusion-controlled period) with function of pressure as well as temperature (Mukhopadhyay, 2000).
respect to convection during the extraction process. As the value of ‘k’ Further, with temperature vapor pressure of solute also increases affect-
increases the intraparticle diffusion plays a dominant role in the extrac- ing the solubility. As in the present case, pressure is constant at 30 MPa,
tion process. For the particle sizes of 1.00, 0.75, 0.50 mm, the fixed bed hence, the solubility will be a function of temperature only. At low

Table 1
Variation of mass transfer coefficients for sunflower oil extraction with temperature.

Temperature Z W xk tCER/60(s) tFER/60(s) kYa (s−1) kXa (s−1) k


−3
333 K 1.016044 0.011021 0.733793 51.875410 115.748090 2.568026 × 10 0.452901 × 10−4 0.405307
353 K 3.000000 0.021326 0.990699 11.971004 86.065194 8.437684 × 10−3 0.876378 × 10−4 0.523648
373 K 4.452837 0.030665 0.960285 7.5045910 80.296195 14.107574 × 10−3 1.260159 × 10−4 0.403116
36 A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40

30
Sim. (20 M Pa)
0.750 mm at temperature of 353 K, supercritical CO2 mass flow of
Exp. (20 M Pa) 1.67 × 10−4 kg/s with 5% co-solvent (ethanol) and shown in Fig. 3 as
Cumulative Extraction Yield x 10 -3 (kg)

Sim. (30 M Pa)

