See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/257084611

Analysis and simulation of an industrial

vegetable oil refining process
ARTICLE in JOURNAL OF FOOD ENGINEERING JUNE 2013
Impact Factor: 2.77 DOI: 10.1016/j.jfoodeng.2013.01.034

CITATIONS

READS

455

4 AUTHORS, INCLUDING:
Gabriele Landucci

Gabriele Pannocchia

Universit di Pisa

Universit di Pisa

65 PUBLICATIONS 387 CITATIONS

85 PUBLICATIONS 955 CITATIONS

SEE PROFILE

SEE PROFILE

C. Nicolella
Universit di Pisa
123 PUBLICATIONS 1,844 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

Available from: Gabriele Pannocchia

Retrieved on: 28 February 2016

Journal of Food Engineering 116 (2013) 840851

Contents lists available at SciVerse ScienceDirect

Journal of Food Engineering

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

Analysis and simulation of an industrial vegetable oil rening process

Gabriele Landucci a,, Gabriele Pannocchia a, Luigi Pelagagge b, Cristiano Nicolella a
a
b

Dipartimento di Ingegneria Civile e Industriale, Universit di Pisa, Largo Lucio Lazzarino, 56126 Pisa, Italy
SALOV Societ Alimentare Lucchese Oli E Vini S.p.A. 1582, Via Montramito, San Rocchino 55054, Italy

a r t i c l e

i n f o

Article history:
Received 10 August 2012
Received in revised form 1 November 2012
Accepted 27 January 2013
Available online 4 February 2013
Keywords:
Vegetable oil rening
Process simulation
Advanced thermodynamic models
Formation of ammable mixtures

a b s t r a c t
This work focuses on the performance analysis of an industrial vegetable oil renery. Using a commercial
process simulator, a process model was developed and validated against actual vegetable oil renery eld
data. The simulator allowed investigating both energy and safety aspects related to the presence of residual extraction solvent (extraction grade hexane) in the processed crude vegetable oil. The critical nodes
for hexane accumulation in the process were evaluated, both considering ordinary operative conditions
and undesired process deviations due to increase of the hexane content. In this latter case, the control
actions able to restore the normal operation were simulated, in terms of increased utility consumption
(e.g., motive steam for ejectors and cooling water) or by modifying and optimizing equipment operating
conditions. Finally, the possibility of ammable mixtures formation inside process vent pipes, caused by
the entrainment of air due strong vacuum conditions, was also investigated.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
Edible oil production by extraction processes greatly increased
in the last century due to both higher request and consumption
(FAO, 2011) and the progressive availability of more efcient process technologies and equipment (Bockisch, 1998; Mielke, 1990;
Shahidi, 2005; Veloso et al., 2005; Calliauw et al., 2008; Cuevas
et al., 2009; Haslenda and Jamaludin, 2011; Szydowska-Czerniak
et al., 2011; Zulkurnain et al., 2012). A critical phase of the edible
oil production chain is the nal rening aimed at removing free
fatty acids, which, in too high concentrations, lead to the rancidity
of the oil (Cavanagh, 1976; Sullivan, 1976; Keurentjes et al., 1991;
Bhosle and Subramanian, 2005; Martinello et al., 2007; Calliauw
et al., 2008; Cuevas et al., 2009; Carmona et al., 2010; Akterian,
2011), and other minor components such as phospholipids, pigments, proteins, oxidation products and the possible residual content of the solvent used for the extraction process. The main
operations involved in conventional rening for removing the
mentioned components are degumming, neutralization, washing,
drying, bleaching, deodorization and ltration (Gunstone et al.,
1994; Mag, 1990; Loft, 1990; Shahidi, 2005; Santori et al., 2012).
This stage of the production chain is crucial for the quality
enhancement of the nal product.
One the more critical aspects of vegetable oil rening is related to
the presence of residual volatile solvent used for the extraction. In
particular, due to the low vapor pressure, the residual solvent may
cause a loss of efciency in high temperature vacuum operations
Corresponding author. Tel.: +39 050 2217907; fax: +39 050 2217866.
E-mail address: gabriele.landucci@diccism.unipi.it (G. Landucci).
0260-8774/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jfoodeng.2013.01.034

(such as drying, bleaching and deodorization). In these operations,

vacuum conditions are often obtained by ejector systems (Bockisch,
1998; Mag, 1990; Loft, 1990; Muth et al., 1998; Akterian, 2011),
whose costs are mainly related to the consumption of steam and
cooling water for condensation. A possible increase of the residual
solvent concentration has a negative impact on these costs, besides
worsening the environmental impact related due to higher emission
factors (odors, pollutant, etc.) (MRI, 1995; Muth et al., 1998).
Another criticality is due to the fact that the extraction solvent is
typically technical hexane (extraction grade hexane) (Dunford and
Zhang, 2003; MRI, 1995) a highly ammable liquid and vapor (GHS
hazard statement, Shell, 2012). In some critical nodes of the process,
the solvent accumulates in the vapor phase and mixing with air may
occur, potentially leading to the formation of ammable mixtures and
conned explosion of the equipment in case of accidental ignition
(NFPA, 2007; Lees, 1996; Tugnoli et al., 2012). As reported in a previous work (Landucci et al., 2011) this mainly affects crude oil storage
tanks, as also experienced in two recent severe accidents which involved several fatalities (La Repubblica, 2006; El Economista, 2007).
Nevertheless, since very low pressure vacuum operations characterize several stages of the process (Bockisch, 1998; Mag, 1990; Loft,
1990; Shahidi, 2005; Muth et al., 1998; Akterian, 2011; Santori
et al., 2012), a low but signicant amount of atmospheric air is entrained by seals or gaskets mixing with the process vents. This may
lead to the formation of ammable mixtures also in process lines.
Even if the vegetable oil rening process is well known, the
industrial facilities are continuously subjected to modications,
revamping and new technologies implementation in order to
achieve a higher process efciency (Shahidi, 2005). In the literature,
several examples of simulation and experimental analysis of each

841

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

single stage of the rening process are available (Keurentjes et al.,

1991; Wills and Heath, 2005; Zin, 2006; Ceriani and Meirelles,
2006; Didi et al., 2009; Farhoosh et al., 2009; Sampaio et al.,
2011), while a systematic performance analysis, which has been
extensively applied in the framework of process/chemical industry
(Motard et al., 1975; Shaw, 1992; Biegler et al., 1997; Vadapalli and
Seader, 2001; Hoyer et al., 2005; Towler and Sinnott, 2013) and
aimed at taking into account the mentioned critical aspects, is still
lacking.
The present analysis was therefore addressed at investigating
the vegetable oil rening process by the development of detailed
simulation model using the commercial software Honeywell UniSim Design (Honeywell, 2010a,b). The analysis was aimed at
identifying the main process streams, the reference substances,
and quantifying the mass and energy uxes among the rening
plant. The process simulator was applied to case studies representative of the current industrial applications, deriving the input data
from inlet conditions of an actual vegetable oil renery. In
particular, the vegetable oil renery of SALOV S.p.A., located in
San Rocchino (Massarosa) (Italy), was considered in the analysis.
The simulation model was validated against actual eld data of
the same plant and a sensitivity analysis was performed in order to
evaluate the utility consumption and potential safety relevant situations depending on the quality of the input feedstock, in particular evidencing the effect of the residual solvent content on the
whole process efciency.

