Publisher’s version / Version de l'éditeur:
Cement and Concrete Research, 35, January 1, pp. 141-153, 2005-01-01
READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.
https://nrc-publications.canada.ca/eng/copyright
Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la
Questions? Contact the NRC Publications Archive team at
PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the
first page of the publication for their contact information.
NRC Publications Archive
Archives des publications du CNRC
This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. /
La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version
acceptée du manuscrit ou la version de l’éditeur.
For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien
DOI ci-dessous.
https://doi.org/10.1016/j.cemconres.2004.07.016
Access and use of this website and the material on it are subject to the Terms and Conditions set forth at
Modeling ion and fluid transport in unsaturated cement systems for
isothermal conditions
Samson, E.; Marchand, J.; Snyder, K. A.; Beaudoin, J. J.
https://publications-cnrc.canada.ca/fra/droits
L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site
LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.
NRC Publications Record / Notice d'Archives des publications de CNRC:
https://nrc-publications.canada.ca/eng/view/object/?id=0f94b274-f37a-4da9-883d-f4fdd089f29a
https://publications-cnrc.canada.ca/fra/voir/objet/?id=0f94b274-f37a-4da9-883d-f4fdd089f29a
Modeling ion and fluid transport in unsaturated
cement systems in isothermal conditions
Samson, E.; Marchand, J.; Snyder, K.A.;
Beaudoin, J.J.
NRCC-45332
A version of this document is published in / Une version de ce document se trouve dans:
Cement and Concrete Research, v. 35, no. 1, January 2005, pp. 141-153,
Doi:
10.1016/j.cemconres.2004.07.016
ement systems for isothermal onditions E. Samson 1 2 J. Mar hand 1 2 K.A. Snyder 3 J.J. Beaudoin 4 1
CRIB Départementde génie ivil
Université Laval, Ste-Foy (Q ), Canada, G1K 7P4
2
SIMCO Te hnologiesin .
1400, boul. du Par Te hnologique, Québe (Q ),Canada, G1P4R7
3
Buildingand Fire Resear h Laboratory
National Institute of Standards and Te hnology, Gaithersburg,MD 20899, USA
4
Materials Laboratory -Institute for Resear h in Constru tion
National Resear h Coun il,Ottawa (Ontario), Canada, K1A 0R6
June 23, 2004
Abstra t
Ades riptionofioni transportinunsaturatedporousmaterialsduetogradients
intheele tro- hemi al potentialandthe moisture ontent isdevelopedbyaveraging
the relevant mi ros opi transportequations overa representative volume element.
The omplete set of equations onsist of a time-dependent equations for both the
on entration of ioni spe ies within the pore solution and the moisture ontent
within the porespa e. The ele trostati intera tions areassumed to o ur
instan-taneously and the resulting ele tri al potential satises Poisson's equation. Using
the homogenizationte hnique,moisture transportdueto both theliquidand vapor
Corresponding author. Current address at SIMCO Te hnologies in . Tel.: 418.656.1003, Fax:
ture ontent is found. The nal transport equations ontain transport oe ients
that an be unambiguously related to experimental quantities. The approa h has
the advantage of making the distin tion between mi ros opi and bulk quantities
expli it. KEYWORDS: Diusion (C) Transportproperties (C) Degradation(C) Modeling (E) Homogenization 1 Introdu tion
Overthe past de ade, agreatdeal ofeorthas been spe i allydevoted tothe
investiga-tion ofiontransportme hanismsinunsaturated ementsystems. The topi isimportant
sin e, in many ases, on rete stru tures exposed to ioni solutions are also frequently
subje ted towettinganddrying y les. The oupledtransportofmoistureandions often
tendstoa eleratephysi aland hemi aldegradationme hanismsandredu ethe servi e
life of the material[1, 2,3℄.
Reports re ently published on the subje t have largely ontributed to larify some
fundamental aspe ts of ion transport me hanisms in unsaturated on rete. Many
inves-tigations have also emphasized the intri ate nature of these phenomena. If most of the
di ulties related to the des ription of transport pro esses in on rete are linked to the
intrinsi omplexity of the material, it appears that part of them also lies with the fa t
that authorshave usedmany dierentapproa hes tostudy thesepro esses. Forinstan e,
the denition of the state variables used todes ribe the various transportpro esses tend
to vary signi antly from one study to another. This is most unfortunate sin e the la k
transport me hanisms are des ribed using a well established mathemati al pro edure,
the homogenization te hnique. The te hnique has been re ently used to investigate the
diusion of ions in saturated systems [4℄. A ording to this approa h, the transport
equationsare rst writtenatthe pore s ale. They are then averaged overthe s aleof the
material. Themainadvantageof thehomogenizationte hniqueliesinthe leardenition
of the state variables.