25 Exp. (30 M Pa) a function of time. The result shows that as the pressure increases the
Sim. (40 M Pa)
Exp. (40 M Pa) cumulative extraction yield of sunflower oil increases and also at
20 MPa, extracted a quantity of sunflower oil is very small. This behavior
20 is well known and explain by the fact that the density of supercritical
CO2 reached to the density of liquid with increasing pressure which im-
15
Temperature: 353 K proves the solvation capacity of supercritical CO2 (Mukhopadhyay,
Particle Size: 0.75 mm
2000). This fact is also supported by the initial part of the extraction
CO2 flow rate: 1.67 x 10-4 kg/s
curve that the value of the tangent of extraction curve increases as the
Co-solvent : 5% of CO2
10 operating pressure increases. The solubility of sunflower oil increases
from 0.003904 kg oil/kg CO2 to 0.0279352 kg oil/kg CO2 as pressure
5
moves from 20 MPa to 40 MPa. The various investigators also studied
the effect of operating pressure on extraction yield at constant temper-
ature and found the similar trend for neem seed (Johnson & Morgan,
0 1997), Rose hip seed, physic nut seed (Machmudah et al., 2008), hiprose
0 50 100 150 200 250 300 seed oil (Reverchon, Kaziunas, & Marrone, 2000). As the pressure in-
Time/60 (s) creases from 20 to 30 MPa and 30 MPa to 40 MPa an increment of 3.5
and 1.2 times in the total extraction yield is observed. However, at
Fig. 3. Effect of pressure on cumulative extraction yield of sunflower seed oil. 20 MPa, the total extraction yield is very low due to very low solubility
of sunflower oil and proposed, broken and intact cell model does not
temperature, such as 333 and 353 K, the cumulative extraction yield is agree with experimental data, particularly in the second half of extrac-
almost same for starting 120 min and after that, it is increases with tion period. This behavior is also observed by Salgin et al. (2006) for
time. The increase in extraction yield with temperature can be ex- sunflower oil extraction. Table 2 shows mass transfer coefficient values
plained through the values of mass transfer coefficients shown in in the fluid phase (kYa), and in the solid phase (kXa) for different pres-
Table 1. The mass transfer coefficient in the fluid phase (kYa) and that sure values. As the pressure increases from 30 to 40 MPa, the value of
in the solid phase (kXa), increases with increase in temperature due to mass transfer coefficient increases due to decrease in resistances in
increase in molecular and effective diffusivity of sunflower oil in super- the solid and the fluid phase. The similar trend is also observed by var-
critical carbon dioxide. At temperature of 333 K, the total oil extracted at ious researchers for the extraction of different type of oil such as canola
the end of the transition period (t = tFER, tFER = 115.7 min) is about seed at 30 MPa and 33.5 MPa (Pederssetti et al., 2011), marigold at 12
69.05 wt% of total oil extracted. Up to transition period mass transfer and 15 MPa (Campos, Michielin, Danielski, & Ferreira, 2005). However,
is governed by convection as well as intraparticle diffusion. The remain- some authors also reports the decreasing value of fluid phase mass
ing 30.95 wt% oil at 333 K is extracted in the third period (t ≥ tFER) in transfer coefficient (kYa) for the extraction of β-carotene (Doker,
which mass transfer is governed by diffusion. Whereas, at a tempera- Salgın, Sanal, Mehmetoglu, & Calimli, 2004), apricot oil (Ozkal et al.,
ture of 353 K, the total oil extracted at the end of the transition period 2005a, 2005b), rosemary (Bensebia, Barth, Bensebia, & Dahmani,
(t = tFER, tFER = 86.1 min) is about 54.10 wt% of total oil extracted. 2009) and nutmeg oil (Machmudah, Sulaswatty, Sasaki, Goto, &
The remaining 45.90 wt% oil is extracted in the third period (t ≥ tFER) Hirose, 2006). At higher pressure of 40 MPa, the broken and intact cell
in which mass transfer is governed by diffusion. It can be seen from model agrees with absolute average relative deviation of 1.62%. More-
the Table 1 that the value of kXa which governs the diffusion phenomena over, about 99.6 wt% of the total oil has been extracted within a time
after t ≥ tFER at 353 K is nearly twice of that at 333 K and hence the cu- of 250 min and about 95% of the total oil has been extracted with in
mulative extraction curves are distinctly separate. Salgin et al. (2006) the first 165 min of experiment. The mathematical model proposed by
also found that the extraction of sunflower oil has been very fast at Sovová (1994) agrees with the experimental data very well at high
30 MPa however, they did not get sufficient information to show the ef- pressure with the average relative error of 1.62%.
fect of temperature on cumulative extraction yield (g oil/g kernel). Var-
ious authors reported that, the solubility of oil derived from different 4.1.3. Effect of particle size
seeds such as apricot, amaranth seed, almond, sunflower also increase The effect of mean particle size on the cumulative extraction yield
with extraction temperature specially at higher pressures i.e. at has been demonstrated through Fig. 4, when pressure, temperature, su-
30 MPa or above (Salgin et al., 2006; Eggers, 1996; Ozkal, Yener, & percritical carbon dioxide mass flow and added co-solvent (%) rate is
Bayindirli, 2005a, 2005b; Marrone, Poletto, Reverchon, & Stassi, 1998; kept at 30 MPa, at 353 K, at 1.67 × 10−4 kg/s and 5% of the CO2 mass
Westerman, Santos, Bosley, Rogers, & Al-Duri, 2006). Fig. 2 also shows, flow respectively. The extractor feed during experiments uses average
the broken and intact cell model is well fitted with the experimental particle size of 0.50 or 0.75 or 1.00 mm to see the effect on cumulative
data. The total extraction yield obtained after 250 min is about 1.3 extraction yield. As particle size decreases surface area per unit mass in-
times as temperature increases from 333 to 373 K. creases, hence extraction yield increases. Further, during grinding of the
seed kernel to create small particles, the cell walls are disrupted and this
4.1.2. Effect of pressure increases the ratio of broken to intact cells and subsequently offers more
The effect of operating pressure varying from 20 to 40 MPa on the accessible oil to the supercritical carbon dioxide. Contrary to this, as the
sunflower oil extraction is studied by using particle size of diameter size of seed particles increases, cumulative extraction yield decreases,

Table 2
Variation of mass transfer coefficient for sunflower oil extraction with pressure.