2. Materials and methods

2.1. Methodological approach
The owchart of the methodology is reported in Fig. 1, and is
based on the approach followed in a previous work by Landucci
et al. (2011) for the analysis of crude vegetable oil storage systems.
The rst step of the methodology was related to characterization of the crude vegetable oil composition, which, for each type
of seed or fruit, is determined by environmental conditions during
plant grow and farming soil characteristics. A reference composition representative of different types of oil was used to perform
the further steps of the methodology. The second step (see Fig. 1)

Characterization of the
crude vegetable oil
composition

Schematization of the oil

refining process

Thermodynamic model
for the estimation of
vapor/liquid equilibrium

Software implementation
of the refining process

Analysis of a case study

and optimization of
process conditions

Sensitivity analysis

Safety aspects

Collection of typical
operations and
process conditions
from actual plants
Validation with
experimental data

UniSim tool
Set up of optimal
equipment operative
conditions

Assessment of
utilities requirement
Increase of residual
solvent concentration

Fig. 1. Flowchart of the methodology.

consisted in the schematization of the typical process operations

for oil rening, with denition of operative conditions for process
equipment and evaluation of energy requirements (steam consumption and other utilities). Then, a thermodynamic model was
applied in order to reproduce the vapor/liquid equilibrium of the
crude vegetable oil system (step 3 in Fig. 1), implementing the
presence of water and residual solvent content. The model was validated against available experimental data.
Next (step 4 in Fig. 1), the rening process was simulated with
Honeywell UniSim Design. Specic subroutines were implemented for the simulation of non-standard utilities such as the
ejectors used for keeping vacuum conditions in process vessels
and the deodorization operation.
The process simulator was used to perform the optimization of
operative conditions given the optimal composition of the feedstock, in order to minimize the costs related to utilities (step 5 in
Fig. 1). A sensitivity analysis was performed (step 6 in Fig. 1) aimed
at identifying the system response to the increasing residual solvent content in the feedstock and possible restoration control measures. Finally, the possibility of formation of ammable mixtures
inside process lines was investigated (step 7 in Fig. 1).

2.2. Characterization of the crude vegetable oil

Crude edible oil is a complex multicomponent system. Recent
studies were focused on the detailed experimental or numerical
characterization of the vapor/liquid equilibrium of this system
(Christov and Dohrn, 2002; Rodrigues et al., 2004; Calliauw et al.,
2008; Ceriani et al., in press). Furthermore, advanced modeling
tools were implemented for the analysis of the rening process
taking into account different relevant triacylglycerols (TAGs), partial acylglycerols (monoacylglycerols MAGs, diacylglycerols DAGs),
and the possible residual acid components, such as free fatty acids
of different type (Rodrigues et al., 2004; Farhoosh et al., 2009; Chiyoda et al., 2010; Silva et al., 2011; Sampaio et al., 2011; Gerasimenko and Turyan, 2012; Teles dos Santos et al., in press;
Ceriani et al., in press). Nevertheless, since the aim of the present
study was to evaluate the effect of residual hexane content on
the safety and energy performance of process equipment, a simplied reference composition was considered. The same approach
was followed in several studies on edible oil processing available
in the literature (Zhang et al., 2003; Ruiz-Mendez and Dobarganes,
2007; Cerutti et al., 2012).
The reference composition implemented in the simulation model is reported in Table 1. Such composition is based on the typical
crude sunower oil feedstock used in SALOV S.p.A. vegetable oil
renery, as already considered by Landucci et al. (2011). The oil
phase of the edible oil was schematized as pure triolein (reference
TAG), while the free fatty acids content is assumed as pure oleic
acid. Minor components such as sterols, tocopherols and squalene
are also present and were implemented in the UniSim Design list
of components as hypo component (Honeywell, 2010a). The hexane residual content (schematized as pure n-hexane) was taken as

Table 1
Reference composition of the crude vegetable oil
considered in the present study based on SALOV
renery data.
Components

Mass fraction (%)

Triolein
Oleic acid
n-Hexane
n-C29H60
Sterols
Tocopherols

97.29
2.00
0.10
0.15
0.40
0.06

842

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

0.1% in the baseline case, the maximum value allowed for the crude
oil processed by SALOV S.p.A. Nevertheless, this value may be higher, up to 810 times the reference value, depending on the type of
seed and oil origin. Moreover, the presence of residual and/or process water was also taken into account in the evaluation of vapor/
liquid equilibrium (see Section 2.4).

fatty acids are stripped by steam in a tray tower and then condensed in a spray tower, while steam with non-condensable vapors
are sent to the ejectors section. Ejectors are also used to keep the
required vacuum conditions in other low-pressure sections (ash
separator and bleaching reactor, see Fig. 2).
Table 2 provides the detailed operative conditions used in each
section of the rening process.

2.3. Schematization of the oil rening process

2.4. Thermodynamic model
In order to consider typical rening process conditions, the
purication of sunower oil was simulated assuming a free fatty
acid content of 2% to be reduced up to 0.04% (percentages are expressed on a weight basis). A schematization of SALOV process is
reported in the Process Flow Diagram (PFD) shown in Fig. 2. This
process is similar to others reported in the literature (Bockisch,
1998; Ceriani and Meirelles, 2006; Mag, 1990; Loft, 1990; Muth
et al., 1998; Shahidi, 2005; Santori et al., 2012). The oil is rst neutralized by adding sodium hydroxide to an intermediate grade of
acidity removing the neutralized soaps and waxes with a centrifugal separator. Next, the oil is degummed by adding water and subsequently it is sent to centrifugal separation to split the oil fraction
from the solid waste. During this step, the oil is washed with water
and consequently it is dried in a ash separator under vacuum conditions. Next the oil is sent to the bleaching treatment, aimed at the
removal of color-producing substances and further impurities. In
this operation the oil is mixed with bleaching earth and activated
carbon in a stirred reactor operating under vacuum conditions
for the adsorption of the mentioned contaminants. The stream containing bleaching earth and activated carbon is modeled as pure
water in the process simulator.
Next the oil is ltered and sent to the deodorization treatment.
This section consists of a physical neutralization with low pressure steam at high temperature under vacuum conditions. The free

The choice and the software implementation of the thermodynamic model is a crucial step for a sound modeling of the rening
process, since it allows determining the operative conditions in
each equipment unit. The UniSim Design software can implement
the thermodynamic model with different property-packages
(Honeywell, 2010b) for determining the correct vapor/liquid equilibrium of complex mixtures. The use of the process simulator for
the thermodynamic modeling of complex multicomponent systems is extensively diffused in both scientic and technical studies
(Harwardt et al., 2008; Szabo et al., 2011; Towler and Sinnott,
2013). It is worth mentioning that equation of state models, in general, and the PengRobinson one and its variants, in particular, are
recommended models in most commercial simulators for hydrocarbon mixtures, also in the presence of water, over a wide range
of pressure and temperature combinations. More details on the
UniSim Design code validation are reported elsewhere (Honeywell, 2010a,b).
In the present study, the selected Property Package is based on
the PengRobinson equations (Peng and Robinson, 1976) corrected
with the Twu Alpha function (Twu et al., 1995; Honeywell, 2010b),
which takes into account the excess free energy in order to have
more accurate prediction of vapor pressure. More details on the
thermodynamic model implemented in software are reported

V
E5

E6

E8

W
W
EJ3c
E7
W

MPS
EJ1a/b

EJ3a/b

EJ2a/b

EA

E4

C2
PI1

W
SH

P1

P2 E1b

E1a

E2

F1

G5

TI4

FFA

FO
C1
R3

S1
R1

R2

EE
TI3 FI1

TI1

G1

SW

G2

WW

G3

G4
TI2

E3

RO

LPS
NEUTRALIZATION
DEGUMMING

WASHING

DRYING

BLEACHING

DEODORIZATION

Equipment items: C: column; E: heat exchanger/condenser; EJ : steam ejector; F: filter; G: pump; P: centrifugal separator; R: reactor; S: flash separator.
Material streams : EA: bleaching earth & activated carbon; EE: exhausted earth; FFA: free fatty acids; FO: feedstock oil; LPS: low pressure steam; MPS:
medium pressure steam; RO: refined oil; SH: sodium hydroxide; SW: soaps & waxes; V: vents; WW: Waste water; W: Water.
Fig. 2. Schematization of the vegetable oil rening process. Tags represent the process variables used for model validation.

