Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Mathematical Model for Water Diffusion During Brining of

Hard and Semi-hard Cheese


JULIO A. LUNA and MdNlCA S. CHAVEZ

ABSTRACT tally. Due to this, theseauthorsalso suggested a relationbe-


The integral method, which is an approximate analytic method, was
tweensaltandmoisturetransportin orderto obtaininformation
used to obtain the theoretical profile of moisture changes during brin- aboutwaterbehaviorduringsalting.This relationshiphaddif-
ing of cheese. A mass layer was defined in the model where all water ficultiesand lack of clarity in mathematical
waterloss studies
variations take place. The model was applied at different times to (Guineeand Fox, 1983).
semi-hard cheese brining and compared with experimental data. A To completeour mathematicalmodelfor brining cheesea
moisture pseudo diffusion coefficient was obtained from experimental pseudodiffusion coefficientwas definedfor a specific posi-
data as part of the proposed model. Agreement was good, which tion. This pseudocoefficientwas usedto producea moisture
demonstratedthe applicability of the model. profile. It wasprovidedby differentapproximate methods:one
Key Words: Cheese, brining, mathematical model, water diffusion
coefficientwas obtainedfrom a graph in normal scale, the
otherfrom probabilitygraphpaper.Thevaluesobtainedthrough
eachmethodwere similar. Finally, the validity of the model
INTRODUCTION wasverifiedusingexperimental datafrom Geurts’et al. (1974)
andGuineeandFox (1983),and our own data.
DURING BRINING, salt is absorbedfrom sodiumchloride
solutioninto cheese,andsimultaneously wateris lost, resulting
in a cheesevolumereductionbecauselesssalt is absorbedthan THEORY
moistureis lost. Brining involvesheatand masstransferfur-
thermore,both processes influencethe final quality of cheese. The systemunderstudy is a cheeseconsideredas a rigid
Governingequationsof the systemare coupledby the water semi-infinitemediumimmersedin an infinite sodiumchloride
contentvariablein the liquid phaseof the porousmediumbut, solutionas describedin Fig. 1. Initially, the moistureconcen-
throughLuna and Bressan(1985)analysisbasedon the char- trationin thecheeseis constant,thenit changesduringbrining.
acteristicof the process,heatandmasstransferduringbrining We assumedthattherewasno chemicalreactionandtherewas
can be treatedseparately.However, in most of the brining no convectivemasstransferin the system;thus we can treat
cheeseanalysiswaterlosshasbeenneglected,so specificstud- the processas a one-dimensional isothermaldiffusion. During
ies on it havenot yet beenreported. brining thereis no changein moistureand sodiumchloride
Thus, the objectiveof our work was to designa mathemat- concentration at the cheesesurface.The diffusion coefficient
ical modelto describemoisturechangesduringhardandsemi- is thena functionof the waterandsodiumchlorideconcentra-
hard cheesebrining. To achievethis, the transferequation, tion insidethe cheese.
subjectto proposedboundaryconditions,was solvedby the The mathematical formulationof this problemis:
IntegralMethod.The methodbasiswasto definea masslayer,
whereall water variationoccurredso a masslayer thickness ac a
D$ ; x > 0; t > 0 (1)
was obtained.First, the proposeddifferential equationwas at=ax ( 1
integratedover the masslayer to removethe spacevariable.
Then, a polynomialprofile was chosento havemoisturecon- whereD is the water diffusioncoefficientin the cheese.
centrationdistributionover the layerandfitted usingproposed Equation(1) is subjectedto the boundaryconditions:
boundaryconditions. The polynomial, thus obtained, de-
pendedon time only. Then, it was introducedinto the mass c = c, atx=O,t>O (2)
integralequationand, after performingthe indicatedopera- c = co atxrO,t=O (3)
tions, an ordinarydifferentialequationin terms of the mass
layerwas obtained.Solutionof the lastequation,togetherwith
the profile previouslyobtained,gavethe moistureconcentra-
tion profile over the masslayer.
The outwardmigrationof wateris relatedto the penetration
1
of saltinto thecheese.In fact, it canbe describedasan impeded
mutualdiffusionprocess.Differentfactorsaffectthis phenom-
enon(Geurtset al., 1974).Theseauthorsproposeda cheese
matrix structuremode1which providedsalt concentration pro-
file, a pseudodiffusioncoefficient,andmoisturechangeanaly- Brining medium
sis duringthe saltingprocess.At brining,while moisturecontent
increasesgradually,salt contentdecreases in the sameway,
from the edgeof the cheeseinward. Moisturecontentaway
from the rind approaches initial moisturecontentasymptoti-