The paper rst addresses the pro ess of moisturetransport inanunsaturated porous
material. Forthe ompletely oupledtransportofionsinanunsaturatedmedia,dynami al
equations are required to express the moisture ontent as a fun tion of time. This is
a hieved by averagingmi ros opi alequationsforbothliquidand watervaportransport.
The mathemati al development yieldsRi hards' equation, and the moisture ontent and
the transport oe ients are well-dened.
The se ondpartof thepaperisdevotedtothe oupledtransportofionsand moisture
in the system. Here, the eld quantity is the on entration of the ions within the pore
solution. The homogenization te hnique is applied to a mi ros opi equation for both
diusive and onve tive transport. Whilediusive equationsalready exist,reformulating
the bulk equations using homogenization ensures that the transport oe ients are well
dened (pore spa e versus mi ros opi quantities) and an, therefore, be unambiguously
relatedto experimental quantities.
2 Water transport in unsaturated porous materials
The rst obje tive isto develop anequation to hara terize the mass transport of water
inanunsaturatedporousmaterial. Ri hards[5℄wasamongthe rstauthorstostudy the
me hanisms of water transport in unsaturated porous solids. In 1931, he proposed the
followingequation todes ribe the ow of water under apillary su tion:
t
apillary potential.
This relationship, known as Ri hards' equation, was later modied to express the
transportof masssolelyasa fun tionofthe gradientinwater ontent. Thismodi ation
is based on the assumption that the apillary potential is a dierentiable fun tion of
the moisture ontent :
=f() (2)
This allows to write:
grad =
d
d
grad (3)
Substituting (3)into (1), one nds:
t div(D grad) (4) whereD
=K(d =d)isthenonlinearwaterdiusivity oe ient. Equation(4)iswidely
used to model the evolution of water ontent in a porous material kept in isothermal
onditions. Equation (4)is alsoknown as Ri hards' equation.
While Ri hards' equation is ommonly a epted among s ientists, its use over the
past de ades has ontributed to some onfusion on how to des ribe moisture transport
me hanismsinunsaturatedporousmaterials. Ri hardsoriginallywrotethe equationwith
the water ontentexpressed in ubi entimetersof waterpergram ofdrymaterial. Over
the years, some authors have preferred to dene water ontent in kilograms of moisture
perkilogramofdrymaterial[6℄orinkilogramsof waterper ubi meterofmaterial[7,8℄.
However, mostauthorshavetraditionally hosentoexpressthe variablein ubi meterof
water per ubi meterofmaterial[9,10,11℄. Toaddtothe onfusion,manyauthorstend
to dene the moisture ontent as the sum of liquid water and vapor while some others
only onsider the liquidphase.
2.1 General onsiderations
In an attempt to larify these on epts, Ri hards' equation will be derived using the
deformations) kept under isothermal onditions (i.e. the transport of water is solely due
to apillary su tion). Other assumptions willariseduring the development of the model.
The mathemati alrules oftheaveragingte hnique an befound intextbooks [12,13℄.
Onlythe basi denitions willbeexposed in the followingparagraphs. More information
onthete hnique analsobefoundinreferen e[4℄. Thete hniqueisoutlinedherebe ause
it isat the ore of development of allthe transportequations.
As previously mentioned, the homogenization te hnique starts with a onservation
and a transport equation at the mi ros opi level (i.e., at the s ale of the pore). These
equations are then integrated over a Representative ElementaryVolume (REV), su h as
the one depi ted in Figure 1. The size of the volume depends onthe intrinsi properties
of the material. For instan e, for on rete and mortar mixtures, the size of the REV
depends on the maximum diameter of the aggregate parti les. For the hydrated ement
paste, the REV is typi allya few ubi entimeters.
The totalvolumeof theREVisgiven byV
o
. Thevolumeo upied bythe liquidphase
isdesignatedby V
L
o
. The volumetri fra tionofliquid
L
isthe ratioofthe liquidvolume
tothe total volume:
L = V L o V o (5)
The gaseous phase o upies a volume V
G
o
. It is a mixture of air and water vapor. It is
assumed that both air and the water vapor ll the whole gaseous phase volume. As for
the liquid phase, the volumetri fra tion of gas
G
is the ratio of the gas volume to the
total volume: G = V G o V o (6)
In the remainder of the text, the supers ripts L and G will designate the liquid and
gaseous phases, respe tively. Furthermore, the supers ript V will represent the water
vapor phase withinthe totalgaseous phase.