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

Table 2
Operative conditions of the main sections of the rening process.
Process section

Operative temperature (C)

Operative pressure (kPa)

Neutralization
Degumming
Washing
Drying
Bleaching
Deodorization

20
6070
90
90
105
230

100
100
100
5
6
0.2

843

accumulated, both in ordinary process conditions and following

unexpected process deviations. For the sake of brevity only the main
issues related to vegetable oil rening simulator and innovative aspects connected with the analysis of the more important equipment
are summarized in the following sections. In order to highlight the
complexity of the developed process simulation model and the
potentialities of the method, the Supplementary information le reports samples of the UniSim Design process ow diagrams (PFDs).

The process simulation model, implemented in the UniSim

Design software, was aimed at evaluating the energy consumption
of the plant and the more critical nodes in which hexane is

2.5.1. Condensers
The condensers are critical units under the point of view of
energetic efciency of the process. These units are aimed at condensing the steam outlets from the ejectors connected to the main
process equipment to keep vacuum conditions (see specic
description in Section 2.5.3) by the use of cooling water available
in the renery plant. Fig. 2 shows the condensers associated to
the ejector of the drying section (E5), bleaching (E6) and deodorization (E7 for the rst and second stage ejectors, E8 for the third
stage ejector). The sample UniSim Design PFD for the condenser
E5 is shown in Supplementary information.
The cooling water owrate is the variable manipulated by the
software (ADJ 1 operator) which determines its value by imposing
a xed temperature of 20 C for the condensate. This implementation allows for a better stability of the model in presence of input
deviations on the crude oil composition. The condenser parameters
were determined after a preliminary rating operation. The typical
range of cooling water owrates, derived from actual plant design
data, was imposed in a preliminary dedicated simulation model together with the geometry documented in the equipment datasheets, thus calculating in the so-called rating mode an average
value for the pressure drops and heat transfer coefcient.
Then, condensers are implemented in the overall simulation
model by imposing the pressure drops on both tubes and shell sides,
and the product of the geometry area times the overall heat transfer
coefcient (design mode, see Honeywell (2010a) for more details).
This modeling approach was associated to the condensers E5,
E6 and E7 (see Fig. 2), while for condenser E8 a different approach
was followed. Since this unit receives the cooling water already
used in condenser E7, associated to ejectors EJ3a and EJ3b (see
Fig. 2), its modeling using an a priori xed value for the overall
transfer coefcient may be inaccurate. In fact, the cooling water
is manipulated to satisfy specications on other upstream units
and may vary signicantly. Therefore, the so-called rating mode
(see Honeywell (2010a) for more details on this procedure) was
used, in which one species the exchanger geometry (number/
dimensions/arrangements of tubes, shell passes, etc.) and appropriate correlations are internally used to evaluate the heat transfer
coefcients and pressure drops on the basis of actual owrates.

Fig. 3. Validation of the thermodynamic model developed in UniSim Design. HEX:

residual hexane content in the crude vegetable oil (% by weight basis). Experimental
data were derived from Smith and Wechter (1950).

2.5.2. Deodorization column

The deodorization stage is aimed at removing minor components (e.g., squalene and polycyclic aromatic compounds) which
cause odor and the loss of quality of the nal product. The deodorization column (C1 in Fig. 2) is a stripping column made of ve
chambers, each fed with low pressure steam (LPS, at 1.5 bar). The
total LPS mass owrate is set as the 1.8% of the total rened oil
owrate. The hot exhausted vapors from each chamber are collected and fed to a water scrubber (C2 in Fig. 2), where the fatty
acids are removed and purged.
In order to reach the required strong vacuum conditions (in particular, 0.2 kPa pressure and temperature higher than 220 C) the
ejector system depicted in Fig. 2 is required.
The column was modeled in the UniSim Design software by
implementing six separators in series, aimed at representing the
ve chambers of the column C1 plus the bottom of the column,
in which the separation is also carried out thus reaching the

elsewhere (Honeywell, 2010b), while Appendix A summarizes the

key parameters and equations used to predict enthalpy, entropy,
the fugacity coefcients for each component of the mixture and
thus the vapor/liquid equilibrium.
In order to test the validity of the model, a comparison with
available experimental data was carried out. A signicant number
of literature studies focuses on vegetable oil/hexane mixtures at
high concentrations of hexane in the liquid phase (Fornari et al.,
1994; Ceriani and Meirelles, 2004; Smith and Florence, 1951), typical of extraction processes. The only available data for diluted
solutions, which are signicant in the present case, are reported
by Smith and Wechter (1950). Data are referred to the soybean
oil/n-hexane solutions with a residual solvent content in the range
0.21.32% by weight. The hexane vapor pressure is measured in the
experiments as a function of the temperature. The model was tted
on the experimental results by setting the trioleinhexane binary
interaction coefcient to 0.095 (Honeywell, 2010b). Notice that
for all other pairs of compounds, the default values of binary interaction coefcients were used. All binary interaction coefcients are
reported for completeness of exposition in Appendix A.
Fig. 3 reports a comparison between experimental data and values calculated with Unisim Design of n-hexane partial pressure in
the vapor phase as a function of temperature and hexane concentration in the oil phase. As can be observed in this gure, the model
gives a quite accurate prediction with major deviations on the safe
side (e.g., 17% overprediction of n-hexane vapor pressure). The data
were linearly extrapolated for temperatures lower than 75 C as already performed in a recent publication (Landucci et al., 2011), in
which, however, the effect of water on the vapor phase composition was neglected and the model was set up only for the analysis
of storage conditions.
2.5. Simulation model implementation

844

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

vapor/liquid equilibrium conditions. For the rst four separators an

energy stream is added in addition to the LPS stripping stream in
order to simulate the presence of high pressure steam (saturated
steam at 40 bar) fed to internal heating coils inside the C1 column
chambers in order to keep high temperature conditions. The Unisim Design PFD is reported in Supplementary information.

2.5.3. Ejectors
Several steam driven ejectors are used in the renery to obtain
the needed vacuum conditions in the process equipment. As evidenced in Section 2.5.1 these pieces of equipment are critical for
the energy performance assessment of the renery plant. However,
no dedicated model is available in the process simulator for ejectors. Thus, a specic modeling tool was implemented in the software in order to achieve an accurate performance evaluation
exploiting the UniSim Design software User Unit Operation

function. The function allows inserting the data derived from actual ejector systems datasheets, in particular the design curves.
These curves report the entrainment ratio (1/l), given by the suction ow related to air at 20 C respect to the motive steam ow, as
a function of the ratio between the discharge and suction pressures
(Pd/Ps). The curves vary according to the parameter given by the ratio between suction and motive steam pressures (Ps/Pm). The analysis of the design curves and optimization of ejector systems is
extensively described in the technical literature (Meherwan,
1999; Akterian, 2011).
Hence, by setting the pressures of the equipment in vacuum
conditions (e.g., Ps), of the motive steam (e.g. Pm) and of the discharge (Pd) it is possible to derive by reading on the curves the
entrainment ratio and calculating the necessary mass ows as
follows:

X1

X2

0.001
0.002
0.005
0.010
0.020
0.050

4.14
3.81
3.38
3.03
2.70
2.26

0.983
0.910
0.732
0.673
0.615
0.489

ma 1
MS K ej

where ma is the entrained ow of air at 20 C, MS is the ow of motive steam and Kej is a correction factor for suction ows other than
air, expressed as follows:

Table 3
Fitting parameters for the approximation of the ejectors
design curves (see Eq. (3)).
Parameter (Ps/Pm)

s
RS T S
K ej
RL T L

where RS is the gas constant of suction ow, RL the gas constant of

air (=287 J kg1 K1), TS the temperature (in K) of suction ow, TL
the reference air temperature for the ejector (=293 K).

