ww.sciencedirect.com

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4

Available online at w

ScienceDirect

journal homepage: www.elsev ier .com/locate/ issn/15375110

Research Paper

Olive oil quality improvement using a naturalsedimentation plant at industrial scale

Giuseppe Altieri a,*, Giovanni C. Di Renzo a, Francesco Genovese a,Antonella Tauriello a, Maurizio D’Auria b, Rocco Racioppi b, Licia Viggiani b

a Scuola di Scienze Agrarie, Forestali, Alimentari e Ambientali, Universita degli Studi della Basilicata,

Viale dell’Ateneo Lucano, 10, 85100 Potenza, ItalybDipartimento di Scienze, Universita degli Studi della Basilicata, Viale dell’Ateneo Lucano 10, 85100 Potenza, Italy

a r t i c l e i n f o

Article history:

Received 2 December 2013

Received in revised form

27 March 2014

Accepted 6 April 2014

Published online

Keywords:

Sedimentation plant

Olive oil quality

Feedback control

* Corresponding author.E-mail address: giuseppe.altieri@unibas.i

http://dx.doi.org/10.1016/j.biosystemseng.201537-5110/ª 2014 IAgrE. Published by Elsevie

Olive oil extraction is mainly carried out using continuous extraction by decanter centri-

fuge with efficiency of approximately 80e90%. After centrifugal extraction, olive oil is

generally cleaned using a vertical disc stack centrifuge separator, which is suspected of

being the major cause of decreased final olive oil quality. Experiments were carried out at

industrial scale to compare the olive oil properties after improved processes of sedimen-

tation (Sedoil) or centrifugation (Cenoil) with respect to raw olive oil obtained at the decanter

exit (Control). Peroxide, polyphenol, chlorophyll, carotenoid, turbidity and K232 average

values were significantly different between Sedoil and Cenoil, which confirmed that the use

of disc stack centrifuges represents an important source of oxidative reactions. Analysis

showed that storage time dramatically affects the oxidation level of the olive oil. All pa-

rameters used to monitor the oxidation level (i.e., free acidity, peroxide value and K232)

increased after 180 d of storage, and the content of natural antioxidants and pigments

decreased as expected. The residual presence of water during long-term storage repre-

sented the most important source of oxidation, and an effective cleaning operation is

necessary to preserve oil quality during its storage life. The analyses performed using 1H

and 13C NMR showed that Sedoil was more similar in composition to Control than to Cenoil.

The use of sedimentation plant allows the employment of the disk stack centrifuge to be

reduced improving both energy saving and the quality of clean olive oil.

ª 2014 IAgrE. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Olive oil is a vegetable oil obtained directly from olive fruits

(Olea europaea) by mechanical extraction. It is essential to the

“Mediterranean diet” and is rich in triacylglycerol, glyceridic

compounds and polyphenols.

t (G. Altieri).14.04.007r Ltd. All rights reserved

At present, mainly due to economic pressures, the evolu-

tion of the oil extraction process has led to the replacement of

traditional discontinuous pressure systems with continuous

centrifugal extraction. Themechanical oil extraction ismainly

carried out using a continuous process based on centrifuga-

tion by decanter centrifuge (a horizontal centrifugewith screw

conveyor and rotating bowl), and it has an extraction

.

Nomenclature

NMR nuclear magnetic resonance1H NMR hydrogen-1 nuclear magnetic resonance13C NMR carbon-13 nuclear magnetic resonance

PID proportional integral derivative

VFD variable frequency driver

FWER familywise error rate

FA free acidity (g (100 g)�1)

PV peroxide value (meq kg�1)

K232 or K232 specific extinction coefficient at 232 nm

K270 or K270 specific extinction coefficient at 270 nm

Vph volume (ml) of solution titrated with potassium

hydroxide

c concentration (mol l�1) of the solution of

potassium hydroxide

M molar weight (g mol�1) of oleic acid (¼ 282)

m mass of sample of olive oil (g)

Vts volume of the sodium thiosulphate solution

(ml)

T normality of the sodium thiosulphate solution

(N)

CHLO chlorophyll content (mg kg�1)

CAR carotenoid content (mg kg�1)

UV ultraviolet

E0 coefficient for specific extinction

A670 absorbance at 670 nm

A470 absorbance at 470 nm

A absorbance

d optical path length

POLYPH total polyphenol content (mg l�1)

TUR turbidity

PCA principal component analysis

PC1 first axis of the principal component analysis

PC2 second axis of the principal component

analysis

AVGDIST “mean Euclidean distance” between paired

samples after principal component analysis

H0 null hypothesis

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4100

efficiency of approximately 80e90% of the oil contained in the

fruit without any addition of other materials. The decanter is

widely used in olive oil extraction and allows the processing of

large amounts of olives in a short amount of time (Catalano,

Pipitone, Calafatello, & Leone, 2003; Piacquadio, De Stefano,

& Sciancalepore, 1998; Ranalli, De Mattia, & Ferrante, 1997).

Changes in olive paste rheological characteristics related to

its water content, temperature, fruit variety and maturity level

dramatically affect the extraction process quantitatively and

even qualitatively (Boncinelli, Catalano, & Cini, 2013; Di

Giovacchino, Sestili, & Di Vincenzo, 2002; Di Renzo & Colelli,

1997). Furthermore, the addition of lukewarm water to the

olive paste during malaxation or extraction in the continuous

extraction process by decanter centrifuge improves the sepa-

ration of oil from vegetable water and husks (Amirante, Cini,

Montel,&Pasqualone, 2001;Amirante,DiRenzo,&Colelli, 1995).

The final quality and extraction yield of olive oil are strictly

related to bothmachine characteristics (e.g., type of crusher or

decanter geometry) and processing parameters (e.g., the

amount of added clean water, malaxing time and tempera-

ture, and amount of oxygen dissolved into the oil during the

process) (Boskou, 2006; Del Caro, Vacca, Poiana, Fenu, & Piga,

2006; Di Giovacchino, Costantini, Ferrante, & Serraiocco, 2002;

Ranalli, Cabras, Iannucci, & Contento, 2001; Salvador, Aranda,

Gomez-Alonso, & Fregapane, 2003; ).

Therefore, a correct control of the extraction process is

essential to achieve both high extraction yields and olive oil

quality (Altieri, 2010; Altieri, Di Renzo, & Genovese, 2013;

Boncinelli, Daou, Cini, & Catalano, 2009; Furferi, Carfagni, &

Daou, 2007; Jimenez, Beltran, Aguilera, & Uceda, 2008). In

particular, the control of temperature and oxygen concen-

tration during the malaxation phase has been demonstrated

to have advantageous effects on both the extraction yield and

the quality of the extracted olive oil (Aiello et al., 2012;

Angerosa, Mostallino, Basti, & Vito, 2001; Catania et al., 2013;

Servili, Selvaggini, Taticchi, Esposto, & Montedoro, 2003).

Indeed the malaxing phase decreases the apparent viscosity

of olive paste and enhances the enzymes activity, conse-

quently affecting the volatile compounds, the phenolic

composition and sensory quality of olive oil (Kalua et al., 2007).

The initial volatile and non-volatile compounds composition

establishes the olive oil shelf-life, which indeed is determined

by the lipid oxidation reaction rate that cause the olive oil

quality decrease at the expenses of the antioxidant content in

the product (Gomez-Alonso, Mancebo-Campos, Salvador, &

Fregapane, 2007).

