Development of a new tool for the prediction of fruit juices microfiltration performance

(1)

ICOM 2014 Layal DAHDOUH 24th_{of 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

(2)

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 24th_{of July 2014}

Aim 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

(3)

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 24th_{of 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

(4)

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 24th_{of 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

(5)

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 24th_{of 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 )

(6)

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 24th_{of July 2014}

Elemental 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 P

(7)

Particle size measurement

Measurment method

42% Obscuration

1500 rpm

D

.

=

∑

n

i

d

i3 ∑

n

_i

d

_i2

n: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 9

Jus

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 24th_{of July 2014}

Rheological 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 11,5

n

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

(8)

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 24th_{of 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

(9)

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 18

In

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 24th_{of 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 IP

model-wise method (Afanador et

al., 2013)

PLS-VIP method (Afanador et al.,

2013)

(10)

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

2

9

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 24th_{of July 2014}

Predictor variables > number of samples

Variables correlated

PLS (Partial Least Squares

regression)

Statistical analysis and Prediction of fruit juices filterability

Conclusion

It is possible to predict juices filterability

according to their intrinsic characteristics

Filterability can be satisfactorily predicted

according to Only five simple predictor

variables

0 1 2 3 4 5 6 V IP Studied variables

Possibility of extrapolating this strategy to

industrial scale since the selected variables

(11)