Table 4
Comparison between the process parameters evaluated by the model and the available eld data. For tags locations, see Fig. 4.
TAG

Description

Units

Model results

Field data

FI1
PI1
TI1
TI2
TI3
TI4

Rened oil exit ow

Pressure in the deodorization column
Temperature of the bleaching reactor
Temperature of crude oil at the deodorization inlet
Rened oil exit temperature
Temperature of the deodorization column top side

kg/h
kPa
C
C
C
C

14,558
0.2
104.8
231.7
160.9
135.8

14,075
0.22
110.1
230.7
154.8
153.0

E1 C1 C2 V1

CW2

Crude oil from

neutralization

C3 C4

V2

Drying
CW4

C5 E5 E6 E7 V3

CW1
3
Bleaching
earth &
activated
carbon

CW6

Bleaching
5

H1 H2 H3
4
W1

CW3

Refined oil
to storage

Deodorization

LEGEND:
C
CW
E
H
V
W

Condensed steam
Cooling water
Energy stream
Low or medium pressure steam
Vent
Process waste
Material stream tag

H4 H5 H6 E2

W2

CW5

H7 H8 H9 E3 E4

Fig. 4. Schematic representation of the heat and material balance on the analyzed plant sections.

845

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

Table 5
Heat and material balance on the plant sections analyzed in the present study. For the identication of the streams, refer to Fig. 4. Composition is expressed in percentages by
weight basis.
Item

Material streams
1

Temperature (C)
Pressure (kPa)
Flowrate (kg/h)
Triolein (%)
Water (%)
n-Hexane (%)
Oleic acid (%)
Other (%)
a
b
c

2
a

90.0
195.0
14,887.5
98.14
0.55
0.10
0.60
0.61

84.3
200.0a
14,795.2
98.74
0.01
0.02
0.61
0.62

4
a

105.2
210.0a
14,695.0
98.75
0.01
0.01
0.61
0.62

20.0
186.0
14,558.4
99.62
0.0
0.0
0.0
0.38

5b

W1

W2

25.0a
200.0a
14.8
0.0
100.0
0.0
0.0
0.0

105.0
8.0
97.8
100.0
0.0
0.0
0.0
0.0

48.4
0.2
133.0a
5.2
0.0
67.3
0.0
27.5c

Value imposed to process simulator.

The stream containing bleaching earth and activated carbon is modeled as pure water.
Spent bleaching earth.

Table 6
Heat and material balance on the plant utilities. For the identication of the streams, refer to Fig. 4. C = steam condensate; CW = cooling water; E = energy stream; H = steam;
V = vent.
ID

Physical
state

Drying
C1
C2
CW1
CW2
H1
H2
H3
V1
E1

section
Liquid
Liquid
Liquid
Liquid
Vapor
Vapor
Vapor
Vapor

Bleaching section
C3
Liquid
C4
Liquid
CW3 Liquid
CW4 Liquid
H4
Vapor
H5
Vapor
H6
Vapor
V2
Vapor
E2

Deodorization section
C5
Liquid
CW5 Liquid
CW6 Liquid
H7
Vapor
H8
Vapor
H9
Vapor
V3
Vapor
E3
E4
E5
E6
E7

Description

Thermal power
(kW)

Steam condensate associated to ejector EJ1a

Steam condensate associated to ejector EJ1b
Cooling water fed to the drying section condensers
Cooling water exiting the drying section condensers
Motive steam fed to rst stage ejector EJ1a
Motive steam fed to second stage ejector EJ1b
Drying steam pre-heating in E1a
Vent exiting from drying section
Heat removed in downstream degumming section with heat exchanger
Steam condensate associated to ejector EJ1a
Steam condensate associated to ejector EJ1b
Cooling water fed to the bleaching section condensers
Cooling water exiting the bleaching section condensers
Motive steam fed to rst stage ejector EJ2a
Motive steam fed to second stage ejector EJ2b
Bleaching steam pre-heating in E1b
Vent exiting from bleaching section
Bleaching pre-heating


X 2
ma
X1

MS
Pd =P s

where X1 and X2 are tting constants reported in Table 3 for different values of the parameter Ps/Pm.
In the process simulator, for each equipment operating in vacuum conditions the suction temperature, the suction pressure
and the motive steam pressure are specied as input parameters;
hence the software applies Eqs. (1)(3) to evaluate the motive

Temp.
(C)

Pressure
(kPa)

150.1
1153.0
9282.0
9282.0
70.1
53.4
1153.0
70.6

19.0
127.5
8.0
18.0
175.5
175.5
127.5
123.2

16.5
250.0
150.0
149.9
900.0
900.0
250.0
108.0

301.0
30.4
1180.8
1180.8
15.6
27.6
301.0
35.4

127.5
19.8
8.0
20.0
175.5
175.5
127.5
134.0

250.0
16.5
150.0
150.0
900.0
900.0
250.0
108.0

1537.0
240,000.0
240,000.0
1100.1
157.1
26.0
33.8

19.8
8.0
12.0
175.5
175.5
175.5
132.4

102.5
150.0
140.9
900.0
900.0
900.0
108.0

142.0

11.0

Steam condensate associated to ejector EJ3

Cooling water fed to the deodorization section condensers
Cooling water exiting the deodorization section condensers
Motive steam fed to rst stage ejector EJ3a
Motive steam fed to second stage ejector EJ3b
Motive steam fed to third stage ejector EJ3c
Total vent owrate exiting from deodorization section condensers (E7 and
E8)
C1 chambers external coil heating
Steam (40 bar) for oil preheating
Cooling of scrubber C2 recycle
Air cooler
Cooling unit

In order to obtain more realistic results, the actual datasheet of

industrial ejector systems were obtained (Krting Hannover AG,
1994) inserting in the UniSim Design software Unit Operation
function the numerical interpolation of the design chart curves
as follows:

Flowrate
(kg/h)

89.0
448.0
53.0
1055.0
163.0

steam ow which is necessary to keep an imposed discharge

pressure.
Therefore, by varying the input conditions, e.g. due to deviations in the process (in particular, the increase of volatile compounds affect the suction ow), the energetic consumptions are
evaluated by calculating the necessary motive steam ow needed
to restore the optimum process conditions.
3. Results and discussion
3.1. Model validation and case study analysis
In order to validate the process simulator, actual eld data were
derived from SALOV S.p.A. renery during typical working

846

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

Fig. 6. Variation of the main process parameters as a function of the increasing

hexane concentration in the inlet crude oil: (a) cooling utility (water) and heating
utility (ejectors motive steam) consumption; (b) hexane removal (% of the initial
residual content in the crude oil) in the different renery sections.

Fig. 5. (a) Example of optimization for ejector EJ1a/b for the base case with hexane
residual content of 0.1% by weight basis; (b) optimization of intermediate pressure
(Pint) as a function of different hexane residual content in the inlet crude oil (% by
weight basis) for the three ejectors groups reported in Fig. 2. MS = motive steam.

Table 7
Air inltration considered in the process vents of the different plant sections
operating under vacuum conditions.
Type of inltration

operations, and compared with the ones predicted by the model in

the correspondent locations. The results of the validation are reported in Table 4, in which the tags of the monitored process variables are indicated in the PFD shown in Fig. 2. As shown in the
table, the simulator allows for a good reproducibility of actual process conditions, such as temperature, pressure and material
streams, with a maximum relative error of 11%.
The simulator was used to investigate the criticalities of the vegetable oil rening process and the inuence of the residual solvent
content on the process efciency. In particular, the simulator allowed identifying the more critical nodes in which the solvent is
accumulated and tracing the different sections respect to the initial
crude oil content. The main hexane accumulation node is the drying
ash, in which 76.6% of hexane is removed, while minor residual are
accumulated in the other sections, in particular 12.7% and 10.7%
respectively in bleaching and deodorization sections. Thus, a possible increase of hexane residual may lead to process efciency decrement, in terms of motive steam consumption for the ejector system.
In order to systematically quantify the renery energy consumption and to determine the critical factors affecting the efciency, the process simulator results were analyzed.
Fig. 4 reports the block diagram of the process evidencing the
main material streams together with the energy and utility lines.
The main product streams are marked together with the possible
process vent and wastes/residuals (respectively labeled with V
and W in Fig. 4). The cooling utility is mainly water (CW in
Fig. 4), while steam at different pressures is the heating utility, also

Type 1
Type 2a
Type 3
a

Air inltration (kg/h)

Drying

Bleaching

Deodorization

3
5
8

3
5
8

3
6
10

Value derived from manufacturer data (Krting Hannover AG,1994).