Let a
denote the amountperunit volume of some extensive quantity A inthe phase
, eithersolid,liquidorgas. Con entration ormass density an serveasexamplesfora
.
a = 1 V o Z V o a dV (7)
The volumetri intrinsi phase average is dened as:
a = 1 V o Z V o a dV (8)
The two values are relatedby the following relationship:
a = a (9)
2.2 Transport of liquid water
The ontinuity equation forliquid water is given by [14, 15℄:
L t +div( L v L )=0 (10) where L
isthemassofliquidwaterperunitvolumeofliquidphase,andv
L
isthevelo ity
of water. The bulk equationis obtained by averagingEquation (10) over the REV:
1 V o Z V L o L t +div( L v L ) dV =0 (11)
This integral an be divided intwo parts:
1 V o Z V L o L t dV + 1 V o Z V L o div( L v L )dV =0 (12)
Usingthe denition of the volumetri phaseaverage (equation 7),one an write:
L t +div( L v L )=0 (13)
The average of the time derivative gives [12, 13℄:
( L L L ) t = L t + 1 V o Z S LG o L un LG dS+ 1 V o Z S LS o L un LS dS (14) where S LG o
is the surfa eof the liquid/gas interfa e, S
LS
o
isthe surfa e of the liquid/solid
interfa e, u is the velo ity of the interfa e, n
LG
LS
liquidphaseatthe liquid/solidinterfa e. Sin eitisassumed thatthedeformations ofthe
solid matrix ould be negle ted, the last integral on the right-hand side of equation (14)
an be dropped, whi h leaves:
( L L L ) t = L t + 1 V o Z S LG o L un LG dS (15)
The average of the divergen e in equation(13) is given by [12, 13℄:
div( L v L )=div L ( L v L L ) + 1 V o Z S LG o L v L n LG dS+ 1 V o Z S LS o L v L n LS dS (16)
At the solid/liquid interfa e, it is assumed that the liquid velo ity is zero (the so- alled
no-slip ondition of uid me hani s [16℄). Hen e, the last integral onthe right-hand side
of equation (16) an be negle ted, whi h leaves:
div( L v L )=div L ( L v L L ) + 1 V o Z S LG o L v L n LG dS (17)
Substituting equations(15) and (17) in equation (13), one nds:
( L L L ) t +div L ( L v L L ) + 1 V o Z S LG o L (v L u)n LG dS =0 (18)
A ording to Whitaker [15℄, the integral in equation (18) orresponds to the rate of
vaporizationperunitvolumeoftheliquidphaseattheliquid/gasinterfa e,andisdenoted
by m._ Also,the average value
L L
orresponds tothe density of the liquid
L
, whi h an
be assumed onstant. Equation (18) an thus besimplied:
L L t + L div( L v L L )+m_ =0 (19)
The next step onsists of determining the average value of the liquid velo ity. The
starting pointis the Dar y equation [17℄:
v L = K (gradP + L g) (20) The quantity v L
is the bulkvelo ity of the liquid, K isthe permeability of the material,
is the vis osity of the uid, P is the pressure on the liquid, and g is gravitational
in materialshavingvery smallpores the apillary for es are dominant: v L = k L gradp (21) Thequantityp
isthe apillarypressureand k isthepermeabilityof theliquid-lledpore
spa e.
Equation (21) isbasedonthe assumptions that gravity ee ts are negligibleand that
the pressure is uniform throughout the liquid and gaseous phases. It should also be
emphasized that the validity of the equation alsorests on the hypothesis that the liquid
phaseis ontinuous. The latterassumptionwillbefurther dis ussed inthelastse tionof
this report.
The bulk velo ity of the liquidv
L
an be related to its intrinsi average ounterpart
through: v L = L v L L (22)
Substituting equations(21) and (22) intoequation (19) gives:
L L t L div K L gradp +m_ =0 (23) Sin e p =f( L
)[15℄, the hain rule allows to write:
gradp = dp d L grad L (24)
The substitution of equation (24) inequation (23) gives:
L L t L div K L dp d L grad L +m_ =0 (25) Let D L = K L dp d L (26)
Equation(25)isnowexpressedasafun tionofasingleeldquantity
L
togivea omplete
des ription of liquidwater transport:
L L t L div(D L grad L )+m_ =0 (27)
L
to awater diusion oe ient. However, it should be emphasized that the movement of
liquid water onsidered in this se tion arises by apillary su tion. It is not, per se, a
diusive phenomenon.
With the denition of D
L
given in equation (26), ombined with equation (24), the
velo ity of the liquidphase (equation 21) an now be writtenas:
v L = D L grad L (28)
2.3 Transport of water vapor
The treatment of the gas transport phenomenon is more ompli ated sin e two phases
have to be onsidered: air and water vapor. However, the problem an be simplied
by onsidering the following assumptions. As mentioned in the previous se tion, the
development of equation (21) rests on the hypothesis that pressure is uniform over the
gaseous phase. This impliesthat there is nobulk movement of airin the gaseousphase.