The process consists of four phases:

1. Rupture of the olive fruits, which is carried out by several

types of crushers. This operation produces a fluid

composed of a mixture of two distinct liquid phases (raw

oil and water) and an extremely heterogeneous solid phase

(pit, skin, and pulp fragments);

2. Kneading of the paste to facilitate cohesion of smaller oil

droplets into larger droplets that are easier to separate and

to complete the tissue rupture using the action of the pit

fragments;

3. Centrifugation to separate the different phases. This pro-

cess occurs in a horizontal screw conveyor centrifuge with

continuous discharge of the solid phase (i.e., a decanter);

4. Cleaning of the olive oil.

Final cleaning of the olive oil is required because the

extracted oil still contains a certain amount of residual water

and impurities. This further cleaning step is generally per-

formed by a vertical disc stack centrifuge separator. Consider

that there is a great interest, from a commercial point of view

and on the basis of market requests, in minimally processed

(raw) olive oil that does not shine because it is considered by

consumers to be closer, in terms of composition, to the orig-

inal oil present in the olive fruit (Boskou, 2006; Tsimidou,

Georgiou, Koidis, & Boskou, 2005). Furthermore, the latest

developments in decanter centrifuge design, modelling and

construction (Altieri, 2010; Altieri et al., 2013; Amirante &

Catalano, 2000, 1993; Boncinelli et al., 2009; Catalano et al.,

2003; Daou, Furferi, Recchia, & Cini, 2007) allow the

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 101

extraction of oil that is much more clean than in the past in

terms of both suspended solids and residual water content.

An aspect that has not been investigated in depth is the

influence of the vertical disc stack centrifuge on the olive oil

quality. Indeed the use of vertical disc stack centrifuge sepa-

rator, which is the most efficient system for final olive oil

cleaning, is suspected to be responsible for negative effects on

the olive oil quality (e.g., loss of aroma) and loss of stability of

the final product (due to oxidative reactions) (Di Giovacchino,

Solinas, &Miccoli, 1994; Masella, Parenti, Spugnoli, & Calamai,

2009) mainly due to oil heating and to the increased amount of

dissolved oxygen (Masella, Parenti, Spugnoli, & Calamai, 2012;

Parenti, Spugnoli, Masella, & Calamai, 2007). In this paper the

olive oil properties were analysed in order to define the best

operating parameters of a natural sedimentation prototype

plant allowing the olive oil to separate with higher quality in

comparison with traditional centrifugal-separated oil and raw

oil (extracted oil without cleaning operation).

To this purpose, both chemical and spectroscopic charac-

terisations of the olive oil were performed. In particular, NMR

spectroscopy (and particularly 1H NMR) was used on olive oil

to determine its fatty acid composition, the presence of vol-

atile organic compounds (Mannina et al., 1999; Sacchi et al.,

1995, 1998) and to evaluate minor components such as alde-

hydes and sterols, which are very useful to determine the

geographical origin of olive oils (Mannina et al., 1999; Sacchi

et al., 1998). Furthermore, the 13C NMR technique provides

information about the fatty acid composition and the acyl

positional distribution (1,3-acyl and 2-acyl) of glycerol tri-

esters in olive oil samples (Mannina, Fontanazza, Patumi,

Ansanelli, & Segre, 2001).

2. Materials and methods

An automatic feedback control systemwas used to control the

mass flow rate of the olive paste as it was fed to the decanter

centrifuge (Altieri et al., 2013). The mass flow rate delivered by

the paste pump varies due to hydraulic pressure variations in

the malaxer unit during olive paste discharge (Altieri et al.,

2013). The level of olive paste in the malaxer unit affects the

hydraulic pressure as the olive paste is fed into the pump,

Fig. 1 e Layout of the experimental apparatus and the feedback

supply pump. The setup allows the regulation of olive paste mas

which leads to a change in pump hydraulic efficiency and

represents, together with variations in olive paste tempera-

ture and viscosity, additional interference during the extrac-

tion process by decanter centrifuge. Therefore, using the

feedback control system to regulate themass flow rate of olive

paste makes it possible to operate under constant conditions

without significant variations in paste mass flow rates,

avoiding any change in the extracted oil quality.

The control system was based on a standard software

proportional integral derivative (PID) feedback control system

(see Fig. 1) consisting of the following elements:

- a notebook computer with data acquisition board DAQCard

AI-16X-E50 (National Instruments Corporation, Austin,

Texas, USA);

- in-housemanagement software built using LabView� 6.0.2

(National Instruments Corporation), which constitutes the

feedback control software;

- Corimass G300þ mass flow rate sensor (Krohne Mes-

stechnik GmbH,Duisburg, Germany) connected through an

RS485 slave serial interface (Modbus protocol);

- olive paste pump (Moineau type pump) coupled with a

three-phase motor (3 kW of mechanical power);

- variable-frequency drive (VFD) (3 kW, Siemens Master

Vector, Siemens AG, Munich, Germany) connected through

an RS485 slave serial interface (USS protocol).

The feedback control system allowed regulation of the

olive paste mass flow rate by varying the rotational speed of

the olive paste pump (Moineau type pump) through the three-

phase motor using the VFD.

The feedback control system was used in olive oil milling

operations to regulate the olive paste flow rate fed to a triple-

phase decanter centrifuge at a constant value of 1800 kg h�1;

theadditional lukewarmwatermassflowratewasalsoheldat a

constant value (900 kg h�1) in order to achieve a dilution coef-

ficient of 0.50 that is recommend for this type of decanter

(Boskou, 2006) related to the water content of the olive paste.

The triple-phasedecantercentrifugeemployedfor the trialshad

amaximumworkingcapacityof 2000kgh�1 of processedolives.

The tests were carried out over five non-consecutive days

with one repetition per day of approximately 12,500 kg of

control system coupled with the three-phase motor of the

s flow rate as the material is fed to the decanter centrifuge.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4102

homogeneous olive drupes (O. europaea L., cv. Coratina) each.

This timing was created to ensure the capture of maximum

variability in the olive samples. Each day, five batches of

approximately 2500 kg of olives were processed, and sampling

of the extracted oil took place with the fourth batch of the day.

The oil samples were extracted from the olives and

collected after the following processing operations (treat-

ments) (see Fig. 1):

- Extraction by centrifuge and direct collection upon

decanter exit (Control);

- Improved natural settling (sedimentation) performed in a

prototype plant (Sedoil);

- Centrifugal separation performed in a vertical disc stack

centrifuge, which represents the traditional oil cleaning

operation (Cenoil).

After undergoing the various treatments, samples were

decanted for 48 h to simulate the usual decanting operation

performed in olive oil processing plants and analysed. Other

samples were stored and maintained in a sealed container at

18 �C for 180 d in a dark room before being analysed.

Five olive oil samples were analysed in triplicate for each

treatment; Sedoil and Cenoil samples were evaluated with

reference to the respective Control samples. Outliers were

removed on the basis of the data’s interquartile range. The

statistical analysis has been carried out using three Wilcoxon

signed rank test, one-sided (left and right tail) and double-

Fig. 2 e Schematic diagram of th

sided, on the differences of the paired samples in order to

fully assess the existence of any significantly difference be-

tween the treatments. The familywise error rate (FWER) was

set to 10% because of the great variability of vegetable prod-

ucts even when produced in the same orchard. The statistical

analysis was carried out using Matlab� software (Matlab

R2013a, The MathWorks Inc., Natick, Massachusetts, USA).

2.1. The prototype processing plant used

To define the design criteria for the innovative plant proto-

type, experiments were carried out using a laboratory pilot

plant consisting of a twin cylindrical oil-water impurity

separator (two cylindrical columns connected in series) based

on a gravity separating system (see Fig. 2). Each cylindrical

column (2 m high, 0.27 m internal diameter, and 0.29 m

external diameter and made of a transparent material (poly

methyl-methacrylate) that is chemically compatible with

olive oil) has a total volume of approximately 114 L.

The laboratory plant was equipped with 4 volumetric

pumps (2 for each column) with flexible impellers and one

pump to fed the oily must (overall installed power of 1 kW);

each is controlled by its own variable-frequency drive (VFD) in

order to allow the user to balance the inlet flow rate with the

outlet flow rate for both clean oil and separated background

material. During the operations the plant prototype was

covered by panels to prevent the light influence on the pro-

cessed olive oil.

e experimental apparatus.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 103

2.2. Oil cleaning process

For the Sedoil treatment, raw oil was obtained by decanter

separation and was fed into the first column at a constant rate

of approximately 1 lmin�1. After approximately 90min, and in

consideration of the internal volume of the column, the raw

oil separates into a background at the bottom of the column

that is composed of high-density suspended solids and re-

sidual water and cleaned oil and low-density suspended

solids, which float to the surface of the cylinder. Cleaned oil

was extracted by the second pump very close to the surface

level (40 mm under the free surface level) and transferred into

the second column at a constant rate equal to the inlet rate.