employed in the mentioned ejector system. The steam entering

each block are labeled with H in Fig. 4, while the exit condensate
is labeled with C. In order to simulate further heat exchanges in
and out of the simulator boundaries and passing between unit
operations (steam coils, air coolers, etc.) several energy streams
were added to the scheme (labeled with E in Fig. 4) using a specic UniSim Design software function.
The quantication of the heat and material balance for the
scheme (Fig. 4) is reported in Tables 5 and 6, respectively for process streams and utilities.
As can be seen in Table 5, the oil content (schematized as pure
triolein) increases passing through the different sections. The major part of water is eliminated in the drying section, as expected,
while the acid fats content, residual of the upstream neutralization
is totally removed in the deodorization section.
Considering the energy consumptions, synthetically represented by the results shown in Table 6, the bleaching section features the lowest thermal requirements, both in terms of hot and
cold utilities. On the contrary, it clearly appears that the most critical section, under the point of view of energy requirements, is the

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

Fig. 7. Comparison between the ammability range of hexane considering two

inert reference gases (carbon dioxide and nitrogen) and vapor concentration in the
venting line for (a) drying, (b) bleaching and (c) deodorization considering a
residual hexane content of 0.1% by weight basis in the inlet crude oil. For air
inltration types characterization, see Table 7.

deodorization. Both steam and cooling water utilities have the

highest requirements in order to keep the severe operative conditions imposed by the process. In particular, low pressure (0.2 kPa)
leads to major motive steam consumption and associated cooling
water for condensation, while the high operative temperature of
the column (230 C) is kept also by the use of additional heating
(energy stream E4 in Table 6) carried out with high pressure steam.
Besides, additional heat exchangers are needed for cooling the
scrubber C2 (see Fig. 2) recycle and the vents before the treatment
and the discharge in the atmosphere.

3.2. Process optimization and sensitivity analysis

The analysis of the renery in the baseline case (0.1% hexane by
weight basis in the inlet crude oil) highlighted the criticalities related to the energy consumptions in the renery low pressure units
(e.g., drying, bleaching and deodorization). Since the ejector systems operative conditions affect the whole renery energetic performance, the process simulator was applied in order to optimize

847

the operating conditions for the minimization of motive steam

consumption. The optimization was carried out on the three ejector systems considering that the motive steam is available in the
plant at the same pressure (medium pressure steam, MPS at 9 bar).
Fig. 5a reports an example of optimization, in particular related
to the ejector system connected to the drying ash drum (EJ1a/b
with condenser E5, see Fig. 2). As can be seen in the scheme, the
ejector is constituted by two different sections in which Ps is the
suction pressure, representative of the equipment operative
conditions, Pout the system discharge pressure, MSA and MSB the
motive steam streams respectively for the rst and second stage,
and Pint is the intermediate pressure, which is the degree of freedom (DOF) to specify for the optimization. The optimization is carried out by varying both MSA and MSB and nally obtaining the Pint
which minimizes the overall steam consumption (e.g., the sum of
MSA and MSB), as shown in Fig. 5a. Determining the intermediate
ejectors pressure allows for the process energetic efciency
enhancement.
The described optimization method can be performed also by
considering a possible increase of the inlet residual hexane content, as reported in Fig. 5b. In particular the gure shows the optimized intermediate pressure for all the considered ejector systems
(see Fig. 2 for tags and equipment representation). These outcomes
might be potentially applied when a different feedstock quality is
accepted and processed by the renery for a mid- or long-term
period, with the need of a systematic improvement of the operating conditions. As shown in Fig. 5b, the increase of the residual
hexane content has a stronger inuence on the drying and bleaching sections respect to the deodorization, since in these sections
the major part of hexane is removed (see Section 3.1). This results
in the increase of the intermediate pressure for optimizing the motive steam consumption.
The results of the sensitivity analysis carried out by varying the
inlet hexane concentration and optimizing the operating conditions and process variables are reported in Table B1 of Appendix
B. The table allows determining the optimized operating conditions referring to the base case discussed in Section 3.1.
On the basis of the sensitivity analysis results, the overall utilities requirements were derived and shown in Fig. 6. Fig. 6a shows
the increase of the overall motive steam and cooling water consumption by varying the inlet hexane concentration of one order
of magnitude (e.g., ranging from 0.1% to 1.5% by weight basis). Motive steam consumption is increased by 40%, showing a more signicant variation respect to cooling water utility, which increase
is limited to 1%. This is due to the fact that the highest owrate
of cooling water is a xed simulation parameter, since it is fed to
the condenser of the third ejector (EJ3c, see detailed description
of simulation set up in Section 2.5.1). This owrate is almost
twenty times higher than the sum of the other cooling water utilities, which can be varied in order to control the condensate temperature (see Section 2.5.1).
In order to determine the variation in the process vents behavior due to the increase of inlet hexane concentration, Fig. 6b presents the change in the hexane removal percentage (thus,
starting from the values evaluated at 0.1% residual hexane content,
see Section 3.1) in each process section. The results highlight that
the excess hexane is mainly removed in the drying section, due to
the oversizing of the equipment. Hence, this allows decreasing the
hexane amount fed to the downstream units, which hexane removal decreases as shown in Fig. 6b.
Therefore, the sensitivity analysis allowed determining the
change in process parameters and utility requirements for restoring the process operating conditions given unforeseen changes of
the inlet feedstock. It clearly appears that the increase of volatile
solvent residual has a negative impact on the energetic costs of
the rening process.

848

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

3.3. Formation of ammable mixtures inside process streams

The process simulator pointed out the more critical nodes for
hexane accumulation, also considering the potential variation of
the initial hexane residual in the process feed. Among the possible hazards related to the presence of hexane inside process
pipes, one of the critical issues is related to the possibility of
air entrainment from gaskets and seals in strong vacuum operating pipes, thus leading to the formation of ammable mixtures
in conned spaces. This might lead to re and explosion hazards
in case of accidental ignition of the ammable mixture, already
highlighted for the storage equipment in a previous work (Landucci et al., 2011).
Therefore, the process simulator was employed to investigate
this problem, considering an additional air ow in the three vent
lines (V1, V2 and V3, see Fig. 4) given a reference air entrainment
value, specied by the ejector manufacturer (Krting Hannover
AG,1994) for the vent discharge line. Table 7 reports the considered
entrainment value (inltration type 2), also considering a possible
negative or positive variations respect to this reference value
(respectively, inltration types 1 and 3 in Table 7).
Fig. 7 reports the evaluated residual hexane concentration in
the vent lines evidencing the possibility of formation of ammable
mixtures in the drying (Fig. 7a), bleaching (Fig. 7b) and deodorization (Fig. 7c) sections as a function of the different air entrainment
rates given a xed hexane residual content in crude oil feed (e.g.,

0.1% by weight basis). A ammable mixture is potentially formed

if the calculated concentration point enters inside the ammable
range, i.e. the region of the chart included inside the reference continuous lines. In absence of data for water as inerting uid, the effect of nitrogen (bright lines in Fig. 7) and carbon dioxide (dark
lines in Fig. 7) as diluents was taken into account in order to obtain
preliminary indications for the methodology (Mashuga and Crowl,
1998; Zabetakis, 1965). Furthermore, the ammability range is affected by operative pressure and temperature, but the use of data
referred to 25 C temperature and 1.01 bar allows for evaluation of
the ammability hazards on the safe side in the considered process
sections (Lees, 1996).
The results make clear that in the case of higher hexane concentration in the vent line, the entrained air is not sufcient to form
ammable mixtures, thus leading to a less hazardous situation.
This is the case of the drying section, in which the major part of
hexane is removed and, as shown in Fig. 7a, and in which none
of the calculated points fall under the ammable region even for
high air entrainment rates. On the contrary, for the other two sections, the hexane vent content is lower and some points calculated
for high air entrainment rates especially in the deodorization section vent (see Fig. 7c), fall into the hazardous zone. This evidences a
safety criticality for strong vacuum operating equipment in presence of ammable vapors.
Hence, this type of hazard might be taken into account during
the vent pipeline design and in maintenance operations.