Consequently, there will be no onve tive transport of water vapor within the material
pore stru ture. Still, there an be movement of mole ules in the gaseous phase as a
resultof their thermalagitation. The otherassumption isthat gravity doesnot haveany
signi ant ee t onthe behavior of the water vapor.
The ontinuityequationforwatervapor omponentofagaseousphaseisthefollowing
[15℄: V t +div( V v V )=0 (29) The quantity V
is the mass of water vapor per unit volume of gaseous phase, and v
V
is the velo ity of water vapor. The water vapor will be in movement as a result of its
thermal agitation. It is therefore a diusive pro ess. A ording toDaian [10℄, the water
vapor uxis given as:
V v V = Dgrad V (30)
V t div(Dgrad V )=0 (31)
The bulk equation is al ulated from the integration of equation (31) overthe REV:
1 V o Z V G o V t div(Dgrad V ) dV =0 (32)
This integral an be divided intwo parts,whi h yields:
1 V o Z V G o V t dV 1 V o Z V G o div(Dgrad V )dV =0 (33)
A ording to the denition of the volumetri phase average (equation 7), equation (33)
an be written as: V t div(Dgrad V )=0 (34)
The average of the time derivative isgiven by:
( G V G ) t = V t + 1 V o Z S GL o V un GL dS+ 1 V o Z S GS o V un GS dS (35) where S GL o
is the surfa e of the liquid/gas interfa e, S
GS
o
is the surfa e of the gas/solid
interfa e, u is the velo ity of the interfa e, n
GL
is a unit ve tor pointing outward the
gaseous phase atthe liquid/gasinterfa e, and n
GS
is a unit ve tor pointing outward the
gaseousphase atthe gas/solidinterfa e. Sin e itisassumed that the deformationsof the
solid matrix are negligible, the last integral on the right-hand side of equation (35) an
be negle ted, whi h leaves:
( G V G ) t = V t + 1 V o Z S GL o V un GL dS (36)
The average of the divergen e gives:
div(Dgrad V )=div G (Dgrad V G ) + 1 V o Z S GS o Dgrad V n GS dS+ 1 V o Z S GL o Dgrad V n GL dS (37)
The rst integral on the right-hand side of equation (37) is negle ted sin e there is no
ex hangeof watervaporbetweenthe solidand thegaseousphases. A ordingly, equation
(37) an besimpliedas:
div(Dgrad V )=div G (Dgrad V G ) + 1 V o Z S GL o Dgrad V n GL dS (38)
as: div(Dgrad V )=div( G Dgrad V G )+ 1 V o Z S GL o Dgrad V n GL dS (39)
The average of the gradient is given by [12, 13℄:
grad V G = G grad V G + 1 V o Z S GS o Æ x(grad V n GS )dS+ 1 V o Z S GL o Æ x(grad V n GL )dS (40)
The quantity isreferred toby Ba hmatand Bear[12℄ as the tortuosityof the material.
Con eptually,itisthe ratio ofthe ma ros opi system lengthtothe shortest path length
through the pore (liquidor gas) spa e. As su h, it is a quantity that stri tly equalto or
less than one. The parameter
Æ
x is dened as
Æ
x = x x
o
, where x is a position ve tor
withinthe REV andx
o
isthe positionve tor ofthe enter ofthe REV. Therst integral
ontheright-handsideofequation(40)involvesthesolid/gasinterfa e. Ex eptforthevery
low water ontent onditions, there will be no dire t onta t between these two phases
be ause water willbe adsorbed onthe surfa e of the solid. A ordingly, the integral an
be negle ted. It is assumed that the term (grad
V
n
GL
) in the se ond integral on the
right-hand side of equation (40) varies very slightly over the surfa e S
GL
o
. Under this
assumption, itleavesan integralof a position ve tor times a s alarover a losed surfa e,
whi hgiveszero. Equation (40) is thussimpliedas:
grad V G = G grad V G (41)
Repla ingequations (36), (39) and (41) into equation(34) gives:
( G V G ) t div( G D G grad V G )+ 1 V o Z S GL o (Dgrad V V u)n GL dS =0 (42)
Substituting equation(30) in equation (42),one nds:
( G V G ) t div( G D G grad V G ) 1 V o Z S GL o V (v V u)n GL dS =0 (43)
Withaker[15℄ showed that the integral inequation (43) has the same value as the one in
equation (18). It represents the rate of ondensationper unit volume of the water vapor
phase atthe liquid/gasinterfa e. Therefore, equation (43) an bewritten as:
( G V G ) t div( G D G grad V G ) m_ =0 (44)
L V
f(
L
) [10℄. Applying the hain rule, it gives:
( G V G ) t div G D G d V G d L grad L _ m=0 (45)
The quantities pre eding the gradient within the parenthesis an be lumped together to
forma single vapor diusion oe ient:
D V = G D G (46)
Equation (45) an be writtensu in tly:
( G V G ) t div D V d V G d L grad L _ m=0 (47)
2.4 Total moisture transport
Inthe previousse tions, the transportequationsforthe liquidand thevaporphaseswere
onsideredseparately(equations27and47). Inordertoget a omplete des riptionofthe