Maintaining as constant the oil inlet rate and the outlet rates

in the two columns, approximately 90% of the material was

clean oil after a settling time of approximately 180 min while

the residual dirty oil mixed with water and settled materials

were treated in a centrifugal separator.

For the Cenoil treatment, raw oil was obtained by decanter

separation and was fed into a vertical disc stack centrifuge

separator (motor power of 5.5 kWwith automatic intermittent

discharge of solids, 6300 rpm, and throughput capacity of

2700 l h�1).

2.3. Analytical methods

Free acidity (FA), peroxide value (PV), and UV-specific extinc-

tion coefficients (K232 and K270) were determined according to

the analytical methods of the European Official Method of

Analysis (EU Regulations 2568/91). FA was determined by

titration of a 1:2 (v/v) ratio mixture of oil in ethanol/ether with

potassium hydroxide at 0.1 N. The results expressed as a per-

centage of oleic acid were found with the following formula:

FA ¼ Vph� c�M=ð10�mÞwhere Vph is the volume (ml) of solution titrated with potas-

siumhydroxide; c is the concentration (mol l�1) of the solution

of potassium hydroxide; M is the molar weight (g mol�1) of

oleic acid (¼ 282) and m is the mass of sample of olive oil (g).

PVwas determined as follows: a 2:3 (v/v) ratiomixture of oil

and chloroform/acetic acid was left to react in darkness with a

saturated potassium iodine solution; the released free iodine

was titrated with a sodium thiosulphate solution (0.01 N). PV

was expressed in milliequivalents of active oxygen per kilo-

gram of oil (meq kg�1) using the following formula:

PV ¼ ðVts� T=mÞ � 1000

where Vts is the volume of the sodium thiosulphate solution

(ml); T is the normality of the sodium thiosulphate solution (N)

and m is the mass of the sample of olive oil (g).

K232 and K270 extinction coefficients were calculated from

absorption of oil samples at 232 nm (K232) and 270 nm (K270),

respectively, with a spectrophotometer (UV/Vis Spectropho-

tometer, Ultrospec 2100 Pro, Biochrom Ltd., Cambridge, En-

gland) using isooctane as a blank.

Chlorophyll (CHLO) and carotenoid (CAR) concentrations

were determined by measuring the absorbance at fixed

wavelengths of 670 nm and 470 nm, respectively, in hexane

using spectroscopy. The maximum absorption at 670 nm is

related to the chlorophyll fraction while the maximum ab-

sorption at 470 nm is related to the carotenoid fraction. The

values of the applied coefficients for specific extinction were

E0¼ 613 for pheophytin, amajor component in the chlorophyll

fraction, and E0 ¼ 2000 for lutein, a major component in the

carotenoid fraction. Thus, the pigment contents were calcu-

lated as follows:

Chlorophyll�mg kg�1� ¼ �

A670 � 106�=ð613� 100� dÞ

Carotenoid�mg kg�1� ¼ �

A470 � 106�=ð2000� 100� dÞ

where A is the absorbance and d is the optical path length

(10 mm). CHLO and CAR contents are expressed as milligrams

of “pheophytin a” or “lutein” per kilogram of oil, respectively.

Total polyphenol (POLYPH) contents were evaluated using

a spectrophotometric method. Phenolic compounds were

isolated by a double extraction of oil (10 g) with a meth-

anolewater mixture (80:20 v/v). The FolineCiocalteau reagent

was added to a suitable aliquot of the extract, and the ab-

sorption of the solution was measured at 765 nm. Gallic acid

standard solutions were used to calibrate the method (at

linear concentration intervals of 15e500 mg/l). The results are

expressed as mg l�1 of gallic acid.

Turbidity measurements (TUR) were carried out by the

nephelometric method with a turbidity meter (Delta Ohm

HD25.2, Delta Ohm S.r.L., Caselle di Selvazzano, Padova, Italy).

The turbidity meter used in the study was carefully calibrated

with formazine standards (0.05e800 NTU).

2.4. 1H NMR analysis

Olive oil samples (20 ml) were placed into 5mmNMR tubes and

dissolved in chloroform-d (700 ml). The NMR spectra were

recordedwith a Varian Inova 500 instrument (Varian Inc., Palo

Alto, California, USA) operating at 499.60 MHz. 1H NMR FIDs

were recorded using the following acquisition parameters:

32 K acquired points, 32 K processed points, 14 ppm spectral

width, 2 s relaxation delay, p/2 pulse, 1.5 s acquisition time,

and 4000 scans. The intensities of the selected resonances

were compared with those at 1.55 ppm normalised to 1000.

The selected 1H resonances are b-sytosterol (0.62 ppm),

squalene (1.62 ppm), four resonances due to terpenes in the

spectral range of 4.45e5.00 ppm, formaldehyde (8.00 ppm),

trans-2-hexenal (9.45 ppm), hexanal (9.70 ppm) and two other

unsaturated aldehydes (9.53 ppm and 9.61 ppm).

2.5. 13C NMR analysis

Olive oil samples (100 ml) were placed in 5 mmNMR tubes and

dissolved in chloroform-d (600 ml). 13C spectrawere recorded at

300 K on a Varian Inova 500 instrument operating at

125.62 MHz using the following acquisition parameters: 256 K

acquired points, 128 K processed points, 195 ppm spectral

width, digital resolution of 0.22 Hz per point, and 8 s relaxation

delay. The intensities of the following carbonyl signals were

measured: sn 1,3 (173.271 ppm) palmitic and stearic chains; sn

1,3 (173.261 ppm) eicosenoic and cis-vaccenic chains; sn 1,3

(173.241 ppm) and sn 2 (172.833 ppm) oleic chains; sn 1,3

(173.230 ppm); and sn 2 (172.821 ppm) linoleic chains. These

ent

ep-

�0.41

1.00

0.16

1.00

0.03*

0.03*

0.03*

1.00

0.84

0.06*

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4104

intensities were normalised by dividing each of them by their

total sum.

Table

1e

Analysisofoliveoilafter48h.Meanvalueswithstandard

deviationforeach

treatm

ent.ForSed

oilandCen

oil,in

parenth

esisis

calculatedth

erelativeperc

difference

withresp

ect

toth

eCon

trol’s

meanvalue.ThreeW

ilco

xonsignedra

nktest,one-sided(left

andrighttail)anddouble-sided,havebeenca

rriedoutonth

edifference

softh

epairedsa

mplesin

ord

erto

fullyass

ess

theexistence

ofanysignifica

ntlydifference

betw

eenth

etreatm

ents.A

10%

fam

ilywiseerrorra

tewasuse

d,th

valuem

ark

edwithanasterisk

meanssignifica

ntlydifference

forth

etest

andth

erejectionofth

erelatednullhypoth

esisH0.

Parameter

Con

trol

Sed

oil

Cen

oil

H0:Con

trol

vs.

Sed

oilp-value

H0:Con

trol

vs.

Cen

oilp-value

H0:Sed

oilvs.