Table A1
Main parameters and equations implemented in the thermodynamic model (Honeywell, 2010b).
ID

Equation

Description

Parameters

Eq. (1)

RT
a
P Vb
 VVbbVb

PengRobinson state equation

Eq. (2)

Z3 - (1 - B)Z2 + (A - 2B - 3B2)Z - (AB - B2 - B3) = 0

PengRobinson expressed in terms

of the compressibility factor Z

Eq. (3)
Eq. (4)
Eq. (5)

A = aP/(RT)2
B = bP/(RT)2
P
RT c;i
b N
i1 xi bi ; bi 0:077796 Pc;i

Parameter in Eq. (2)

Parameter in Eq. (2)
1st PengRobinson equation
coefcient for mixtures

Eq. (6)

P = Pressure (Pa)
R = 8314 (J kmol1 K1) universal gas
constant
T = Temperature (K)
V = Volume (m3)
a = see Eq. (6)
b = see Eq. (5)
Z = Compressibility factor = (PV)/
(RT)
A = see Eq. (3)
B = see Eq. (4)
a = see Eq. (6)
b = see Eq. (5)
xi = mass fraction of the ith
component of the mixture of N
components.
Tc,i = critical temperature of the ith
component
Pc,i = critical pressure of the ith
component
Tr,i = T/Tc,i
kij = system specic experimental
binary interaction factor
mi = see Eq. (7)
xi = Acentric factor of the ith
component

PN PN
i1

2nd PengRobinson equation

coefcient for mixtures original
formulation

0:5
1  kij ; ai ac;i i
j1 xi xj ai aj
RT c;i 2
0:5
;
1 mi 1  T 0:5
r;i
i
Pc;i

ac;i 0:457235

Eq. (7)

mi 0:37464 1:5422xi  0:26992x2i ; xi 6 0:49

mi 0:379642 1:48503  0:164423  0:016666xi
xi xi ; xi > 0:49

Polynomial factor for Eq. (6)

original formulation

Eq. (8)

ai T Nr;ii =Mi 1 expLi 1  T Nr;ii Mi

Twu Alpha function for Peng

Robinson correction for Eq. (6)

Eq. (9)

HHID
RT

Eq. (10)

SSID
R

Eq. (11)


 V 20:5 1b
da
ln V 20:5 1b
Z  1  21:51bRT a  T dT
lnZ  b  lnP=P  

A
T da
21:5 bRT a dT

ln

V20:5 1b
V20:5 1b

Enthalpy equation



 



PN
bi
Pb
1
0:5
ln /i  ln Z  RT
Z  1 bbi  1:5a
2a0:5
j1 xj aj 1  kij  b 
i
2 bRT a


0:5
ln V 20:5 1b
V 2

1b

Entropy equation

Evaluation of fugacity coefcient

Li, Mi, Ni = Parameters of pure ith

substance (see details in Honeywell
(2010b))
H = predicted enthalpy
HID = reference enthalpy evaluated
at 25 C and 1.01 bar
S = predicted entropy
SID = reference entropy evaluated at
25 C and 1.01 bar
P pressure in the reference state
(1.01 bar)
/ = mixture fugacity coefcient of
for the ith component

849

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

4. Conclusions

Acknowledgement

In the present work a quantitative methodology was developed

for the performance analysis of the vegetable oil rening process.
An advanced thermodynamic model was implemented in order
to reproduce the vapor/liquid equilibrium of crude vegetable oil
residual solvent system. The model was validated against available
experimental data and was implemented in the rening process
simulator, developed on the Honeywell UniSim Design software.
The simulator allowed for a detailed performance analysis of
the process. The results were compared with eld data obtained
from an actual vegetable oil renery showing good agreement in
reproducing the rening process in the reference conditions.
The effect of the residual solvent content increase on the process efciency was investigated, determining the most signicant
nodes of solvent accumulation among the plant process operations
and evaluating its inuence on the global energy requirements. In
particular, the ejector systems, aimed at keeping vacuum operating
conditions, were deeply investigated, evaluating the utility consumption increment. Both motive steam and cooling water for condensers were analyzed by varying the residual hexane content in
the input crude oil and determining the modication in the operative conditions for minimizing the energy costs. The study evidenced the criticalities related to the management of inlet crude
oil quality, in terms of residual solvent content control, for the
enhancement of the global process efciency.
Finally, the simulator also allowed investigating the potential
hazards due to formation of ammable mixtures inside the process
vent lines, in presence of purged hexane vapors and air entrained
by gaskets and/or seals of vacuum operating pipelines. The results
evidenced the conditions in which ammable mixtures might
potentially be formed inside the process vents, with re and explosion hazards in presence of accidental ignition.

The authors gratefully acknowledge nancial support received

from Regione Toscana (Bando Unico R&S n.2009DUA/526090469/
1).
Appendix A
The present section provides details on the thermodynamic
model implemented in Unisim Design (Honeywell, 2010a,b).
The selected model is based on the PengRobinson equations (Peng
and Robinson, 1976) corrected with the Twu Alpha function (Twu
et al., 1995; Honeywell, 2010b), which takes into account the excess free energy in order to have more accurate prediction of vapor
pressure. Table A1 summarizes the key parameters and equations
used to predict enthalpy, entropy, the fugacity coefcients for each
component of the mixture and thus the vapor/liquid equilibrium.
Tables A2 and A3 report the specic parameters selected for each
substance considered in the present study.
Appendix B
Table B1 reports the results of the process optimization and
sensitivity analysis, comparing the baseline case results (BC) and
the optimized cases (OCs) by varying the residual hexane content
(HEX in the following, expressed in % by weight basis) up to one
order of magnitude respect to the BC, which features HEX = 0.1%.
The rst column of the table reports the process variable of
interest (EJ: ejector, MS: motive steam, CW: cooling water, see
Figs. 4 and 5). The second column report the results obtained for
the baseline case with HEX = 0.1%, while the third column shows
the correspondent optimization of process variables aimed at

Table A2
Main parameters selected for the present analysis (Honeywell, 2010b). For parameters denition see Table A1.
Parameter (see Table A1)

Equation (see Table A1)

Units (SI)

Assigned parameter for each component Unisim Design library

Triolein

Oleic acid

n-Hexane

n-C29H60

Sterols

Tocopherols

Water

Tc,i
Pc,i
Li
Mi
Ni
L0
M0
N0
L1
M1
N1

5
5
8
8
8
see
see
see
see
see
see
see

C
kPa

680.9
360.2
a
a
a
0.1253
0.9118
1.9482
0.5116
0.7841
2.8125
1.6862

496.9
1390
0.7760
0.8235
0.8235

234.7
3032
0.1363
0.8620
0.8620

564.9
826
0.3688
0.8247
0.8247

668.1
999.7
a
a
a
0.1253
0.9118
1.9482
0.5116
0.7841
2.8125
0.9863

646.7
945.9
a
a
a
0.1253
0.9118
1.9482
0.5116
0.7841
2.8125
0.9624

374.1
22,120
0.3831
0.8701
0.8701

xi

note
note
note
note
note
note
note

(a)
(a)
(a)
(a)
(a)
(a)
(a)

The parameters Li, Mi and Ni depend on individual compounds and were retrieved from UniSim Design library for the application of Eq. (8) of Table A1. Nevertheless, for
0
1
0
0
N0=M01
non-library compounds, the Twu alpha function can be estimated by the following expressions: ai ai T xi ai T  ai T where ai T r;i
1
N1=M11
expL01  T N0M0
; ai T r;i
expL11  T N1M1
; T r;i T=T c;i .
r;i
r;i
In this case, Table A2 reports the relevant parameters for the estimation of the Twu alpha function (L0, M0, N0, L1, M1, N1 and xi).