transport, both equationsshould beadded together:
t ( L L + G V G ) div L D L + d V G d L D V grad L =0 (48)
Asthedensityofwatervaporhasamu hlowervaluethanthe oneofliquidwater(
V G L ) and G L
,equation (48) an be simpliedas:
L L t div L D L + d V G d L D V grad L =0 (49) Let D = L D L + d V G d L D V L (50)
Substituting equation(50) into(49) gives:
L t div(D grad L )=0 (51)
This isRi hards' equation. As an beseen, the equationfully des ribes the movementof
both vapor and liquid water on the basis of a single variable
L
tothe vaportransportanda se ondonerelatedtothe transportofthe liquidphase. The
demonstration also indi ates that the variable of the original equation (4) stands for
the volumetri liquidwater ontent, whi h isexpressed in ubi meter of water per ubi
meter of material.
2.5 Determination of the moisture transport properties of
hy-drated ement systems
The des ription of moisturetransportme hanismson the basis of equation (51) requires
the determination of the fun tion D
. An interesting dis ussion of the variation of this
fun tion with the water ontent of the material has re ently been published by Martys
[19℄. The author learly emphasizes the non-linear hara ter of this fun tion.
Measurementsmadeonasand olumn[20℄showthatwhenthehumidityinthemedium
ishigherthan4%(by weight),the ontributionofthe vaporphasetothe overall moisture
transfer is negligible. In that ase, one an assume that D
=D
L .
Over the years, numerous experimental te hniques have been used to determine the
moisturetransportproperties ofhydrated ement systems. Athorough dis ussion of this
subje tis beyond the s ope ofthis paper. Comprehensive riti alreviews ofthis problem
an be found in referen es [21, 22℄.
3 Ioni transport in unsaturated porous materials
Several mathemati al models have been developed to predi t the movement of ions in
ement-basedmaterials. Mostoftheseapproa hesare single-ionmodels, onsideringonly
hlorideanditsdetrimentalee tonthedurabilityofthematerial. Mostofthetime,su h
models onsider the transport of ions under the ee t of diusion and adve tion (uid
ow). Also onsidered is the ee t of the hemi al rea tions involving the onsidered
However, these models oversimplify some basi physi al phenomena. For instan e,
the ele tri al oupling between the ions [18℄ and its ee t on their movements is often
overlooked. This is parti ularlytrue for ement-based materials sin e they ontain
on- entrated porous solution. The ele tri al oupling between the ions for on entrated
solutions was re ently put in eviden e in two papers by Snyder [26, 27℄ that report on
diusion experiments through non rea tive erami frits. Multiioni models taking into
a ountele tri al ouplingwere re entlypublishedbyMasielal. [28℄andTru etal. [29℄.
Unfortunately,asitwasthe ase withRi hards'equation,thereisala kof agreement
with regard to the denition and the use of some parameters in these models. For
ex-ample, the diusion oe ient is sometimes alled the intrinsi diusion oe ient, the
apparent diusion oe ient, or the ee tive diusion oe ient. On e again, the
aver-aging pro edure is used to generate an ioni transport model. The method will larify
some of the basi on epts behind the modeling of ioni transport. Su h a work was
previously published [4℄ but was applied only to non-rea tive saturated materials. The
modelpresented inthe following se tionsis more general.
3.1 Transport of ions in the liquid phase
Thetransportmodelisbasedonthe observationthat thetransportof ionsonlyo urs in
the liquidphase. Hen e, noequation has tobe developed forthe solid orgaseousphases.
The onservation equation foran ioni spe ies i inthe liquidphase atpore s ale isgiven
by the following mi ros opi equation [13℄:
i t +div(j i )+r i =0 (52) The quantity i is the on entration, j i
is the ioni ux and r
i
is a sour e/sink term
a ountingforthehomogeneous hemi alrea tions[39℄betweenionsinsolution. Thebulk
equationis obtained fromaveragingthis equationover the REV,followingthe pro edure
that lead toequations (13) and (34):
i t +div(j i )+r i =0 (53)
( L i L ) t = i t + 1 V o Z S LS o i un LS dS+ 1 V o Z S LG o i un LG dS (54)
The rst integral on the right-hand side of equation (54) ontains a term that a ounts
for the velo ity of the solid/liquidinterfa e. Whilethis interfa e may possibly move as a
resultofsomedissolution/pre ipitation hemi alrea tions,itwilldosoveryslowly. It an
thus be negle ted. The other integral involvesthe movement of the liquid/gas interfa e.