The authors are with lnstituto de Desarrollo Tecnoldgico para


la lndustria Qu/inica (INTEC)- Consejo National de lnvestiga-
ciones Cientificas y Tknicas (CONICET), Univ. Nat. del Litoral. -I
Giiemes 3450, C. C. No. 91,3000-Santa Fe, Reptiblica Argentina.
Fig. 1 -System under study.

Volume 57, No. 1, 1992-JOURNAL OF FOOD SCIENCE-55


DIFFUSION OF WATER IN CHEESE. . .

Cheese When this profile is introducedinto the massintegralequa-


tion, an ordinary differential equationfor s(t) is obtained:
di3
12 D, = 6 dt t>o

subjectto
Flow of
NoCl
39
s=o t=O (12)
I
I
%
I X
The solution of Eq. (11) with (12) gives:

Knowing s(t) the moisture concentrationprofile is deter-


mined accordingto Eq. (10).
Fig. 2-Schematic view of the moisture profile inside the cheese, D, is a pseudodiffusion coefficientdefinedin x = 0 where
and the mass layer. D(C,, C,,,) = D, = constant.D, was obtainedusing two
different graphicmethod.
Using the Integral Method in the problem defined by Eq.
(1) to (3) involves integratingEq. (1) over a penetrationdis- MATERIALS & METHODS
tanceS(t), which is a function of time, called the masslayer,
indicatedin Fig. 2. The masslayer is definedas the distance EXPERIMENTAL GOUDA cheeseswere provided by a local factory.
beyondwhich thereare no changesin the moistureconcentra- They were cylindrical (135 mm high, 200 mm diameter, 1 kg in
tion. weight)with aninitialmoistureof 45%.Thebrinesolutioncontained
Equation(1) is integratedin relationwith the spacevariable 20.50% sodium chloride and 1.5% calcium chloride; pH 5.5, tem-
from the position x = 0 to x = 6(t) to obtain: perature 12.6”C.
To avoid simultaneous salt diffusion from different directions, the
cheeses were waxed on the side and on one flat surface so that only
(4) one unwaxed surface was in contact with the brine solution. Ten
cheeses were brined and taken out of the brine in pairs for moisture
analysis after 0, 2, 4, 6, and 8 days of salting. A 20 mm diameter
The rule of differentiationin the integralis usedin the right- cylinder was cut from the cheese center, and from this cylinder 50
hand side of Eq. (4); using integral limits basedon: mm high slices were cut moving from the unwaxed surface. The slices
were analyzed in duplicate for moisture using a CEM microwave
oven.

from which
RESULTS & DISCUSSION
THE MASS LAYER thickness,6, must be previously deter-
mined to apply Eq. (10) to any proposedcase.We obtainedit
is obtained. using two different graphic methods.The first one selectsS
where, from moistureconcentrationvs. penetrateddistancegraph, in
w commonscale,as the distanceat which the following relation
e = I x-o CM dx is satisfied:

Equation(6) is calledthe massintegralequationin the pres- (C-G)