Pressure Z W xk tCER/60 (s) tFER/60 (s) kYa (s−1) kXa (s−1) k


−3
20 MPa 2.057369 0.835874 0.016199 320.849849 989.839839 7.264519 × 10 34.34973 × 10−4 1.685564
30 MPa 3.000000 0.021326 0.990699 11.971004 86.065194 8.437684 × 10−3 0.876378 × 10−4 0.523648
40 MPa 3.998800 0.053728 0.866394 6.714385 58.904932 10.186670 × 10−3 2.207918 × 10−4 0.416710

(Due to bad fitting of model observed value is not tuned with experiment).
A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40 37

30
seed of the low mean particle size (0.50 mm). The lowest particle size
Sim. (1.00 mm) (0.50 mm) gives approximately 93.45 wt% cumulative extraction yield
Cumulative Extraction Yield x 10-3 (kg)

Exp. (1.00 mm) at 250 min whereas 78.17 wt% and 55.00 wt% cumulative extraction
25
Sim. (0.75 mm)
yields has been obtained for 0.75 and 1.00 mm mean particle sizes re-
Exp. (0.75 mm)
Sim. (0.50 mm) spectively in the same time of extraction.
20
Exp. (0.50 mm)
4.1.4. Effect of solvent mass flow
15 The effect of the solvent mass flow on the cumulative extraction
yield has been studied using Fig. 5(a–b), when pressure, temperature,
particle size and co-solvent mass flow are kept at 30 MPa, at 353 K, at
10 0.75 mm and at 5% of supercritical carbon dioxide mass flow respective-
Temperature: 353 K ly. The Fig. 5(a) are plotted for cumulative extraction yield as a function
5 Pressure : 30 M Pa of time and Fig. 5(b) are plotted for cumulative extraction yield as a
CO2 flow rate: 1.67 x 10-4 kg/s function of the amount of supercritical carbon dioxide consumed. The
Co-solvent (EtOH) : 5% of CO2 mass flow of supercritical carbon dioxide for experiments is varied at
0 three different levels, i.e. 0.833 × 10− 4, 1.67 × 10− 4 and
0 50 100 150 200 250
Time/60 (s)
2.5 × 10−4 kg/s. the Fig. 5(a) shows as the solvent mass flow is in-
creased, the extraction rate, which is denoted by the tangent of the
Fig. 4. Effect of particle size on cumulative extraction yield of sunflower seed oil. curve is also increased and at higher mass flow, such as 2.5 × 10−4 kg/
s, extraction curve reaching to an asymptotic value. Fig. 5(a) clearly
shows that extraction rate is higher for larger flow rate because more
which indicate that, oil is not easily transported from unbroken cell carbon dioxide passes through the fixed bed of the sample and hereby
walls, and only accessible oil present on the surface is coming out more sunflower oil extracting. However, Fig. 5(b) shows that the extrac-
with the supercritical carbon dioxide. Salgin et al. (2006) obtained sim- tion rate is constant and not influenced by the flow rate at the beginning
ilar results for the extraction of sunflower oil using supercritical carbon of the all experiments and is only a function of the quantity of carbon di-
dioxide. Table 3 shows, the magnitudes of fluid side mass transfer coef- oxide that passes through the fixed bed of the sample. This indicates, at
ficient (kYa) are sufficiently large as compared to mass transfer coeffi- the beginning, that the exit concentration of sunflower oil is indepen-
cient (kXa) in the solid phase. The mass transfer coefficient in the solid dent of flow rate, and hence equilibrium is assumed to exist. The various
phase (kXa) contributes toward the diffusion of inaccessible oil in the researchers suggested that, if solid phase mass transfer resistance
seed particles and hence, it depends on the properties of the kernel, (intraparticle diffusion resistance) is considered to be important, the
the permeability of cell walls, and the capacity of the supercritical car- slop of the extraction curve in the Fig. 5(b) becomes smaller for higher
bon dioxide to access the inner parts of the particles (Ozkal et al., flow rate, but in our case it appears little bit higher for higher flow
2005a, 2005b). Fig. 4 clearly shows that, the cumulative extraction rate, which conform that intraparticle diffusion also have some impact
yield has not reached up to the asymptotic value for 1.00 mm and on the extraction (Goto, Sato, & Hirose, 1996; Salgin et al., 2006;
0.75 mm particle size, which reveals that penetration of supercritical Tonthubhimthong, Chuaprasert, Douglas, & Luewisutthichat, 2001).
carbon dioxide in to seed particle has been very low and hence, diffusion Hence, the model simulated values for solid phase mass transfer coeffi-
is also very low due to the physical structure of the seed. This behavior is cient shows some increments. Moreover, as discussed earlier, the mini-
also observed by various researchers for different type of seed particles mum space time are 10, 5 and 3.33 min for 5, 10 and 15 g/min
such as apricot (Ozkal et al., 2005a, 2005b), sunflower Salgin et al. supercritical carbon dioxide mass flow respectively, The Fig. 5(b) clearly
(2006), black pepper (Sovová, Jez, Bartlova, & Stastova, 1995), sea buck- shows that, the first part of the extraction curve almost coincided with
thorn berries (Stastova, Jez, Bartlova, & Sovová, 1996). They also con- each other which means during the aforementioned space time, the su-
cluded that supercritical carbon dioxide is not able to reach all of the percritical carbon dioxide is totally saturated with extracted oil during
oil inside the cells. the all three experiments. Hence, low space time for large flow rate
The Table 3 shows that, mass transfer coefficient in the fluid phase, does not affect the extraction rate.
(kYa), and mass transfer coefficient in the solid phase, (kXa), increases Table 4 demonstrates that, as the solvent mass flow increases the
with the decrease in the particle size. The probable reason for this incre- value of solid phase mass transfer coefficient (kXa) and supercritical
ment is due to the fact that, the increase in the surface area and the de- phase mass transfer coefficient (kYa) increases and consequently the
crease in the diffusion path length (Ozkal et al., 2005a, 2005b). For the value of resistances to mass transfer decreases. The reason of decreasing
large size particle, the extraction in considerably less in the third extrac- mass transfer resistances is due to increase in convection and decrease
tion period, which is diffusion controlled and hence the particle size re- in the film thickness around the seed particle. This explanation is in con-
duction is necessary to decrease the extraction time. In the present formity to the finding of various researchers for different seed oil extrac-
study, for large size particles the broken and intact cell model is fitted tion, such as apricot oil (Ozkal et al., 2005a, 2005b), grape seed oil
with the AARD of 3.95%, which is quite large as compared to small size (Sovová, Kucera, & Jez, 1994), parsley seed oil (Louli et al., 2004) and
particle which offers a value of AARD as 0.53%. The broken and intact egg yolk oil (Wu & Hou, 2001). Hence, higher solvent mass flow should
cell model fits the experimental result with the average relative error be used to obtain maximum efficiency of the SFE process.
of 0.53% and 1.67% for 0.50 mm and 0.75 mm particle sizes respectively. The broken and intact cell model predictions and the experimental
The highest cumulative extraction yield has been obtained from the data are in good agreement for the given experimental conditions. The