Cen

oilp-value

�¼

��

¼�

�¼

Waterco

ntent[%

]0.28�

0.15

0.08�

0.01(�

70.42%)

0.08�

0.02(�

71.83%)

1.00

0.06*

0.03*

1.00

0.06*

0.03*

0.78

0.81

Pro

cessingtemperatu

re(�C)

32.9

�1.4

29.7

�0.5

(�9.84%)

33.7

�1.2

(2.37%)

1.00

0.06*

0.03*

0.03*

0.06*

1.00

0.03*

0.06*

Freeacidity[g

(100g)�

1]

0.26�

0.06

0.20�

0.04(�

23.44%)

0.18�

0.03(�

31.23%)

0.97

0.13

0.06*

0.97

0.13

0.06*

0.91

0.31

Pero

xidevalue[m

eqkg�1]

3.4

�0.8

2.9

�0.9

(�15.44%)

5.1

�1.1

(48.95%)

0.94

0.19

0.09*

0.03*

0.06*

1.00

0.03*

0.06*

Totalpolyphenols

[mgl�

1]

182.2

�9.6

181.8

�9.5

(�0.26%)

179.0

�9.2

(�1.80%)

0.97

0.13

0.06*

1.00

0.06*

0.03*

1.00

0.06*

Chloro

phyll[m

gkg�1]

18.36�

1.33

18.10�

1.77(�

1.44%)

16.88�

1.32(�

8.05%)

0.69

0.81

0.41

0.97

0.13

0.06*

1.00

0.06*

Caro

tenoid

[mgkg�1]

11.48�

0.66

11.05�

0.77(�

3.69%)

10.77�

0.63(�

6.19%)

1.00

0.06*

0.03*

1.00

0.06*

0.03*

1.00

0.06*

K232[A

U]

1.836�

0.061

1.684�

0.130(�

8.28%)

1.842�

0.092(0.30%)

1.00

0.06*

0.03*

0.50

1.00

0.59

0.03*

0.06*

K270[A

U]

0.160�

0.017

0.134�

0.016(�

16.18%)

0.146�

0.021(�

8.95%)

1.00

0.06*

0.03*

0.81

0.63

0.31

0.22

0.44

Turb

idity[N

TU]

93.6

�1.2

54.1

�2.7

(�42.22%)

46.3

�5.4

(�50.50%)

1.00

0.06*

0.03*

1.00

0.06*

0.03*

0.97

0.13

3. Results and discussion

3.1. Results 48 h after extraction

Table 1 displays the results obtained after performing chem-

ical and spectrophotometric analyses associatedwith the data

collected during the processing operations.

The measured average water content was 0.08% in both

Sedoil and Cenoil (see Table 1). The values were significantly

different when referred to the Control, which proves that using

the enhanced sedimentation process allows efficient water

separation with reductions of water content of �70.4% and

�71.8%, respectively.

Furthermore, the oil temperature measured at the end of

the process was significantly different between Sedoil, Cenoil

and Control (see Table 1): the first process causes a tempera-

ture reduction of the oil soon after extraction from the

decanter exit from 32.9 �C to 29.7 �C, which represents a sig-

nificant reduction of �9.8%. Conversely, the oil sampled after

centrifuge cleaning reaches an average temperature of 33.7 �C,which represents a significant increase of the olive oil tem-

perature by þ2.4% from when it was extracted by decanter

centrifuge.

The average FA values of Sedoil and Cenoilwere significantly

lower than Control, moreover the average FA value of Sedoil

and Cenoil were not significantly different. This parameter

highlights the activity of hydrolytic enzymes in the presence

of water and at optimal temperatures in the range from 30 to

40 �C. The average water content measured in the samples

appears to be related with the trend in FA values. The

measured FA of Cenoil and Sedoil was lower than that of the

Control samples. This result confirms that the oil extracted

using the new sedimentation plant is not clean enough to be

considered stable and suitable for long-term storage.

PV measures the formation of hydroperoxides from poly-

unsaturated fatty acids through a radical mechanism in the

presence of oxygen; its presence indicates how the processing

operation affects product quality in terms of oxygen contact

between environmental air and olive oil during the cleaning

operation (Boskou, 2006). In the Cenoil olive oil samples, the

average PV was equal to 5.1 meq kg�1, which was significantly

greater than the average value of 2.9 meq kg�1 measured in

the Sedoil samples (see Table 1) and the average value of

3.4 meq kg�1 measured in the Control samples.

This result confirms that prolonged contact between the oil

and environmental air (oxygen) consequent to the use of the

disc stack centrifuge (as used in the olive oil industry) repre-

sents an important source of oxidative alterations (Di

Giovacchino et al., 1994; Masella et al., 2009). This is espe-

cially true considering that the centrifugation operation for oil

cleaning is performed while maintaining the oil at a temper-

ature over 30 �C as measured during the experimental trials.

The lower average PV measured in Sedoil relative to Control

(�15.4%) confirms the necessity of cleaning the olive oil after

centrifugal extraction. The presence of residual water in the

oil represents an important source of oxidation even though it

Fig. 3 e Principal component analysis of olive oil parameters after 48 h.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 105

has a smaller effect if comparedwith the oxygen action during

the centrifugal separation.

The average POLYPH value in Sedoil was 181.8 mg l�1, this

result is significantly lower (�0.3%) than that observed in

Control, moreover the average value measured in Cenoil is

significantly lower (�1.8%) than that measured in Control. This

result again confirms the oxidative action produced by the

cleaning operation performed by vertical disk stack

Fig. 4 e Principal component analysis of differences (SedoileCo

centrifugation. Considering the high water solubility of poly-

phenols, the high value of this parameter measured in Control

samples (182.2 mg l�1) could be related to the higher residual

water still present after decanting (0.28%) compared to Sedoil

and Cenoil, which were equal to 0.08%.

CHLO and CAR average values show significant differences

among the treatments, the higher water content of the sam-

ples and lower mechanical action on the oils produced

ntrol) vs. (CenoileControl) of olive oil parameters after 48 h.

Table

2e

Analysisofoliveoilafter6m

onth

s.Meanvalueswithstandard

deviationforeach

treatm

ent.ForSed

oilandCen

oil,in

parenth

esisis

calculatedth

erelative

percentdifference

withresp

ect

toth

eCon

trol’s

meanvalue.T

hreeW

ilco

xonsignedra

nktest,o

ne-sided(leftandrighttail)a

nddouble-sided,h

avebeenca

rriedoutonth

edifference

softh

epairedsa

mplesin

ord

erto

fullyass

ess

theexistence

ofanysignifica

ntlydifference

betw

eenth

etreatm

ents.A

10%

fam

ilywiseerrorra

tewasuse

d,thep-

valuem

ark

edwithanasterisk

meanssignifica

ntlydifference

forth

etest

andth

erejectionofth

erelatednullhypoth

esisH0.

Param

eter

Con

trol

Sed

oil

Cen

oil

H0:Con

trol

vs.

Sed

oilp-value

H0:Con

trol

vs.

Cen

oilp-value

H0:Sed

oilvs.

Cen

oilp-value

�¼

��

¼�

�¼

�Freeacidity[g

(100g)�

1]

0.46�

0.03

0.35�

0.04(�

24.12%)

0.36�

0.04(�

21.93%)

1.00

0.06*

0.03*

1.00

0.06*

0.03*

0.25

0.50

1.00

Pero

xidevalue[m

eqkg�1]

6.8

�1.1

5.2

�0.5

(�23.80%)

6.3

�0.4

(�6.32%)

1.00

0.06*

0.03*

0.91

0.31

0.16

0.03*

0.06*

1.00

Totalpolyphenols

[mgl�

1]

73.8

�9.8

79.2

�4.5

(7.37%)

79.3

�5.7

(7.51%)

0.09*

0.19

0.94

0.06*

0.13

0.97

0.50

1.00

0.56

Chloro

phyll[m

gkg�1]

5.98�

0.56

6.10�

0.85(1.94%)

5.67�

0.80(�

5.25%)

0.22

0.44

0.84

1.00

0.06*

0.03*

1.00

0.06*

0.03*

Caro

tenoid

[mgkg�1]

5.20�

0.84

5.01�

0.67(�

3.54%)

4.92�

0.63(�

5.39%)

0.97

0.13

0.06*

1.00

0.06*

0.03*

1.00

0.06*

0.03*

K232[A

U]

2.489�

0.285

1.923�

0.123(�

22.75%)

2.209�

0.438(�

11.24%)

1.00

0.06*

0.03*

0.91

0.31

0.16

0.03*

0.06*

1.00

K270[A

U]

0.179�

0.022

0.177�

0.016(�

1.00%)

0.189�

0.022(5.69%)

0.69

0.81

0.41

0.03*

0.06*

1.00

0.03*

0.06*

1.00

Turb

idity[N

TU]

51.7

�3.3

40.0

�2.5

(�22.58%)

38.2

�2.3

(�26.03%)

1.00

0.06*

0.03*

1.00

0.06*

0.03*

0.94

0.19

0.09*

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4106

significantly higher values in the Control and Sedoil samples

than in Cenoil.