-= 0 gg
ent problem. Its solution is chosenas a polynomialprofile for (Co-cd) -
(14)
the moistureconcentrationdistributionover the masslayer:
C(x,t) = a, + a,x + a& + a4x3 (8) HereC, andC, aresurfaceandinitial moistureconcentration
equaled30.36% and 45.08%, respectively;and 0.99 was cho-
The first threecoefficientsin the polynomialare determined sendueto the estimationof the maximumrelativeerror, which
by using the boundaryconditionsgiven by Eq. (5), and the was 0.75%, so a relative error of 1% was suitable.
fourth coefficient is obtainedfrom using equation(1) at x = From the samegraphicmethodusingvisual appreciation,6
6, where the moistureconcentrationequalsC$ was also selected.Mass layer indicateswhere C approaches
Co asymptotically,which meansthat the curve slopeis null in
(9) that range. This importantassumptionwas implicitly used in
the visual appreciation.
Application of Eq. (5) and (9) to Eq. (8) yields the moisture Table 1 shows6 values using Eq. (14), &, and visual ap-
concentrationprofile in the form: preciation,S2. Both a1 and a2 were used in linear regression
to obtain D,r and D,, accordingto (13). A correlationcoeffi-
cient helpedus to selectthe most suitableD, value. For visual
C(x.t) = c, + (Co - C,) ;x - ; x2 + $x3 (10) appreciationR2 equals0.97 and for Eq. (14) 0.89 so, Ds2 =
( 1 9.56x lo-11m2/secwas taken.
so: The secondgraphicmethodproposedby Geurtset al. (1974)
providedD, using probability graph paper.Theseauthorsap-
a,=C,; a2=i (CO-CJ; a3=$ (CO-C,); ad=+ (Co-C,) proximatedsalt penetrationbehaviorduring brining to an error
function which allowed the use of probability paper. Water

56-JOURNAL OF FOOD SCIENCE-Volume 57, No. 1, 1992


Table 1 -Mass layer thickness obtained from Eq. (14), &, and visual ap- Table 3-Percent deviation between experimental and theoretical mois-
preciation, & ture concentration during brining process
Period of Experimental Experimental VE according to % Deviation* % Deviation’
brining (days) 6, h-d a2 Imm) (brining days) using D,? using DSJ
2 20 20 2 3.04 2.55
4 2.32 2.67
: 31 30
33 6 2.65 3.54
8 39 40 a 2.88 3.59
a With respect to VT, which was obtained from Eq (13) and (10).

Table 2-Theoretical mass layers at surface moisture concentration 30.36%


and initial concentration 45.08% Table *-Percent deviation between experimental and theoretical mois-
Theoretical Theoretical ture concentration during brining process
Period of
brining (days) S2 (mm) S3 b-m) VE according to: % Deviation’
2 19.91 22.59 Fig.1 -Geurts et al. (1974) 1.64
4 28.16 31.59 Fig.lO-Geurts et al. (1974) 1.14
6 34.46 39.12 Guinee and Fox (1983) 4.05
8 39.82 45.18 0 With respect to VT, which is obtained from Eq (13) and (10).