Table A3
Determination of system specic binary interaction factor ki,j (i: columns; j: rows) (see Eq. (11) in Table A1).
ki,j i ? j;

Triolein

Oleic acid

n-Hexane

n-C29H60

Sterols

Tocopherols

Water

Triolein
Oleic acid
n-Hexane
n-C29H60
Sterols
Tocopherols
Water

0
0.095
0
0
0
0

0
0
0
0
0

0.095
0

0.031
0
0
0.48

0
0
0.031

0
0
0.48

0
0
0
0

0
0

0
0
0
0
0

0
0
0.48
0.48
0
0

850

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

Table B1
Results of the sensitivity analysis. BC = base case; OC: optimized case; RHC: residual hexane content.
Process variable

EJ1a/b operative pressure (kPa)

EJ2a/b operative pressure (kPa)
EJ3a/b operative pressure (kPa)
EJ3c operative pressure (kPa)
MSA for EJ1a/b owrate (kg/h)
MSB for EJ1a/b owrate (kg/h)
MSA for EJ2a/b owrate (kg/h)
MSB for EJ2a/b owrate (kg/h)
MSA for EJ3a/b owrate (kg/h)
MSB for EJ3a/b owrate (kg/h)
MSA for EJ3c owrate (kg/h)
CW1&CW2 owrate (kg/h)
CW3&CW4 owrate (kg/h)
CW5&CW6 owrate (kg/h)
a

HEX = 0.1%

HEX = 0.5%

HEX = 1.0%
a

BC

OC

BC

OC

BC

OCa

16.5
16.5
2.8
20.0
70.1
53.4
15.6
27.6
1100.1
157.2
26.0
9282.2
1180.8
240,000.0

14.0
20.0
2.9
11.0
55.2
64.1
20.7
21.7
1138.3
55.0
50.3
9061.0
1316.0
240,000.0

16.5
16.5
2.8
20.0
90.4
177.1
18.8
47.5
1113.9
216.6
42.0
10,330.0
1375.0
240,000.0

25.5
22.5
3.0
12.0
165.7
66.0
28.6
32.7
1171.7
78.7
69.0
12,026.2
1664.2
240,000.0

16.5
16.5
2.8
20.0
115.2
342.3
22.2
68.7
1127.3
275.4
61.3
11,604.5
1637.4
240,000.0

26.0
24.0
3.0
12.0
223.1
63.0
38.4
42.5
1205.3
95.7
94.5
14,534.7
2099.6
240,000.0

Respect to the base case.

keeping the same operative condition in process equipment. The

other column of the table shows the results in case of higher
HEX values. In particular, the third column shows the variation
of the process variables able to restore the normal operative conditions in presence of HEX = 0.5%, while the fourth column reports
the correspondent optimized process variables and operative conditions. The same type of results are shown in the fth and sixth
column for HEX = 1.0%.
Appendix C. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jfoodeng.2013.
01.034.
References
Akterian, S., 2011. Improving the energy efciency of traditional multi-stage steamjet-ejector vacuum systems for deodorizing edible oils. 11th International
Congress on Engineering and Food (ICEF11). Procedia Food Science 1, 1785
1791.
Bhosle, B.M., Subramanian, R., 2005. New approaches in deacidication of edible
oils a review. Journal of Food Engineering 69 (4), 481494.
Biegler, L.T., Grossmann, I.E., Westerberg, A.W., 1997. Systematic Methods of
Chemical Process Design. Prentice Hall, Upper Saddle River.
Bockisch, M., 1998. Fats and Oils Handbook. AOCS Press, Champaign (IL).
Calliauw, G., Vila Ayala, J., Gibon, V., Wouters, J., De Greyt, W., Foubert, I.,
Dewettinck, K., 2008. Models for FFA-removal and changes in phase behavior of
cocoa butter by packed column steam rening. Journal of Food Engineering 89
(3), 274284.
Carmona, M.A., Jimnez, C., Jimnez-Sanchidrin, C., Pea, F., Ruiz, J.R., 2010.
Isolation of sterols from sunower oil deodorizer distillate. Journal of Food
Engineering 101 (2), 210213.
Cavanagh, G.C., 1976. Miscella rening. Journal of the American Oil Chemists
Society 53 (6), 361363.
Ceriani, R., Meirelles, A.J.A., 2004. Predicting vaporliquid equilibria of fatty
systems. Fluid Phase Equilibria 215 (3), 227236.
Ceriani, R., Meirelles, A.J.A., 2006. Simulation of continuous physical reners for
edible oil deacidication. Journal of Food Engineering 76 (3), 261271.
Ceriani, R., Gani, R., Liu, Y.A., in press. Prediction of vapor pressure and heats of
vaporization of edible oil/fat compounds by group contribution, Fluid Phase
Equilibria, doi:10.1016/j.uid.2012.09.039.
Cerutti, M.L.M.N., Ulson de Souza, A.A., Ulson de Souza, Ulson., 2012. Solvent
extraction of vegetable oils: numerical and experimental study. Food and
Bioproducts Processing 90 (2), 199204.
Chiyoda, C., Peixoto, E.C.D., Meirelles, A.J.A., Rodrigues, C.E.C., 2010. Liquidliquid
equilibria for systems composed of rened soybean oil, free fatty acids, ethanol,
and water at different temperatures. Fluid Phase Equilibria 299 (1), 141147.
Christov, M., Dohrn, R., 2002. High-pressure uid phase equilibria-experimental
methods and systems investigated (19941999). Fluid Phase Equilibria 202 (1),
153218.
Cuevas, M.S., Rodrigues, C.E.C., Meirelles, A.J.A., 2009. Effect of solvent hydration
and temperature in the deacidication process of sunower oil using ethanol.
Journal of Food Engineering 95 (2), 291297.

Didi, M.A., Makhoukhi, B., Azzouz, A., Villemin, D., 2009. Colza oil bleaching through
optimized acid activation of bentonite. A comparative study. Applied Clay
Science 42 (34), 336344.
Dunford, N.T., Zhang, M., 2003. Pressurized solvent extraction of wheat germ oil.
Food Research International 36 (910), 905909.
El Economista, January, 19th 2007, <http://www.eleconomista.es/economia/
noticias/131964/01/07/Muere-un-obrero-y-otro-herido-grave-en-explosionen-tanque-aceite-en-Sevilla.html>.
FAO Food and Agriculture Organization, 2011. The State of Food and Agriculture
20102011. FAO, Rome (I).
Farhoosh, R., Einafshar, S., Sharayei, P., 2009. The effect of commercial rening steps
on the rancidity measures of soybean and canola oils. Food Chemistry 115 (3),
933938.
Fornari, T., Bottini, S., Brignole, A.E., 1994. Application of UNIFAC to vegetable oilalkane mixture. Journal of the American Oil Chemists Society 71 (4), 391395.
Gerasimenko, E.O., Turyan, Ya.I., 2012. Automated ow pH-method for the
determination of total free fatty acids content in edible oils. Food Chemistry
132 (3), 15621565.
Gunstone, F.D., Harwood, J.L., Padley, F.B., 1994. The Lipid Handbook, second ed.
Chapman & Hall, London, pp. 265279.
Harwardt, A., Kossack, S., Marquardt, W., 2008. Optimal column sequencing for
multicomponent mixtures. Computer Aided Chemical Engineering 25, 9196.
Haslenda, H., Jamaludin, M.Z., 2011. Industry to industry by-products exchange
network towards zero waste in palm oil rening processes. Resources,
Conservation and Recycling 55 (7), 713718.
Honeywell, 2010a. UniSim Design Design User Guide. Honeywell, London,
Ontario.
Honeywell, 2010b. UniSim Design Design Simulation Basis Reference Guide.
Honeywell, London, Ontario.
Hoyer, M., Schumann, R., Premier, G.C., 2005. An approach for integrating process
and control simulation into the plant engineering process. Computer Aided
Chemical Engineering 20, 16031608.
Keurentjes, J.T.F., Doornbusch, G.I., Vant Riet, K., 1991. The removal of fatty acids
from edible oil. Removal of the dispersed phase of a water-in-oil dispersion by a
hydrophilic membrane. Separation Science and Technology 26 (3), 409423.
Krting Hannover AG, 1994. Reference Data for Application of Jet Ejectors and
Vacuum Processing. Krting Hannover AG, Hannover.
La Repubblica, 2006. Section: Perugia local news. Gruppo Editoriale LEspresso,
Rome. November, 26th 2006 p. 6.
Landucci, G., Nucci, B., Pelagagge, L., Nicolella, C., 2011. Hazard assessment of edible
oil rening: formation of ammable mixtures in storage tanks. Journal of Food
Engineering 105 (1), 105111.
Lees, F.P., 1996. Loss Prevention in the Process Industries, second ed. ButterworthHeinemann, Oxford.
Loft, S.C., 1990. Deodorization theory and practice. In: Erickson, D.R. (Ed.), Edible
Fats and Oils Processing: Basic Principles and Modern Practices. The American
Oil Chemists Society, Boulder Urbana, pp. 117123.
Mag, T.K., 1990. Bleaching theory and practice. In: Erickson, D.R. (Ed.), Edible Fats
and Oils Processing: Basic Principles and Modern Practices. The American Oil
Chemists Society, Boulder Urbana, pp. 107116.
Martinello, M., Hecker, G., Pramparo, M.d.C., 2007. Grape seed oil deacidication by
molecular distillation: analysis of operative variables inuence using the
response surface methodology. Journal of Food Engineering 81 (1), 6064.
Mashuga, C.V., Crowl, D.A., 1998. Application of the ammability diagram for
evaluation of re and explosion hazards of ammable vapors. Process Safety
Progress 17 (3), 176183.
Meherwan, P.B., 1999. Transport and storage of uids, section 10. Perrys Chemical
Engineers Handbook, seventh ed. McGraw Hill, New York.
Midwest Research Institute MRI, 1995. Emission factor documentation for AP-42
(section 9.11.1) Vegetable oil processing nal report. Midwest Research