Absorbance in the ultraviolet region (K232) for the Sedoil

samples indicated a significant lower value between this

sample and the Cenoil and Control samples. It appears that the

effect of oxygenation on the oxidation level of olive oil during

the stack centrifuge cleaning operation produces results

similar to the effect of high residual water content as

measured in the Control samples, while no difference was

found significant between Control and Cenoil.

The K270 parameter is used to measure the presence of

conjugate alkenes in the samples and did produce signifi-

cantly different average values only between Control and Sedoil

treatments, the Sedoil value results were significantly lower

(�16.2%) than that of Control. These differences are justified

considering that high values of this parameter are measured

in olive oil with chemical defects treated with a refining

process.

The average TUR value of 46.3 in Cenoil was significantly

lower than the average value of 54.1 measured in Sedoil and

the average value of 93.6 measured in Control. This result

confirms that centrifugal cleaning produces olive oil that is

much more “clean” and “brilliant” than the sedimentation

process. Likewise, the Sedoil has a murky aspect due to the

content of small solid particles that are still present in the oil,

which provides an “organic” aspect and could improve its long

term stability (Boskou, 2006; Tsimidou et al., 2005).

The analysis of these data shows that, for POLYPH, CHLO

and CAR, the value of each parameter in the olive oil obtained

through the sedimentation process is very close to the corre-

sponding value in olive oil obtained from the decanter.

Conversely, the value of the same parameter observed in the

olive oil obtained by centrifugation was quite different,

showing that the final centrifugal separation of the olive oil

from water and suspended solids produces modifications of

the olive oil quality parameters with respect to natural

settling. Moreover for the parameters as FA, PV, K232 and

K270, the treatment through sedimentation process decreases

the oxidation status of the treated oily must.

Figure 3 shows a principal component analysis (PCA)

decomposition of the samples after 48 h along the axis related

to Table 1 quality parameters. This was done to show the

parameters responsible, among the treatments, for the sig-

nificant differences and correlations. It is evident that the

three treatments are different, Sedoil and Cenoil have primarily

great differences along PV, K232 and temperature axis, indeed

Sedoil generates samples with lower PV, lower K232 and lower

temperature than Cenoil. The principal axis PC2 in Fig. 3 rep-

resents the oxidation status for the reason that it is positively

correlated to K232 and PV. Because the Sedoil samples are in

the negative region of PC2 (low oxidation status) and Cenoil

samples are in the positive region of PC2 (high oxidation sta-

tus) we could conclude that whereas the effect of Sedoil is that

of decrease the oxidation of the product, the effect of Cenoil is

that of increase the oxidation status of the oily must, thus

once again is demonstrated the oxidative action due to disk

stack centrifuge separators. This evidence is further clarified

by Fig. 4, where the PCA analysis of the differences

(SedoileControl) vs. (CenoileControl) is shown. From Fig. 4 it can

be seen that the difference between Sedoil and Cenoil are an

Fig. 5 e Principal component analysis of differences (SedoileControl) vs. (CenoileControl) of olive oil parameters after 180 d.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 107

increase of POLYPH and turbidity for Sedoil and a increase of

PV, K232 and process temperature for Cenoil, whereas CAR,

CHLO, FA, K270 and water content playing a minor role.

Furthermore, Fig. 4 also confirms the positive high correlation

existing between FA, CHLO, CAR, K270 and water content and

the negative high correlation between K270 and POLYPH.

Fig. 6 e Principal component analysis of olive

3.2. Results 180 days after extraction

The results obtained by performing the chemical and spec-

trophotometric analyses on samples after 180 d of storage are

shown in Table 2, the PCA analysis of the difference between

Sedoil and Cenoil treatments after 180 d is shown in Fig. 5. The

oil parameters after 48 h and after 180 d.

Table

3e

1H

NMRanalysisofoliveoilafter48h.Meanvalueswithstandard

deviationforeach

treatm

ent.ForSed

oilandCen

oil,in

parenth

esisis

calculatedth

erelative

percentdifference

withresp

ect

toth

eCon

trol’s

meanvalue.T

hreeW

ilco

xonsignedra

nktest,o

ne-sided(leftandrighttail)a

nddouble-sided,h

avebeenca

rriedoutonth

edifference

softh

epairedsa

mplesin

ord

erto

fullyass

ess

theexistence

ofanysignifica

ntlydifference

betw

eenth

etreatm

ents.A

10%

fam

ilywiseerrorra

tewasuse

d,thep-

valuem

ark

edwithanasterisk

meanssignifica

ntlydifference

forth

etest

andth

erejectionofth

erelatednullhypoth

esisH0.

Com

pound

Con

trol

Sed

oil

Cen

oil

H0:Con

trol

vs.

Sed

oil

p-value

H0:Con

trol

vs.

Cen

oil

p-value

H0:Sed

oilvs.

Cen

oil

p-value

�¼

��

¼�

�¼

�Sytostero

l1.108�

0.360

0.791�

0.235(�

28.58%)

0.935�

0.390(�

15.62%)

0.97

0.13

0.06*

0.69

0.81

0.41

0.31

0.63

0.78

Squalene

163.658�

53.676

160.572�

47.795(�

1.89%)

137.304�

34.733(�

16.10%)

0.59

1.00

0.50

1.00

0.06*

0.03*

1.00

0.06*

0.03*

Terp

enes

0.885�

0.414

0.668�

0.540(�

24.54%)

0.477�

0.340(�

46.11%)

0.84

0.44

0.22

0.94

0.19

0.09*

0.69

0.81

0.41

Form

aldehydea

0.278�

0.179

0.176�

0.333(�

36.71%)

0.039�

0.053(�

85.90%)

0.84

0.44

0.22

0.94

0.25

0.13

0.69

0.88

0.44

Trans-2-h

exenal

0.115�

0.046

0.063�

0.083(�

44.95%)

0.061�

0.068(�

46.69%)

0.84

0.44

0.22

0.84

0.44

0.22

0.59

1.00

0.50

Unsa

turatedaldehydes

0.145�

0.146

0.112�

0.241(�

22.38%)

0.061�

0.073(�

57.87%)

0.81

0.63

0.31

0.78

0.63

0.31

0.41

0.81

0.69

Hexanal

0.121�

0.097

0.085�

0.177(�

30.31%)

0.013�

0.015(�

89.46%)

0.84

0.44

0.22

1.00

0.06*

0.03*

0.50

1.00

0.59

aOneoutlierwasremovedfrom

data.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4108

results show that storage time, regardless of the used treat-

ment, dramatically affects the oxidation level of the olive oil.

All the parameters used to monitor the oxidation level

(including FA, PV and K232) increased after 180 d of storage.

Conversely, the content of natural antioxidants (POLYPH) and

pigments (CHLO and CAR) in the olive oil decreased.

The average FA observed in Control samples was 0.46 g

(100 g)�1 (see Table 2). This result further asserts that the oil

extracted using the innovative separation system is not

completely stable and suitable for long-term storage. The

values of the same parameter for Cenoil and Sedoil were

significantly lower (0.36 and 0.35 g (100 g)�1, respectively) than

Control, which emphasises that oxygen contact between

environmental air and olive oil during centrifugal cleaning

induces a greater formation of hydroperoxides from poly-

unsaturated fatty acids through a radical mechanism during

long-term storage. Not significant difference was observed

between FA values of Sedoil and Cenoil.