_ 09999 ,

OS

82
OS
8:
02
0t F 2 days of brining
Experimental data
b
0011 1 I 1 I 1 I l 8 days of brining
0 2 4 6 6 I) 12
-model prediction
x/K ~mm/doyV2)
I I I I I
25’ ’
Fig. 3-Moisture concentration from experimental data plotted I 9 18 27 36 45
on probability paper. Distance .fiom cheese surface (mm)
Fig. 4-Experimental and Theoretical valuesof moisture con-
centration during two and eight days of brining.
behavioris analogousto salt penetrationbut oppositeso, we
proposedthe following relationfor waterloss:
(Co- C)
-= Percentdeviationvaluesshowgoodagreement betweenthe
(15)
(Co-Cs) modelproposedand experimentaldatain both cases.
with z = 2 (Dt)rnandresultingK = 1 to satisfythe condition: To extendits applicability,experimentaldatafrom Geurts
et al. (1974)and Guineeand Fox (1983)were analyzed.We
x = 0. -= G-C) , appliedour modeland, finally, comparedresultsobtainedby
’ (G-G) percentdeviationvalues.
In the work by Geurtset al. (1974)Goudacheesewas used
Equation(15)wasplottedon probabilitypaperfor eachtime andsodiumchlorideandmoisturedistributionsweremeasured
of brining (2, 4, 6, 8 days).Our experimentaldatacould be after8.1 daysof brining(pH 5.64,T = 12.6”C,C, = 30.52%,
approximated to a straightline, D, is obtainedfrom the tangent G = 43.83%)and8 daysof brining (pH 4.79, T = 12.8”C,
of it, which provedthat D, was constantfor the samecheese C, = 31.00, Co = 41.13%).Thesehad beenplottedin Fig.
at everybrining time as is shownin Fig. 3. From Fig. 3 it 1 and Fig. 10 of that paper,respectively.We measuredmass
wasfeasibleto getthepseudodiffusioncoefficient,D, amounts layerusingthe visual methodfrom thesefigures:6 equaled5
to 1.23x 10-10cm2/sec. This methodapproaches waterbehav- cm and 4.5 cm, respectively.Data thus obtainedwere intro-
ior to Eq. (15) and thenwaterconcentration datato a normal ducedinto Eq. (10) andtheoreticalprofileswere obtained.
distributionthat is the straightline obtainedin Fig. 3. In the sameway, 6 = 5.75 cm was foundfrom the dataof
Theoreticala2and8, shownin Table2 werecalculatedfrom GuineeandFox (1983)Fig. 3 on experimental moistureprofile
Eq.(13)usingDs2andD, values.Theoreticalmasslayerthick- after 9 daysof brining of Roman0cheese(T = 20”, C, =
nesswasintroducedin Eq. (10)to obtain,finally, a theoretical 30.50%,C,, = 46.29%).Then Eq.(lO) was appliedto these
waterdistributionprofile. Figure4 showsthe goodagreement experimentaldata. Percentdeviationvaluesin Table 4 show
betweenthe theoreticalprofilesand the experimental onesfor good fitting of the proposedmodel to experimentaldata in
2 and8 daysof brining.It alsoshowsthat the proposedmodel hardcheeseduringbrining; thus,the modelvalidity was dem-
is valid only insidethe masslayerfor eachbrining time. onstrated.
Table3 showspercentdeviationvalues(Helderman,1974)
for the theoreticaland experimentaldata,basedon Eq. (17).
CONCLUSION
THE PROPOSEDtheoreticalmodel permittedeasyand ac-
curateestimationof water diffusion during brining in semi-
% deviation= hardcheeses.It was importantto havethe correctvaluesof
J D, or valuesof the masslayer 6(t), becausemoistureconcen-

Volume 57, No. 1, 1992-JOURNAL OF FOOD SCIENCE-57


DIFFUSION OF WATER IN CHEESE. . .
nation profiles dependedstronglyon thosefactors.The prob- VT Theoreticalvalue
ability graph papermethodwas nearly as good as the others X Spatialcoordinate(mm)
and so were its results,but it was mathematicallymore com- 6 Mass layer
plicatedthan the others.The equationsfor the water loss pro- 8 Integral along the masslayer
vided here, togetherwith other studies of sodium chloride Subscripts
diffusion in cheese,provide an important tool for analyzing 0 Initial state
changesin propertiesduring cheeseripening. S At the cheesesurface

NOMENCLATURE
REFERENCES
Polynomialcoefficient
t Moisture concentrationin the cheese Geurts, T.J., Walstra, P., and Mulder, M. 1974. Transport of salt and water
during salting of cheese. 1. Analysts of the process involved. Neth. Mik
Moistureconcentrationin the cheesesurface l+iry J. 28: 102.
2 Moisture concentrationinside the cheesebefore GmmneemTzoand Fox, P.F. 1983. Sodium Chloride and moisture changes
brining begins. e cheese durmg salbn . J. Davy Research. 50: 511.
Helderman, D.?T . 1974. Predicting the rc?a.tronships between unfrozen water
D Diffusion coefficientof water in the system fraction and temperature during food freezmg using freezing point
Diffusion coefficient at x = 0 depression. Trans. ASAE 17: 83.
Luna, J.A. and Bressan, J.A. 1985. Heat transfer during brining of Cuar-
? Pointsconsidered tirolo Argentmo cheese. J. Food Sci. 61: 858.
Time (minutes) MS received l/23/91; revised 6/29/91; accepted 7/15/91.
:E Experimentalvalue