Table 3
Variation of mass transfer coefficient for sunflower oil extraction with particle size.

Particle size Z W xk tCER/60 (s) tFER/60 (s) kYa (s−1) kXa (s−1) k
−3
0.50 mm 4.191138 0.042508 1.041692 5.348764 75.918572 12.204885 × 10 1.812758 × 10−4 0.634585
0.75 mm 3.000000 0.021326 0.990699 11.971004 86.065194 8.437684 × 10−3 0.876378 × 10−4 0.523648
1.00 mm 2.057369 0.835874 0.016199 320.84984 989.839839 7.264519 × 10−3 34.47515 × 10−4 1.685564

(Due to bad fitting of model observed value is not tuned with experiment).
38 A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40

30 30
Temperature: 353 K
Sim. (0.0% of CO2)
Pressure : 30 M Pa
Cumulative Extraction Yield x 10 (kg)

Particle Size: 0.75 mm Exp. (0.0% of CO2)

Cumulative Extraction Yield x 10 -3 (kg)


25
Co-solvent (EtOH) : 5% of CO2 25 Sim. (5.0% of CO2)
-3

Exp. (5.0% of CO2)


20 Sim. (10% of CO2)
Exp. (10% of CO2)
20
15

15
10 Sim. (0.83 x 10-4 kg/s)
Exp. (0.83 x 10-4 kg/s)
Sim. (1.66 x 10-4 kg/s)
5 Exp. (1.66 x 10-4 kg/s) 10
Sim. (2.50 x 10-4 kg/s)
Exp. (2.50 x 10-4 kg/s) Temperature: 353 K
0
5 Pressure : 30 M Pa
0 50 100 150 200 250
Time/60 (s) Particle Size: 0.75 mm
CO2 flow rate: 1.67 x 10-4 kg/s
0
0 50 100 150 200 250
Time/60 (s)

Fig. 6. Effect of co-solvent on cumulative extraction yield of sunflower seed oil.

(1996) found that solubility of sunflower oil is almost directly propor-


tional to co-solvent amount. As co-solvent (ethanol) concentration in-
creases in the supercritical carbon dioxide, initial slopes of the
extraction curves also increases, indicating that the solubility of sun-
flower oil in supercritical carbon dioxide is increasing due to increase
in the polarity of supercritical carbon dioxide. The addition of 5% ethanol
to supercritical carbon dioxide increases the overall cumulative extrac-
tion yield of sunflower oil 1.45 times compared to cumulative extraction
yield in the pure supercritical carbon dioxide. Similarly, the addition of
10% ethanol to supercritical carbon dioxide increased the cumulative
Fig. 5. (a): Effect of flow rate on cumulative extraction yield of sunflower seed oil. (b):
Effect of flow rate on cumulative extraction yield of sunflower seed oil.
extraction yield of sunflower oil 1.79 times compared to cumulative ex-
traction yield in the pure supercritical carbon dioxide. Table 5 shows
that mass transfer coefficients in the fluid side (kYa) and in the solid
experimental data for the mass flow of 0.833 × 10−4, 1.67 × 10−4 and phase (kXa) increase with the addition of co-solvent. The probable rea-
2.5 × 10− 4 kg/s is fitted with the average relative errors of 1.40%, son for the increase in mass transfer coefficient is due to possible in-
1.67% and 0.9% respectively. creases in the diffusivity of sunflower oil in supercritical carbon
dioxide and as a result, the increase in effective diffusivity of solvent
in particle. Similar effects of co-solvent (ethanol) addition to supercrit-
4.1.5. Effect of co-solvent ical carbon dioxide are reported for the extraction of apricot oil (Ozkal
There are lot of chemicals may be used as a co-solvent such as meth- et al., 2005a, 2005b), pistachio nut oil (Palazoglu & Balaban, 1998),
anol, ethanol, isopropanol, formic acid etc. Out of these chemicals, etha- bioresidues from Posidoniaocenanica (Pilavtepe & Celiktas, 2013). In
nol is one of the most commonly used polar organic solvent to enhance the present study, for the addition of 5% co-solvent, the mass transfer
the extraction yield of vegetable oils because it is safe for human con- coefficient in the solid phase increases, as shows in Table 5. The Broken
sumption. Fig. 6 shows the effect of the increasing percentage of ethanol and Intact cell model fit the experimental data quite well for 0 and 5%
in supercritical CO2 on extraction yield when pressure, temperature, addition of co-solvent. The experimental data for the co-solvent amount
mean size particle and supercritical carbon dioxide mass flow are kept equal to 0%, and 5% of supercritical carbon dioxide mass flow is fitted
at 30 MPa, at 353 K, at 0.75 mm, and at 1.67 × 10−4 kg/s respectively. with the average relative error of 0.89% and 1.66% respectively.
The addition of co-solvent to the supercritical carbon dioxide is varied
from 0%, 5% and 10% of supercritical carbon dioxide mass flow to ob-
serve the effect on cumulative extraction yield. As co-solvent percent- 5. Conclusion
age increases the overall cumulative extraction yield also increases
and at a higher percentage of co-solvent, such as 10%, cumulative ex- The experimental and modelling results show that the operating
traction yield curve reaches to asymptotic value. Cocero and Calvo process parameters taken in the present study have important effect

Table 4
Variation of mass transfer coefficient for sunflower oil extraction with solvent flow rate.