In the Sedoil olive oil samples, the average PV was equal to

5.2 meq kg�1, which is significantly lower than the average

value of 6.3 meq kg�1 measured in the Cenoil samples and the

average value of 6.8 meq kg�1 measured in Control samples.

This result further demonstrates that the prolonged con-

tact between the oil and environmental air (oxygen) during

the centrifugal cleaning operation produces irreversible olive

oil quality deterioration. Moreover the average value of

6.8 meq kg�1 measured in Control, higher than Cenoil and Sedoil

(see Table 2), confirms that the presence of residual water

during long-term storage represents the most important

source of oxidation. For this reason, an effective cleaning

operation is necessary to preserve olive oil quality during its

storage life.

POLYPH average values of 73.8, 79.2 and 79.3 mg l�1 were

measured in Control, Sedoil and Cenoil, respectively, after 180 d

of storage (see Table 2), the value of Control was significantly

lower than Sedoil and Cenoil; all of these values were less than

50% of the valuesmeasured soon after extraction (see Table 1).

The reduction confirms the important preservative action of

polyphenols against the oxidative reactions; unfortunately,

the amount of polyphenols decreased during the storage

despite the fact that the oil was stored in sealed bottles.

Chlorophyll reductions were very high in all the treat-

ments: oxidative action during the storage period produced a

reduction of over 70% of the initial contentmeasured after the

extraction. However, the value measured in Cenoil

(5.67 mg kg�1) was significantly lower than Sedoil

(6.10 mg kg�1) and Control (5.98 mg kg�1).

CAR average values decreased by more than 50% of the

initial value in all the samples. The value of 5.20 mg kg�1

measured in Controlwas significantly higher than 4.92mg kg�1

measured in Cenoil samples and 5.01 mg kg�1 measured in

Sedoil samples, even Sedoil value resulted significantly higher

than Cenoil value.

After storage, the absorbance in the UV region (K232) for

Sedoil (1.923) was found to be significantly lower than the value

of 2.209 for Cenoil and the value of 2.489 for Control. The signif-

icant difference between the average valuesmeasured in Sedoil

andCenoilagainconfirmed theoxidative actionof thedisk stack

centrifuge separators. The average value of 2.489 for theControl

samples, higher than Sedoil and Cenoil samples, confirms that,

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 109

during long-term storage, the oxygen that was dissolved in the

oil during the cleaning operation by the centrifuge is less

dangerous but produces similar results if compared to the high

residual water content in the Control samples.

The K270 measurement did not produce significantly

different average values among Control and Sedoil treatments;

the Cenoil samples average value of 0.189 was significantly

higher than the Control and Sedoil value, 0.179 and 0.177

respectively. This result demonstrates that Cenoil treatment

increases the presence of conjugate alkenes after a long-term

storage.

The TUR average value of 38.2 in Cenoil was significantly

lower than the average values of 40.0 and 51.7 measured in

Sedoil and Control, respectively. Furthermore, the values after

180 d of storage compared with the initial values show re-

ductions of 44%, 26% and 17% in Control, Sedoil and Cenoil,

respectively. From this result, it is possible to conclude that

sedimentation continues during the storage period, and the

content of small solid particles that give an “organic” aspect to

the Sedoil samples is quite stable during storage.

PCA analysis confirms these results (Fig. 5) about the dif-

ferences between Sedoil and Cenoil treatments after 180 d. The

major discriminant axes for the treatments, after 180 d of

storage, are CHLO, K270, FA, PV and K232: these parameters

measure the residual effect of the different treatments with

respect to Control, whereas POLYPH, CAR and turbidityweaken

their influence. From this point of view, the positive value of

PC2 axis represents the region with lower oxidation status

than Control treatment, indeed the Sedoil samples are all in this

region and all the Cenoil samples are in the opposite region.

Figure 6 shows the principal component analysis decom-

position of the oil parameters, as reported in Table 2, after 48 h

and after 180 d of storage. Once build the oxidation level axis

Fig. 7 eAmount of sytosterol in olive oil for each treatment and f

from the sedimentation process (Sedoil); olive oil from the centr

as having an angle which is the average of the axes angles of

the variables K232, K270, FA and PV, then the turbidityecolour

axis is taken perpendicular to the previous one and positively

correlated with turbidity, CHLO and CAR parameters. Conse-

quently from Fig. 6 becomes immediately clear the effect of

the storage on the olive oil samples: they moved from the

region with high turbidityecolour and low oxidation levels to

the regionwith high oxidation levels and low turbidityecolour

values. Furthermore, after the storage time the turbiditye-

colour value of the samples is almost the same independently

from the treatment, this is not true after 48 h from the treat-

ment. From the oxidation level point of view, the Sedoil

treatment allows to lower the oxidation level of the oily must,

furthermore, after the storage period the Sedoil treatment al-

lows to obtain an olive oil with lower oxidation values and

therefore capable of a more prolonged shelf-life.

3.3. 1H NMR and 13C NMR analysis results

Table 3 collects the data obtained from 1H NMR analyses of all

samples. The amount of sytosterol (Fig. 7 and Table 3) in oily

must is similar to that present in olive oil obtained through

sedimentation in samples 1, 3, and 4 while the oil obtained

through centrifugation gave a better result for sample 2. In

sample 5, the olive oil obtained through sedimentation had a

reduced amount of sytosterol while the olive oil obtained

through centrifugation had an increased amount. The Sedoil

treatment significantly lowers the sytosterol content (�28.6%).

The olive oil obtained through sedimentation showed a

relative amount of squalene that was very similar to that

detected in oily must while olive oil obtained through centri-

fugation produced a significantly reduced amount (�16.1%)

with respect to Sedoil and Control (Table 3).

or five samples: olive oil from the decanter (Control); olive oil

ifugation process (Cenoil).

Fig. 8 e Amount of terpenes in olive oil for each treatment and for five samples: olive oil from the decanter (Control); centre:

olive oil from the sedimentation process (Sedoil); right: olive oil from the centrifugation process (Cenoil).

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4110

The presence of terpenes in the oil was monitored, and the

results are reported in Fig. 8. The olive oil obtained through

sedimentation showed amounts similar to those present in

oily must in samples 1, 2, and 5 while the amounts of terpenes

Fig. 9 e Amount of trans-2-hexenal in olive oil for each treatmen

olive oil from the sedimentation process (Sedoil); olive oil from

detected in the olive oil obtained through centrifugation are

significantly lower than in oily must.

In samples 2 and 3, the olive oil obtained through sedi-

mentation gave similar results for the amount of trans-2-

t and for five samples: olive oil from the decanter (Control);

the centrifugation process (Cenoil).

Table 4 e 13C NMR analysis of olive oil after 48 h. Mean values with standard deviation for each treatment. For Sedoil andCenoil, in parenthesis is calculated the relative percent difference with respect to the Control’s mean value. ThreeWilcoxonsigned rank test, one-sided (left and right tail) and double-sided, have been carried out on the differences of the pairedsamples in order to fully assess the existence of any significantly difference between the treatments. A 10% familywiseerror rate was used, the p-valuemarkedwith an asteriskmeans significantly difference for the test and the rejection of therelated null hypothesis H0.