MICROFILTRATION OF SKIM MILK. . .From page 48

A comparisonof the relative volume of control skim milk compositionaldatawith thosefor narrowlydefinedmicelle size
micelles(Fig. 2) with that of micellescomprisingthe permeate fractionsof Donnelly et al. (1984) is necessarilylimited.
and retentatefractions (Fig. 3 and 4) reflects the expected
volume differencesbetweenpermeateand retentatefractions.
A more effective size separationwas obtainedwith the 100 REFERENCES
nm membrane(Fig. 3) with relatively small numbersof mi-
celles larger than 100 nm presentin contrastto permeateob- Davies, D.T. and Law, A.J.R. 1983. Variation in protein composition of
tained using the 200 nm membrane(Fig. 4). Both retentate bovine casein micelles and serum casein in relation to mice&r size and
milh temperature. J. Dairy Res. 60: 67.
fractionsstill containedlarge numbersof micelleswith diam- Donnelly, W.J., McNeill, G.P., Buchheim, W., and McGann, T.C.A. 1984.
eters less than 100 nm indicating that microfiltration would A comprehensive study of the relationship between size.and protein com-
needto be extendedpast the 4:l concentrationlevel, together l;o$on 111natural bovme casem mrcelles. Blochrm. Brophys. Acta 789:
with possibleuse of diafiltration, to effect a greaterremoval Ekstrand! B., Larsson-Raznihiewicz, M., Brannang, E., and Swensson, C.
of smallermicellesfrom the retentatefractions. However,the 1981. Srze distribution of casein micelles related to coagulating proper-
g; ~lco5~parrson between drfferent breeds of cattle. Swedrsh J. Agric.
possibility of rearrangements in the size of micelles in reten-
tatesor permeatesshould also be considered(McGannet al., McG&m,‘T.d.A., Kearney, R.D., and Donnelly W.J. 1979. Developments
on column chromatography for the separati&r and characterization of
1980) since the skim milk systemis in dynamic equilibrium, casein micelles. J. Dairy Res. 46: 307.
dependentupon a variety of environmentalfactors. McGann, T.C.A, Donnell ., W.J., Kearney, R.D., and Buchheim, W. 1980.
Comption and size $ strrbutron
. . of bovine cesein micelles. Biochim.
Biop ys. Acta 630: 261.
Miller, G.L. and Burton, AL. 1959. Spectrophotometric determination of
Micellar caseincomposition aldoses by an iodometric procedure. Anal. Chem. 31: 1790.
Rose, D. and Colvin, J.R. 1966. Appearance and size of micelles from bo-
Resultsof the densitometricanalysisof the micellar casein vine milk. J. Dairy Sci. 49: 1091.
componentcompositionof the permeateandretentatefractions Yoshikawa,. M.? Taheuchi, M., Sasaki, R., and Chiba, H. 1982. Chemical
charactermatron of bovine casein micelles fractionated by size on CP-lO/
are given in Table 4. The Q- and rc-caseincontentswere 300 chromatography. Agric. Biol. Chem. 46: 1043.
comparablein all of the fractions, however, the qZ-casein MS received 2/ll/Sl; revised 7/20/91; accepted 8/11/91.
Mention of brand or firm names does not constitute an endorsement by
contentwas significantly higher and the p-caseinsignificantly the U.S. Department of Agriculture over others of a similar nature not
lower in both permeates.The lower p-caseincontent in the mentioned.
permeatefractions (smaller micelles)was in agreementwith
the report of Donnelly et al. (1984). However, the magnitude
of this differencebetweenlarge and small micelleswas much
greaterin our study. Since the permeatefractions only rep- Wethank Mr. T. Dobson for technicalassistancewith electron microscopyand Mr.
P. Smith for su~vlvinaoasteurizedskim mik
resentan enrichmentof smaller micelles, comparisonof our

58-JOURNAL OF FOOD SCIENCE-Volume 57, No, 1, 1992

You might also like