It is similarto the rst integralin equation (14). While itwas used in the mathemati al
development of the moisture transport, it is assumed that it has only a small ee t on
the ioni transport, and an thus be negle ted,simplifyingequation (54) to:
( L i L ) t = i t (55)
The average of the divergen e is given by:
div(j i )=div( L j i L )+ 1 V o Z S LS o j i n LS dS+ 1 V o Z S LG o j i n LG dS (56)
Thelastintegralontheright-handsideofequation(56)a ountsfortheioni ux rossing
the liquid/gaseous interfa e. The value of this ux is zero sin e ions do not go into the
gaseous phase. The other integral, related to the ux of ions a ross the solid/liquid
interfa e, will be used to model the various hemi al rea tions involving those phases.
A ordingly,equation (56) an beredu ed to:
div(j i )=div( L j i L )+ 1 V o Z S LS o j i n LS dS (57)
Substitutingequations (55)and (57) inequation(53) and averagingthe term r
i by L r i L
(see equation 9),one nds:
( L i L ) t +div( L j i L )+ L r i L + 1 V o Z S LS o j i n LS dS =0 (58)
The next step onsists of writing the proper ux expression at the mi ros opi level
(ions in bulk ele trolyte) and averaging it over the REV. Due to the harged nature
of ions, this expression has to onsider the ele tri al oupling between ioni parti les.
hemi al a tivity. Finally, the movement of the uid itself will have an impa t on the
movement of ions. All these physi al phenomena an be taken into a ount through the
extended Nernst-Plan k modeltowhi his added anadve tion term [18℄:
j i = D i grad i D i z i F R T i grad D i i grad(ln i )+ i v L (59) The parameter D i
is the self-diusion oe ient [30℄ of spe ies i in diluted, free water
onditions,
i
is the hemi al a tivity oe ient, is the ele tri al potential, z
i
is the
valen e number of the ion, F is the Faraday onstant, R is the ideal gas onstant and
T is the absolute temperature. The terms on the right-hand side of equation (59) are
asso iated with diusion, ele tri al oupling between the ions, hemi al a tivity ee ts
and water transport, respe tively.
TheintegrationoftheuxovertheREV,similartothepro edurefollowedinequations
(9)and (11) to(13), leads to:
L j i L = D i L grad i L D i z i F R T L i grad L D i L i grad(ln i ) L + L i v L L (60)
The next steps onsistin averagingthe various gradientsand variablesin equation (60).
The average of the on entration gradient is given by [12, 13℄:
grad i L = L grad i L + 1 V o Z S LS o Æ x(grad i n LS )dS+ 1 V o Z S LG o Æ x(grad i n LG )dS (61) The quantity L
is the tortuosity of the aqueous phase. It is a purely geometri al fa tor
a ounting for the omplexity of the paths the ions must travel through in liquidspa e.
It is a fun tion of the water ontent
L
, sin e it is related tothe volume of liquid in the
pore spa e, and itsvalue is less than one [12℄. The parameter
Æ
xwas rst en ountered in
equation (40).
To evaluatethe surfa e integralsin equation(61), onehas torefer tothe doublelayer
theory [31℄. Figure 2 shows the ross-se tion of a pore and the s hemati shape of the
on entration prole along its radius. The solid bears an ele tri al surfa e harge
solid .
It is neutralizedby harges of the opposite sign in two dierent zones, the Stern and the
diuse layers, bearing respe tively
stern
and
di
solid = stern + di (62)
Theexternallimitofthe Sternlayer, alledthe outerHelmholtzplaneortheshearplane,
separates the solid fromthe aqueous phase, inwhi h ioni diusion o ur. The aqueous
phase is divided in the diuse layer and the free water zone, where ions do not feel
the ee t of the solid/liquid interfa e. A re ent study by Revil [32℄ showed that ioni
transportmay o ur inthe Stern layer. Butit wasalsomentioned inthe paperthat this
phenomenon isnegligiblewithrespe ttotransportin the bulk porewhenthe pore solution
ishighly on entrated,asitisthe ase in ement-basedmaterials. Consequently,onlythe
ioni transport in the aqueous phase is onsidered in this paper. Finally, the des ription
of the ross-se tion of the pore is omplete by onsidering a gaseous phase atthe enter
of the pore, whenthe latter isnot saturated [33℄.