Compound Control Sedoil Cenoil H0: Control vs.Sedoil p-value

H0: Control vs.Cenoil p-value

H0: Sedoil vs.Cenoil p-value

� ¼ � � ¼ � � ¼ �Oleic acid 93.205 � 3.844 93.491 � 3.778 (0.31%) 95.070 � 4.587 (2.00%) 0.31 0.63 0.81 0.13 0.25 0.94 0.31 0.63 0.81

Linoleic acid 6.795 � 3.844 6.509 � 3.778 (�4.21%) 4.930 � 4.587 (�27.46%) 0.81 0.63 0.31 0.94 0.25 0.13 0.81 0.63 0.31

Stearic sn 1,3 0.130 � 0.073 0.134 � 0.077 (3.23%) 0.130 � 0.077 (�0.15%) 0.69 0.88 0.44 0.69 0.81 0.41 0.69 0.81 0.41

Oleic sn 1,3 0.547 � 0.048 0.546 � 0.056 (�0.33%) 0.559 � 0.055 (2.16%) 0.56 1.00 0.50 0.50 1.00 0.59 0.41 0.81 0.69

Oleic sn 2 0.323 � 0.079 0.320 � 0.056 (�0.74%) 0.311 � 0.056 (�3.60%) 0.69 0.81 0.41 0.81 0.63 0.31 0.59 1.00 0.50

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 111

hexenal (Fig. 9) to those determined in oily must. Olive oil

obtained through centrifugation gave an amount of trans-2-

hexenal similar to that observed in oily must only in the case

of sample 1. The hexanal content resulted significantly lower

in Cenoil (�89.5%) than in Control.

Table 4 collects all the results obtained through 13C NMR

analyses of the olive oil samples. Considering the relative

amounts of oleic and linoleic acid (Fig. 10 and Table 4), the

glycerol esters with stearic acid in positions 1 and 3 with oleic

acid in positions 1 and 3 (Table 4), and oleic acid in position 2

(Fig. 11 and Table 4) for all samples, the composition of olive

oil obtained through sedimentation is very similar to that of

the corresponding oily must while the use of the centrifuga-

tion in most cases gave altered results in comparison to the

corresponding oily must. This is also demonstrated calcu-

lating the mean Euclidean distance (AVGDIST) between the

paired samples that could be considered as a similarity

Fig. 10 e Average amount of oleic (dark grey) and linoleic acids

centre: olive oil from the sedimentation process (Sedoil); right: o

function. In this case we have AVGDIST (Sedoil, Control) ¼ 0.63

and AVGDIST (Cenoil, Control) ¼ 3.3, when testing these dis-

tributions with null hypothesis H0:

(SedoileControl) � (CenoileControl) we reach a p-value of 0.03

that significantly allows to reject H0 thus establishing that the

effect on Cenoil is significantly higher than Sedoil, therefore

Sedoil is more similar to Control than Cenoil.

The same concept of similarity could be used with the 1H

NMR compounds, in this case we obtain AVGDIST (Sedoil,

Control) ¼ 3.44 and AVGDIST (Cenoil, Control) ¼ 14.69, when

testing these distributions with null hypothesis H0:

(SedoileControl) � (CenoileControl) we reach a p-value of 0.01

that significantly allows to reject H0 thus establishing that the

effect on Cenoil is significantly higher than Sedoil, therefore

again Sedoil is more similar to Control than Cenoil.

Therefore the technical difference between Sedoil and

Cenoil is in the reduced amount of squalene (�16.1%)

(grey) in olive oil. Left: olive oil from the decanter (Control);

live oil from the centrifugation process (Cenoil).

Fig. 11 e Average amount of glycerol esters with stearic acid in positions 1 and 3 (dark grey), oleic acid in positions 1 and 3

(grey), and oleic acid in position 2 (light grey) in olive oil. Left: olive oil from the decanter (Control); centre: olive oil from the

sedimentation process (Sedoil); right: olive oil from the centrifugation process (Cenoil).

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4112

measured in Cenoil treatment lesser than the value obtained

with the Sedoil treatment (�1.9%) from the oily must;

furthermore, when comparing Sedoil treatment with Control

treatment only sytosterol amount results significantly lower

(�28.5%); finally, when comparing Cenoil treatment with Con-

trol treatment the squalene, terpenes and hexanal measured

values result significantly lower (�16.1%, �46.1% and �89.5%

respectively).

4. Conclusions

After centrifugal extraction, olive oil is generally cleaned using

a vertical disc stack centrifuge separator. The use of vertical

disc stack centrifuge separators influences dramatically the

final oil quality. This is an aspect that has not been investi-

gated in depth before.

The experiments carried out to compare the olive oil

quality after sedimentation and centrifugation demonstrated

that the centrifugation is responsible for negative effect on the

olive oil quality (loss of squalene (�16.1%), terpenes (�46.1%)

and hexanal (�89.5%); fatty acids composition seems not to be

affected) and loss of stability of the final product (increase of

peroxide value (þ49%) and decrease of polyphenols content

(�1.8%)) (Di Giovacchino et al., 1994; Masella et al., 2009) due to

oil heating (increased processing temperature (þ2.4%)) and to

increased amount of dissolved oxygen (increased peroxides

value and decreased polyphenols content) (Masella et al.,

2012; Parenti et al., 2007). The sedimentation significantly

lowers (�28.6%) the systosterol content of the oily must, also

decreasing its oxidation level (decreased peroxide value

(�15.4%), K232 (�8.3%) and K270 (�16.2%)) and allows a low

processing temperature (�9.8%). Both sedimentation and

centrifugation decrease the water content and the free acidity

of the oily must.

The analyses performed using 1H and 13CNMR showed that

Sedoil is more similar in its composition to Control than to

Cenoil. In particular, the amounts of sytosterol and squalene

and the relative amounts of terpenes are in agreement with

this assumption. Moreover, the ester composition as deter-

mined by 13C NMR showed the same trend.

The analysis after 180 d of storage showed that storage

time, regardless of the treatment used, dramatically affects

the oxidation level of the oil. All the parameters used to

monitor the oxidation level (i.e., free acidity, peroxide value

and K232) had increased after 180 d of storage. On the other

hand, the content of natural antioxidants (polyphenols) and

pigments (chlorophylls and carotenoids) in the olive oil

decreased.

Following a principal component analysis decomposition

two axes which identify a “turbidityecolour” region and an

“oxidation level” region were identified. The effect of the

storage causes the movement of olive oil samples from the

region with high turbidityecolour and low oxidation levels to

the regionwith high oxidation levels and low turbidityecolour

values. The turbidityecolour value of the samples is almost

the same, after the storage time, and independent of the

treatment. After the storage period the sedimentation treat-

ment allows olive oil to reach lower oxidation values and

therefore capable of a more prolonged shelf-life.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4 113

The globally improved quality of Sedoil, which is enhanced

by a light content of stable small solid particles that imparts to

the oil an “organic” aspect (Boskou, 2006; Tsimidou et al.,

2005), demonstrates that the cleaning operation using a disk

centrifugal separator produces modifications of the olive oil

composition with consequences on olive oil quality just after

the extraction and within a medium storage period.

Finally, the sedimentation plant used in the trials at in-

dustrial scale has allowed to use of the disk stack centrifuge to

be reduced achieving an effective energy saving per unit mass

of product and at the same time an improved quality of the

olive oil produced. Indeed, the vertical disc stack centrifuge

separator which is normally powered on continuously with

power consumption of around 4.5e5.5 kW whether or not oil

is being cleared. The natural sedimentation system is based

on 5 pumps, equipped with 0.18 kW electric engine, for mov-

ing the oil, the separated water and the solid part from the

different sedimentation columns and the final tank, the

pumps being powered only on the request of the control sys-

tem. So for this reason the authors consider the sedimenta-

tion plant as an energy saving system.

Acknowledgements

The authors thank dott. Racioppi, “Frantoio Oleario F.lli Pace

Srl” and “Oleificio Cooperativo COVAN” for their valuable help

and to have made possible the carrying out of these trials.

r e f e r e n c e s

Aiello, G., Catania, P., Enea, M., La Scalia, G., Pipitone, F., &Vallone, M. (2012). Real time continuous oxygen monitoringsystem during malaxation for the production of Virgin OliveOil. Grasas y Aceites, 63(4), 475e483.

Altieri, G. (2010). Comparative trials and an empirical model toassess throughput indices in olive oil extraction by decantercentrifuge. Journal of Food Engineering, 97(1), 46e56.