Supercritical carbon dioxide Z W xk tCER/60 (s) tFER/60 (s) kYa (s−1) kXa (s−1) k
−4 −3
0.83 × 10 kg/s 2.664029 0.021807 0.842861 31.962000 163.41136 3.746372 × 10 0.448072 × 10−4 0.3668931
1.67 × 10−4 kg/s 1.999459 0.022276 0.711812 38.653800 158.13000 5.623601 × 10−3 0.915418 × 10−4 0.5662887
2.50 × 10−4 kg/s 2.726000 0.043773 0.622147 28.419125 136.70183 11.50056 × 10−3 2.698236 × 10−4 0.6706789
A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40 39

Table 5
Variation of mass transfer coefficient for SF oil extraction with co-solvent.

Co-solvent (% of CO2 flow rate) Z W xk tCER/60 (s) tFER/60 (s) kYa (s−1) kXa (s−1) k

0 2.899034 0.007950 0.9263144 19.696139 98.859801 7.324007 × 10−3 0.295890 × 10−4 0.233485
5 3.000000 0.021326 0.990699 11.971004 86.065194 8.437684 × 10−3 0.876378 × 10−4 0.5236485
10 2.424300 1.253118 0.044081 53.556860 188.221308 7.130195 × 10−3 53.85021 × 10−4 1.2310455

(Due to bad fitting of model observed value is not tuned with experiment).

on the extraction efficiency (yield). The extraction efficiency increases References