G. Landucci et al. / Journal of Food Engineering 116 (2013) 840851

Institute, Ofce of Air Quality Planning and Standards (OAQPS), US
Environmental Protection Agency (EPA), Research Triangle Park.
Mielke, T., 1990. Current world supply, demand and price outlook for oils and fats.
In: Erickson, D.R. (Ed.), Edible Fats and Oils Processing: Basic Principles and
Modern Practices. The American Oil Chemists Society, Boulder Urbana, pp. 19.
Motard, R.L., Shacham, M., Rosen, E.E., 1975. Steady state chemical process
simulation. American Institute of Chemical Engineers AIChE Journal 21 (3),
417436.
Muth, M.K., Depro, B.M., Domanico, J.L., 1998. Vegetable Oil Production: Industry
Prole. Research Triangle Institute, Center for Economics Research, NC.
NFPA National Fire Protection Association, 2007. NFPA 68: standard on explosion
protection by deagration venting. National Fire Protection Association, Quincy.
Peng, D.Y., Robinson, D.B., 1976. A two constant equation of state. Industrial
Engineering Chemical Fundamentals 15 (1), 5964.
Rodrigues, C.E.C., Filho, P.A.P., Meirelles, A.J.A., 2004. Phase equilibrium for the
system rice bran oil + fatty acids + ethanol + water + c-oryzanol + tocols. Fluid
Phase Equilibria 216 (2), 271283.
Ruiz-Mendez, M.V., Dobarganes, M.C., 2007. Combination of chromatographic
techniques for the analysis of complex deodoriser distillates from an edible oil
rening process. Food Chemistry 103 (4), 15021507.
Sampaio, K.A., Ceriani, R., Silva, S.M., Tahama, T., Meirelles, A.J.A., 2011. Steam
deacidication of palm oil. Food and Bioproducts Processing 89 (4), 383390.
Santori, G., Di Nicola, G., Moglie, M., Polonara, F., 2012. A review analyzing the
industrial biodiesel production practice starting from vegetable oil rening.
Applied Energy 92, 109132.
Shahidi, F., 2005. Edible oil and fat products: edible oils, . 6th ed. Baileys Industrial
Oil and Fat Products 6th ed., vol. 2 Wiley & Sons, New York.
Shaw, J.F.G., 1992. The integration of process simulation and engineering design. In:
Proceedings of the European symposium on computer aided process
engineering (ESCAPE-1) No. 1. Elsevier, Kidlington. (pp. S465S472)
Shell Chemicals Europe B.V., 2012. Hexane (extraction grade) material safety
datasheet (version 3.2). Shell Chemicals Europe B.V., Rotterdam.
Silva, S.M., Rocco, S.A., Sampaio, K.A., Taham, T., Meller da Silva, L.H., Ceriani, R.,
Meirelles, A.J.A., 2011. Validation of a method for simultaneous quantication of
total carotenes and tocols in vegetable oils by HPLC. Food Chemistry 129 (4),
18741881.
Smith, A.S., Florence, B., 1951. Vapor pressure of hexanesoybean oil solutions at
high solvent concentrations. Journal of the American Oil Chemists Society 28
(8), 360361.
Smith, A.S., Wechter, F.J., 1950. Vapor pressure of hexanesoybean oil solutions at
low solvent concentrations. Journal of the American Oil Chemists Society 27
(10), 381383.

851

Sullivan, F.E., 1976. Steam Rening. Journal of the American Oil Chemists Society 53
(6), 358360.
Szabo, L., Nemeth, S., Szeifert, F., 2011. Analysis of separation possibilities of
multicomponent mixtures. Computer Aided Chemical Engineering 29, 341345.
Szydowska-Czerniak, A., Trokowski, K., Karlovits, G., Szyk, E., 2011. Effect of
rening processes on antioxidant capacity, total contents of phenolics and
carotenoids in palm oils. Food Chemistry 129 (3), 11871192.
Teles dos Santos, M., Gerbaud, V., Le Roux, G.A.C., in press. Modeling and simulation
of melting curves and chemical interesterication of binary blends of vegetable
oils, Chemical Engineering Science. doi: 10.1016/j.ces.2012.09.026.
Towler, G., Sinnott, R., 2013. Process simulation,, .. Chemical Engineering Design,
second ed. Elsevier, Amsterdam. http://dx.doi.org/10.1016/B978-0-08-0966595.00004-3 (Chapter 4).
Tugnoli, A., Landucci, G., Salzano, E., Cozzani, V., 2012. Supporting the selection of
process and plant design options by Inherent Safety KPIs. Journal of Loss
Prevention in the Process Industries 25 (5), 830842.
Twu, C.H., Coon, J.E., Cunningham, J.R., 1995. A new generalized alpha function for a
cubic equation of state: Part 1. PengRobinson equation. Fluid Phase Equilibria
105, 4959.
Vadapalli, A., Seader, J.D., 2001. A generalized framework for computing bifurcation
diagrams using process simulation programs. Computers and Chemical
Engineering 25 (23), 445464.
Veloso, G.O., Kriouko, V.G., Vielmo, H.A., 2005. Mathematical modeling of vegetable
oil extraction in a counter-current crossed ow horizontal extractor. Journal of
Food Engineering 66 (44), 477486.
Wills, A.G., Heath, W.P., 2005. Application of barrier function based model
predictive control to an edible oil rening process. Journal of Process Control
15 (2), 183200.
Zabetakis, M.G., 1965. Flammability characteristics of combustible gases and
vapors. US dept. of the Interior, Bureau of Mines, Washington.
Zhang, Y., Dub, M.A., McLean, D.D., Kates, M., 2003. Biodiesel production from
waste cooking oil: 1. Process design and technological assessment. Bioresource
Technology 89 (1), 116.
Zin, R.B.M., 2006. Process design in degumming and bleaching of palm oil. Research
vote no. 74198, Centre of lipids engineering and applied research (CLEAR),
University of Technology, Malaysia, Johor Bahru.
Zulkurnain, M., Lai, O.M., Latip, R.A., Nehdi, I.A., Ling, T.C., Tan, C.P., 2012. The effects
of physical rening on the formation of 3-monochloropropane-1,2-diol esters in
relation to palm oil minor components. Food Chemistry 135 (2), 799805.