ICOM 2014 Layal DAHDOUH 24thof July 2014
Development of a new tool for the prediction of fruit juices
microfiltration performance
DAHDOUH Layal, WISNIEWSKI Christelle, KAPITAN-GNIMDU André, SERVENT Adrien,
DORNIER Manuel, DELALONDE Michèle
UMR Qualisud, Montpellier, France
Outlines
Context and aim of the work
Scientific strategy
Results and discussion
1. Juices physicochemical characterization
2. Juices filterability tests
3. Statistical analysis
4. Prediction of fruit juices filterability
Conclusion
Context of the work
Consumption of fruit
juices and drinks
containing fruit juices
•
Safe product
•
Fresh-like quality
•
Thermal treatments
Consumer
Industrial
Cross-flow microfiltration
•
Water treatment
•
Pharmaceutical field
•
Food industries (wine, beer, milk…)
Membrane seperation process
Fruit juices
industries
•
High-quality clarified
juices
•
Beverage made of
clarified juices
•
Step before other
specific treatments
Avantages
Membrane
fouling
Membrane
permeability
diminution
Predicting and
controlling juice
filterability
ICOM 2014 1 24thof July 2014Aim of the Work
Optimization
approaches and
strategies
Pre-treatment
and/or conditioning
of juices to be
filtered
Mechanisms
governing
membrane
fouling
few adequate
prediction tool are
proposed to anticipate
membrane fouling
Develop a simple and
fast juice filterability
prediction tool
Choose the suitable
approach for enhancing
microfiltration
performance
Lab-scale study
Potentiel industrial
application
Most of publications
Lack of knowledge
Study the
possibility of
predicting fruit
juices filterability
according to some
of their intrinsic
characteristics
Scientific strategy
Lab-scale filtration
Fruit juices
(orange juice)
Filterability response
Physicochemical characterization
♦
Conventional parameters:
pH, conductivity, Dry matter, …
♦
Physical Analysis: Particles size distribution, rheological behavior, …
♦
Elemental analysis: Nitrogen, mineral, …
● Variables / intrinsic characteristics of studied juices
● Response / filterability
S
TATISTICAL ANALYSIS
Highlight most important juice characteristics for predicting fruit juices filterability
Macroscopic approach
ICOM 2014 3 24thof July 2014
Juices physicochemical characterization
Juices selection
Nine different commercial brands of orange juices
Studied parameters
Total soluble solids (TSS), suspended insoluble solids (SIS), dry
matter (DM), total acidity (TA), pH, conductivity, turbidity
Stored at -20°C and defrosted before use at 4°C/one night
Juices physicochemical characterization
Juices selection
Nine different commercial brands of orange juices
Studied parameters
Stored at -20°C and defrosted before use at 4°C/one night
conventional
Total soluble solids (TSS), suspended insoluble solids (SIS), dry
matter (DM), total acidity (TA), pH, conductivity, turbidity
Physical measurements
Particle size distribution, rheological behavior, capillary suction
time
ICOM 2014 5 24thof July 2014
Juices physicochemical characterization
Juices selection
Nine different commercial brands of orange juices
Studied parameters
Stored at -20°C and defrosted before use at 4°C/one night
conventional
Total soluble solids (TSS), suspended insoluble solids (SIS), dry
matter (DM), total acidity (TA), pH, conductivity, turbidity
Physical measurements
Particle size distribution, rheological behavior, capillary suction
time
Conventional parameters
2 2,5 3 3,5 4 C 1 C 2 C 3 C 4 C 5 C 6 C 7 C 8 C 9 p H 6 7 8 9 10 11 12 C 1 C 2 C 3 C 4 C 5 C 6 C 7 C 8 C 9 T S S ( g /1 0 0 g ) 2 2,2 2,4 2,6 2,8 3 3,2 3,4 3,6 3,8 4 C 1 C 2 C 3 C 4 C 5 C 6 C 7 C 8 C 9 C o n d u ct iv it y (m S /c m ) 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 C 1 C 2 C 3 C 4 C 5 C 6 C 7 C 8 C 9 T A ( g c it ri c ac id /1 0 0 g )Values of these attributes were in the range of those previously reported for
commercial orange juices
ICOM 2014 7 24thof July 2014
Conventional parameters
0 2 4 6 8 10 12 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 D M ( g \1 0 0 m L) 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 S IS ( g \1 0 0 m L) 0 1000 2000 3000 4000 5000 6000 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 T u rb id it y (N T U )Elemental analysis
0 0,2 0,4 0,6 0,8 1 1,2 1,4 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 g /1 0 0 g l y o p h il iz e d ju ic e K+ Ca2+ Mg2+ Na+ 0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 g / 1 0 0 g l y o p h il iz e d ju ic e Ca2+ Mg2+ Na+ ICOM 2014 9 24thof July 2014Elemental analysis
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 g /1 0 0 g l y o p h il iz e d ju ic e N PParticle size measurement
Measurment method
42% Obscuration
1500 rpm
D
.