with increases with temperature, pressure, solvent mass flow and with
the addition of co-solvent (ethanol). Moreover, particle size has an in- AOAC (1995). Official method of analysis of the AOAC International (16th ed ). Virginia:
verse effect on extraction yield. The experimental data are modelled by USA.
Bensebia, O., Barth, D., Bensebia, B., & Dahmani, A. (2009). Supercritical CO2 extraction of
broken and intact cell model and fitting parameters are optimized by a
rosemary: Effect of extraction parameters and modeling. Journal of Supercritical Fluids,
global optimization Box complex method. The solid and supercritical 49, 161–166.
phase mass transfer coefficients are calculated with optimum fitting pa- Campos, L. A. M. S., Michielin, E. M. Z., Danielski, L., & Ferreira, S. R. S. (2005). Experimental
rameters and in most of the cases the dependence of extraction rate on data and modeling the supercritical fluid extraction of marigold (Calendula officinalis)
oleoresin. Journal of Supercritical Fluids, 34, 163–170.
process parameters explains very well in term of these coefficients. Casas, L., Mantell, C., Rodriguez, M., Lopez, E., & Ossa, E. M. (2009). Industrial design of
multifunctional supercritical extraction plant for agro-food raw materials. Chemical
Engineering Transactions, 17, 1585–1590.
Nomenclature Cocero, M. J., & Calvo, L. (1996). Supercritical fluid extraction of sunflower seed oil with
CO2-ethanol mixtures. Journal of the American Oil Chemists Society, 73, 1573–1578.
F objective function Doker, O., Salgın, U., Sanal, I., Mehmetoglu, U., & Calimli, A. (2004). Modeling of extraction of
β-carotene from apricot bagasse using supercritical CO2 in packed bed extractor. Journal
K the mass of inaccessible solute inside the solid particle of Supercritical Fluids, 28, 11–19.
kf solvent-phase mass transfer coefficient, m·s−1 Eggers, R. (1996). Supercritical fluid extraction of oilseeds/lipids. Supercritical fluid technology
ks solid-phase mass transfer coefficient, m·s−1 in oil and lipid chemistry (pp. 35–60). AOCS Press.
Erdogdu, F., & Balaban, M. O. (2009). Complex Method Optimization: Optimization in Food
kXa mass transfer coefficient in solid phase Engineering. CRC Press, 295–304.
kYa mass transfer coefficient in fluid phase Ferreira, S. R. S., & Meireles, M. A. A. (2002). Modeling the supercritical fluid extraction of
N mass of the solute-free solid phase, kg black pepper (Piper nigrum L.) essential oil. Journal of Food Engineering, 54, 263–269.
Goto, M., Sato, M., & Hirose, T. (1996). Extraction of peppermint oil by supercritical carbon
P total mass of extractable solute in the solid dioxide. Journal of Chemical Engineering of Japan, 26, 401–407.
QCO2 mass flow rate of solvent, kg/s International Organization for Standardization (ISO) (1977). Oilseeds residues — determi-
Q mass flow rate of solvent related to N, s−1 nation of total ash. Geneva: ISO (Standard No. 749).
Johnson, S., & Morgan, E. D. (1997). Supercritical fluid extraction of oil and triterpenoids
X oil concentration in seeds, kg oil/kg oil frees seeds
from neem seeds. Phytochemical Analysis, 8, 228–232.
X center of all search points except XH Louli, V., Folas, G., Voutsas, E., & Magoulas, K. (2004). Extraction of parsley seed oil by su-
x0 overall initial concentration related to solute-free solid phase percritical CO2. Journal of Supercritical Fluids, 30, 163–174.
(kg oil/kg oil-free solid) Machmudah, S., Sulaswatty, A., Sasaki, M., Goto, M., & Hirose, T. (2006). Supercritical CO2 ex-
traction of nutmeg oil: Experiments and modelling. Journal of Supercritical Fluids, 39,
xk hardly or inaccessible solute (kg oil/kg oil-free solid) 30–39.