It is assumed that the on entration proles at the liquid/gas interfa e is at (see
gure 2). Consequently, the se ond integral in equation (61) is negligible sin e there
is no on entration gradient along the radius at the liquid/gas interfa e. The situation
is dierent for the rst integral be ause of the on entration gradient along the radius
at the solid/liquid interfa e aused by the ele tri harge at the surfa e of the solid.
Simple double layer al ulations made with the Gouy-Chapman model [31℄ are shown
on Figure 3. They emphasize that in reasing the ioni strength of a solution in the
vi inity of a harged surfa e de reases dramati ally the thi kness of the double layer.
Sin e ementitiousmaterialsbear ahighly hargedsolution,the doublelayerextend over
a very small region. Outside this region, the on entration prole is unae ted by the
surfa e harge. Followingthis,theterm
R Æ x(grad i n LS )dS inequation(61)isnegle ted, leaving: grad i L = L grad i L (63)
Then one needs to average the term in equation (60) on erned with the ele tri al
oupling between the ions. A ording to the pro edure for averaging a produ t [12, 13℄,
one an write:
i grad L = i L grad L + Æ i Æ grad L (64)
Æ i = i i L (65)
Itisassumedthatthedeviationsleadtosmallterms,whi hallowstonegle tthedeviation
produ t inequation (64): i grad L = i L grad L (66)
Followingthe same pro edure astheone used forthe on entrationgradient,the average
of the potentialgradientgives:
grad L = L grad L + 1 V o Z S LS o Æ x(grad n LS )dS+ 1 V o Z S LG o Æ x(grad n LG )dS (67)
Figure 2 shows a potential prole a ross the se tion of a pore. A ording to the double
layer models [31℄, it has a shape similar to the on entration prole, i.e. it is disturbed
near the solid/liquid interfa e but tends to a at prole toward the enter of the pore.
And like the on entration proles shown on Figure 3, in reasing the ioni strength of
the solution redu es the thi kness of the area where the gradient of potentialis dierent
from zero. A ordingly, the integrals are negle ted, assuming again that the ele tri al
phenomenaneartheinterfa edonotae t ioni movement. Equation(67)thussimplies
to: grad L = L grad L (68)
Substituting equation(68) into(66) gives:
i grad L = i L L grad L (69)
The same approa h is used to average the hemi al a tivity term in equation (60).
The same assumptions on erning the deviations, as well as those on erning the ee t
of the ele tri alphenomena at the solid/liquidinterfa e lead to:
i grad(ln i ) L = i L L grad(ln i L ) (70)
It is assumed that the term ln
i L
orresponds to the hemi al a tivity oe ients
al- ulated with the average on entrations
i L . For simpli ity, ln i L is approximated by ln i L : i grad(ln i ) L = i L L grad(ln i L ) (71)
i v L = i L v L + Æ i Æ v L (72)
The term in equation(72) ontaining the deviations is alled the dispersive ux [13, 34℄.
It is shown inthe previousreferen es that it an be writtenunder aFi kian form:
Æ i Æ v L = D disp grad i L (73) where D disp
is alled the oe ient of adve tive dispersion, and is due to ngering, not
diusion. Consequently,thisterm anbeaddedtotheioni diusiontermthatwouldthen
exhibitanew diusion oe ientbeing thesum ofthe lassi alone plusthe oe ientof
adve tive dispersion. When the uid is in movement under the ee t of a water ontent
gradient, as des ribed in the pre eding se tion, the velo ity is relatively weak. In that
ase, the dispersion term an be negle ted [35℄, leadingto:
i v L = i L v L (74)
Substituting equations (63), (69), (71), and (74) in equation (60), gives the average
ux expression: L j i L = D i L L grad i L D i z i F R T L L i L grad L D i L L i L grad(ln i L )+ L i L v L (75)
The diusion oe ient atthe ma ros opi levelD
i isdened as: D i = L D i (76)
To simplifythe expression, let:
C i i L (77) L (78)
Substituting equations(76) to (78) inequation (75) gives:
L j i L = D i L gradC i D i z i F R T L C i grad D i L C i grad(ln i L )+ L C i v L (79)
the ma ros opi ioni transportequation: ( L C i ) t div D i L gradC i + D i z i F R T L C i grad +D i L C i grad(ln i L ) L C i v L + L r i L + 1 V o Z S LS o j i n LS dS =0 (80)
In order to simplify this equation, the integral must be expressed in a manner that is
more friendly to a further numeri al analysis. The term (j
i
n
LS
) gives the amount of
ions rossingthe solid/aqueous phaseinterfa e, asaresult of dissolution/pre ipitationor
ionex hangerea tions. Itispossible toexpress itdierentlyby performingthe averaging
operationontheionsinthesolidphase[13℄. The onservationequationatthemi ros opi
s ale for the ions in solid phase is:
is t +div(j is )=0 (81)
wherethesubs ript sdesignatesthe solidphase. Contrarytoequation(52),itisassumed
thatno hemi alrea tionso urwithinthe solidphase,sin e allpre ipitation/dissolution
phenomenaaretakingpla eatthesolid/aqueousphaseinterfa e. Averagingequation(81)
over the REV leads to:
( s C is ) t +div( s j is s )+ 1 V o Z S SL o j is n SL dS =0 (82) where s
isthe volumetri fra tion ofsolid phase andn
SL
isanoutward (tothe S-phase)
unit ve tor on the solid/aqueous phase interfa e (designated as S
SL
). The integral in
equation (82) has the same value as the one in equation (80) but with an opposite sign
sin e the ions oming out of the aqueous phase are being bound by the solid phase.