=
∑n
id
i3 ∑n
id
i2n:number of particles
d:diameter of particles
Equipement
Laser diffraction, Malvern Mastersizer 3000
Refractive Indices 1,33 (water) 1,73 (cloud
particles)
Absorption index 0,1 (cloud particles)
0
1
2
3
4
5
6
7
8
0,01
0,1
1
10
100
1000
10000
V
o
lu
m
e
d
e
n
si
ty
(%
)
Particle diameter (μm)
CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9Jus
CJ 1
CJ 2
CJ 3
CJ 4
CJ 5
CJ 6
CJ 7
CJ 8
CJ 9
D[3.2]
27,3 21 100 138 52 30,7 100 81 51 ICOM 2014 11 24thof July 2014Rheological behavior
1
10
100
1000
10000
0,1
1
10
100
1000
(m
P
a
.s
)
(s
-1)
CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 .Juices
CJ 1
CJ 2
CJ 3
CJ 4
CJ 5
CJ 6
CJ 7
CJ 8
CJ 9
K
12,6 12,2 197,9 440,3 96,6 13,4 8,9 9,7 11,5n
0,6 0,6 0,4 0,3 0,4 0,6 0,3 0,3 0,3=
K:consistency coefficient
n:flow behavior index
Capillary suction time
Filter paper
R1 R2
Front of migration Juice
CST-meter, measure of capillary suction time
CST= T2-T1
T1 T2
≠ Pressure
•
Originally developed for assessing
sludge filterability and industrial
suspensions (Gale and Baskerville,
1967, (Ruiz et al., 2010)
•
Not yet tested for fruit juices
0 200 400 600 800 1000 1200 CJ 1 CJ 2 CJ 3 CJ 4 CJ 5 CJ 6 CJ 7 CJ 8 CJ 9 C S T ( s) ICOM 2014 13 24thof July 2014
Juices filterability tests
Filtration cell
300 rpm
1.5 bar
25 mL
Glass fiber membrane 1.2 μm
Evaluate the filterability of the studied juices
0 5 10 15 20 25 0 1000 2000 3000 4000 5000 6000 7000 8000 P e rm e at e w e ig h t (k g ) Time (s) 10-3
Robust Loess
smoothing
J
(kg.h-1.m-2)=
(kg.h
-1.m
-2)
minute
First
Juices
CJ 1
CJ 2
CJ 3
CJ 4
CJ 5
CJ 6
CJ 7
CJ 8
CJ 9
Average
Fluxes (Kg.h
-1.m
-2)
8.6
12.3
6.8
9.5
16.3
11.4
10.6
13.2
14.2
Initial Fluxes
(Kg.h
-1.m
-2)
12.1
15.6
9.5
12.2
22.8
15.1
12.6
16.5
18.9
y = 1,3157x R² = 0,9537 0 5 10 15 20 25 6 8 10 12 14 16 18In
it
ia
l
fl
u
xe
s
((
K
g
.h
-1.m
-2)
Average fluxes (Kg.h
-1.m
-2)
•
Initial fluxes were used as
the only filterability
response variable in the
statistical analysis
ICOM 2014 15 24thof July 2014
Juices filterability tests
Statistical analysis
Predictor variables > sample number
Variables correlated
PLS (Partial Least Squares
regression)
5 7 9 11 13 15 17 19 21 23 25 5 10 15 20 25 P re d ic te d i n it ial f lu x ( K g .h -1.m -2) 0 1 2 3 4 5 6 V IPmodel-wise method (Afanador et
al., 2013)
PLS-VIP method (Afanador et al.,
2013)
5 7 9 11 13 15 17 19 21 23 25 5 10 15 20 25 F lu x i n it ial p ré v u ( K g .h -1.m -2)
Flux initial mesuré (Kg.h-1.m-2)
0 1 2 3 4 5 6 V IP Variables étudiées
model-wise method (Afanador et
al., 2013)
PLS-VIP method (Afanador et al.,
2013)
5 7 9 11 13 15 17 19 21 23 25 5 10 15 20 25 P re d ic te d i n it ial f lu x e s (K g .h -1.m -2)Measured initial fluxes(Kg.h-1.m-2)
Number of
samples
Number of
variables
Mean of the
measured
initial fluxes
Standard
deviation of the
measured initial
fluxes
Minimum of
the
measured
initial fluxes
Maximum
of the
measured
initial fluxes
RMSEC
RMSECV
R
29
17
15.0
4.0
9.4
22.8
0.79
2.10
0.957
9
5
15.0
4.0
9.4
22.8
1.06
2.09
0.923
Leave-one-out cross-validation test
(LOOCV)
(Cawley and Talbot, 2004)
RMSEC et RMSECV < 20 % of the
mean
(Bikindou et al., 2012)
ICOM 2014 17 24thof July 2014