xp easily accessible solute (kg oil/kg oil-free solid) Machmudah, S., Kondo, M., Sasaki, M., Goto, M., Munemasa, J., & Yamagata, M. (2008). Pres-
XU upper bounds of the explicit constraints sure effect in supercritical CO2 extraction of plant seeds. Journal of Supercritical Fluids, 44,
301–307.
XL lower bounds of the explicit constraints Marrone, C., Poletto, M., Reverchon, E., & Stassi, A. (1998). Almond oil extraction by super-
XR reflected point critical CO2: Experiments and modelling. Chemical Engineering Science, 53, 3711–3718.
V volume Martinez, J., & Martinez, J. M. (2008). Fitting the Sovová's supercritical fluid extraction
model by means of a global optimization tool. Computers and Chemical Engineering,
W parameter of slow-extraction period 32, 1735–1745.
yr solubility Mira, B., Blasco, M., & Subirats, S. (1996). Supercritical CO2 extraction of essential oils from
Z dimensionless co-ordinate orange peel. Journal of Supercritical Fluids, 9, 238–243.
Moura, L. S., Carvalho, R. N., Stefanini, M. B., Ming, L. C., & Meireles, M. A. (2005). Super-
Zw parameter for second extraction period
critical fluid extraction from fennel (Foeniculum vulgare): Global yield, composition
and kinetic data. Journal of Supercritical Fluids, 35, 212–219.
Subscripts Mukhopadhyay, M. (2000). Natural extracts using supercritical carbon dioxide. Boca Raton,
FL: CRC Press.
Nimet, G., da Silva, E. A., Palu, F., Dariva, C., Freitas, L. S., Neto, A. M., & Filho, L. C. (2011).
K easily accessible solute; the empiric parameter characterizing Extraction of sunflower (Heliantus annuus L.) oil with supercritical CO2 and subcritical
extraction rate in the second period propane: Experimental and modeling. Chemical Engineering Journal, 168, 262–268.
M start of the extraction from the inside of particles Ozkal, S. G., Yener, M. E., & Bayindirli, L. (2005a). Mass transfer modeling of apricot kernel
oil extraction in supercritical carbon dioxide. Journal of Supercritical Fluids, 35,
N end of the extraction of easily accessible solute 119–127.
O overall initial concentration Ozkal, S. G., Yener, M. E., & Bayindirli, L. (2005b). Response surfaces of apricot kernel oil
yield in supercritical carbon dioxide. LWT-Food Science and Technology, 38, 611–616.
Palazoglu, T. K., & Balaban, M. O. (1998). Supercritical CO2 extraction of lipids from
Greek letters roasted pistachio nuts. Journal of American Society of Agricultural Engineering, 41,
679–684.
Ρ density of solvent phase, kg·m−3 Pederssetti, M. M., Palu, F., Da-Silva, E. A., Rohling, J. H., Cardozo-Filho, L., & Dariva, C.
(2011). Extraction of canola seed (Brassica napus) oil using compressed propane
ρf density of supercritical CO2, kg·m−3 and supercritical carbon dioxide. Journal of Food Engineering, 102, 189–196.
ρs density of seeds, kg·m−3 Pilavtepe, M., & Celiktas, O. Y. (2013). Mathematical modeling and mass transfer consid-
Ε void fraction of the bed of seeds erations in supercritical fluid extraction of Posidoniaoceanica residues. Journal of
Supercritical Fluids, 82, 244–250.
Rai, A., Mohanty, B., & Bhargava, R. (2015). Modeling and response surface analysis of super-
Abbreviations critical extraction of watermelon seed oil using carbon dioxide. Separation and
Purification Technology, 141, 354–365.
Rai, A., Mohanty, B., & Bhargava, R. (2016). Supercritical extraction of sunflower oil: A
AARD average absolute relative deviation
central composite design for extraction variables. Food Chemistry, 192, 354–365.
CER constant extraction rate Ravindran, A., Ragsdell, K. M., & Reklaitis, G. V. (2006). Engineering optimization: Method
FER falling extraction rate and application (2nd ed.). John Wiley & Sons, 305–321.
40 A. Rai et al. / Innovative Food Science and Emerging Technologies 38 (2016) 32–40