Altieri, G., Di Renzo, G. C., & Genovese, F. (2013). Horizontalcentrifuge with screw conveyor (decanter): optimization of oil/water levels and differential speed during olive oil extraction.Journal of Food Engineering, 119(3), 561e572.

Amirante, P., & Catalano, P. (1993). Analisi teorica e sperimentaledell’estrazione dell’olio d’oliva per centrifugazione[Theoretical and experimental analysis of olive oil extractionby centrifugation]. La rivista italiana delle sostanze grasse, LXX,329e335 (in Italian).

Amirante, R., & Catalano, P. (2000). Fluid dynamic analysis of thesolideliquid separation process by centrifugation. Journal ofAgricultural Engineering Research, 77(2), 193e201.

Amirante, R., Cini, E., Montel, G. L., & Pasqualone, A. (2001).Influence of mixing and extraction parameters on virgin oliveoil quality. Grasas y Aceites, 52(34), 198e201.

Amirante, P., Di Renzo, G. C., & Colelli, G. (1995). Estrazionecentrifuga con controllo continuo dei parametri operativi[Extraction by centrifuge with continuous control of theoperating parameters]. Rivista di Ingegneria Agraria, 17, 344e351(in Italian).

Angerosa, F., Mostallino, R., Basti, C., & Vito, R. (2001). Influence ofmalaxation temperature and time on the quality of virgin oliveoils. Food Chemistry, 72, 19e28.

Boncinelli, P., Daou, M., Cini, E., & Catalano, P. (2009). A simplifiedmodel for designing and regulating centrifugal decanters forolive oil production. Transactions of the ASABE, American Societyof Agricultural and Biological Engineers, 52(6), 1961e1968.

Boncinelli, P., Catalano, P., & Cini, E. (2013). Olive pasterheological analysis. Transactions of the ASABE, American Societyof Agricultural and Biological Engineers, 56(1), 237e243.

Boskou, D. (2006). Olive oil: Chemistry and technology (2nd ed.). AOCSPublishing.

Catalano, P., Pipitone, F., Calafatello, A., & Leone, A. (2003).Productive efficiency of decanters with short and variabledynamic pressure cones. Biosystems Engineering, 86(4),459e464.

Catania, P., Vallone, M., Pipitone, F., Inglese, P., Aiello, G., & LaScalia, G. (2013). An oxygen monitoring and control systeminside a malaxation machine to improve extra virgin olive oilquality. Biosystem Engineering, 114, 1e8.

Daou, M., Furferi, R., Recchia, L., & Cini, E. (2007). A modellingapproach to extra virgin olive oil extraction. Journal ofAgricultural Engineering, 38(4), 11e20.

Del Caro, A., Vacca, V., Poiana, M., Fenu, P., & Piga, A. (2006).Influence of technology, storage and exposure on componentsof extra virgin olive oil (Bosana cv) from whole and de-stonedfruits. Food Chemistry, 98, 311e316.

Di Giovacchino, L., Costantini, N., Ferrante, M. L., & Serraiocco, A.(2002). Influence of malaxation time of olive paste on oilextraction yields and chemical and organolepticcharacteristics of virgin olive oil obtained by a centrifugaldecanter at water saving. Grasas y Aceites, 53(2), 179e186.

Di Giovacchino, L., Sestili, S., & Di Vincenzo, D. (2002). Influence ofolive processing on virgin olive oil quality. European Journal ofLipid Science and Technology, 104(9e10), 587e601.

Di Giovacchino, L., Solinas, M., & Miccoli, M. (1994). Effect ofextraction systems on the quality of virgin olive oil. Journal ofthe American Oil Chemists’ Society, 71(11), 1189e1194.

Di Renzo, G. C., & Colelli, G. (1997). Flow behavior of olive paste.Applied Engineering in Agriculture, 13(6), 751e755.

Furferi, R., Carfagni, M., & Daou, M. (2007). Artificial neuralnetwork software for real-time estimation of olive oilqualitative parameters during continuous extraction.Computers and Electronics in Agriculture, 55, 115e131.

Gomez-Alonso, S., Mancebo-Campos, V., Salvador, M. D., &Fregapane, G. (2007). Evolution of major and minorcomponents and oxidation indices of virgin olive oil during 21months storage at room temperature. Food Chemistry, 100,36e42.

Jimenez, A., Beltran, G., Aguilera, M. P., & Uceda, M. (2008). Asensor-software based on artificial neural network for theoptimization of olive oil elaboration process. Sensors andActuators B: Chemical, 129(2), 985e990.

Kalua, C. M., Allen, M. S., Bedgood, D. R., Jr., Bishop, A. G.,Prenzler, P. D., & Robards, K. (2007). Olive oil volatilecompounds, flavour development and quality: a criticalreview. Food Chemistry, 100, 273e286.

Mannina, L., Barone, P., Patumi, M., Fiordiponti, P.,Emanuele, M. C., & Segre, A. L. (1999). Cultivar andpedoclimatic effect in the discrimination of olive oils: a high-field NMR study. Recent Research Developments in Oil Chemistry,3, 85e92.

Mannina, L., Fontanazza, G., Patumi, M., Ansanelli, G., & Segre, A.(2001). Italian and Argentine olive oils: a NMR and gaschromatography study. Grasas y Aceites, 52, 380e388.

Masella, P., Parenti, A., Spugnoli, P., & Calamai, L. (2009). Influenceof vertical centrifugation on extra virgin olive oil quality.Journal of the American Oil Chemists’ Society, 86(11), 1137e1140.

Masella, P., Parenti, A., Spugnoli, P., & Calamai, L. (2012). Verticalcentrifugation of virgin olive oil under inert gas. EuropeanJournal of Lipid Science and Technology, 114(9), 1094e1096.

b i o s y s t em s e n g i n e e r i n g 1 2 2 ( 2 0 1 4 ) 9 9e1 1 4114

Parenti, A., Spugnoli, P., Masella, P., & Calamai, L. (2007). Influenceof the extraction process on dissolved oxygen in olive oil.European Journal of Lipid Science and Technology, 109(12),1180e1185.

Piacqadio, P., De Stefano, G., & Sciancalepore, V. (1998). Quality ofvirgin olive oil extracted with new centrifugation systemusing a two phase decanter. Fett/Lipid, 10, 472e474.

Ranalli, A., Cabras, P., Iannucci, E., & Contento, S. (2001).Lipochromes, vitamins, aromas and other components ofvirgin olive oil are affected by processing technology. FoodChemistry, 73, 445e451.

Ranalli, A., De Mattia, G., & Ferrante, M. L. (1997). Comparativeevaluation of the olive oil given by a new processing system.International Journal of Food Science & Technology, 32, 289e297.

Sacchi, R., Mannina, L., Fiordiponti, P., Barone, P., Paolillo, L.,Patumi, M., et al. (1998). Characterization of Italian extra virginolive oils using 1H-NMR spectroscopy. Journal of Agriculturaland Food Chemistry, 46, 3947e3951.

Sacchi, R., Patumi, M., Fontanazza, G., Barone, P., Fiordiponti, P.,Mannina, L., et al. (1995). A high-field 1H nuclear magneticresonance study of the minor components in virgin olive oils.Journal of American Oil Chemists’ Society, 73, 747e758.

Salvador, M. D., Aranda, F., Gomez-Alonso, S., & Fregapane, G.(2003). Influence of extraction system, production year andarea on Cornicabra virgin olive oil: a study of five crop seasons.Food Chemistry, 80, 359e366.

Servili, M., Selvaggini, R., Taticchi, A., Esposto, S., &Montedoro, G. F. (2003). Volatile compounds and phenoliccomposition of virgin olive oil: optimization of temperatureand time exposure of olive pastes to air contact during themechanical extraction process. Journal of Agricultural and FoodChemistry, 51, 7980e7988.

Tsimidou, M. Z., Georgiou, A., Koidis, A., & Boskou, D. (2005). Lossof stability of veiled cloudy virgin olive oils in storage. FoodChemistry, 93(3), 377e383.