Reverchon, E., Kaziunas, A., & Marrone, C. (2000). Supercritical CO2 extraction of hiprose Tomic, M., Savin, L., Micic, R., Simikic, M., & Furman, T. (2014). Possibility of using biodie-
seed oil: Experiments and mathematical modelling. Chemical Engineering Science, 55, sel from sunflower oil as an additives for the improvement of lubrication properties
2195–2201. of low-sulfur diesel fuel. Energy, 65, 101–108.
Salgin, U., Doker, O., & Calimli, A. (2006). Extraction of sunflower oil with supercritical Tonthubhimthong, P., Chuaprasert, S., Douglas, P., & Luewisutthichat, W. (2001). Super-
CO2: Experiments and modelling. Journal of Supercritical Fluids, 38, 326–331. critical CO2 extraction of nimbin from neem seeds- an experimental study. Journal
Shahidi, F. (2005). Bailey's industrial oil and fat products. Wiley-interscience, John Wiley & of Food Engineering, 47, 289–293.
Sons. Westerman, D., Santos, R. C. D., Bosley, J. A., Rogers, J. S., & Al-Duri, B. (2006). Extraction of
Sovová (1994). Rate of the vegetable oil extraction with supercritical CO2-I. Modelling of amaranth seed oil by supercritical carbon dioxide. Journal of Supercritical Fluids, 37,
extraction curves. Chemical Engineering Science, 49, 409–414. 38–52.
Sovová, H., Kucera, J., & Jez, J. (1994). Rate of the vegetable oil extraction with supercritical Wu, W., & Hou, Y. (2001). Mathematical modeling of extraction of egg yolk oil with su-
CO2–II. Extraction of grape oil. Chemical Engineering Science, 49, 415–420. percritical CO2. Journal of Supercritical Fluids, 19, 149–159.
Sovová, H., Jez, J., Bartlova, M., & Stastova, J. (1995). Supercritical carbon dioxide extrac-
tion of black pepper. Journal of Supercritical Fluids, 8(1995), 295–301.
Stastova, J., Jez, J., Bartlova, M., & Sovová, H. (1996). Rate of the vegetable oil extraction
with supercritical CO2-III. Extraction from sea buckthorn. Chemical Engineering
Science, 51, 4347–4352.

You might also like