Furthermore, the ux j
is
within the solid is zero sin e there is noioni movement in this
phase. This allows one to write:
( s C is ) t = 1 V o Z S SL o j is n SL dS = 1 V o Z S LS o j i n LS dS (83)
Substituting equation(83) in equation (80) gives:
( L C i ) t div D i L gradC i + D i z i F R T L C i grad +D i L C i grad(ln i L ) L C i v L + L r i L + ( s C is ) t =0 (84)
Thisisthegeneralexpressionfortheioni transport inporousmaterialsunderisothermal
To model the transport of ions under the inuen e of apillary su tion, it would seem
straightforward to substitute equation (28) in (84). However, the development of the
water transport equations was made for the ase of pure water in a porous material.
When ions are in solution, the vapour pressure above a solution is lower than in pure
water [36℄. This ee t is quantied through Raoult's law. A ordingly, the relationship
p
=f(
L
)should instead bewritten as:
p =f( L ; i ) (85)
sin e the presen e of ions in solution is likely to disturb the equilibrium between the
aqueous and gaseous phases in a pore. To evaluate in what extent the presen e of ions
will ae t the vapor pressure of water, one an use Raoult's law to al ulate the vapor
pressure hange between pure water and a 500 mmol/L NaCl solution with water as
solvent. A ording toRaoult's law [36℄, the vapor pressure hange is given as:
p v =X solute p Æ v (86) where X solute
is the molar fra tion of solute (NaCl) in the solution and p
Æ v
is the vapor
pressure of pure water. At 25
Æ
C , the vapor pressure of bulk water is 3.17 kPa [36℄.
Knowing that inone literof water thereare 56moles:
X solute
=
0.5 moleNaCl
0.5 moleNaCl +56 moles water
=0:009 (87)
This gives a hange in vapor pressure of p
v
=0:03 kPa, whi h is obviously very weak.
A ording tothe result of this simple al ulation,the ee t ofioni on entration onthe
apillarypressure isnegle ted. Itwasalsonegle tedinthe modelspresented inreferen es
[23, 24,25, 28℄.
Substituting equation (28) inequation (84) gives:
( L C i ) t div D i L gradC i + D i z i F R T L C i grad +D i L C i grad(ln i L )+C i D L grad L + L r i L + ( s C is ) t =0 (88)
This equation an be used to model the transport of ions in unsaturated ement-based
expression to al ulate the hemi al a tivity oe ients. These topi s are addressed in
the followingse tions.
3.3 Cal ulation of the potential
The ele tri al potentialin equation (88) arises in the materialto enfor e the
ele troneu-trality ondition. If two spe ies are diusing ina materialwithone of the spe ies having
a greater self-diusion oe ient. To maintain a neutral solution, the potential reated
slows the fastest ions and a elerates the slowest ones.
The mathemati alrelationship thatrelates ele tri alpotentialtoele tri al harges in
a given media is given by Poisson's equation [18℄
r 2 + =0 (89)
where is the ele tri al harge density and is the medium permittivity. The harge
density an be relatedto the ioni on entration through:
=F N X i=1 z i i (90)
where N isthe numberof ioni spe ies. Substituting equation(90) in(89) gives:
r 2 + F N X i=1 z i i =0 (91)
It may seem awkward tohave anequation from ele trostati s ina model where the ions
aremovingthroughtime. However, sin etheele tromagneti signalismovingmu hmore
rapidly than the ions, Poisson'sequation is perfe tlysuitable.
To use equation(91) in the transport model, ithas tobe averaged overthe REV. As
it was done previously, it is assumed that the boundary ee ts at the liquid/solid and
liquid/gas interfa es are negligible. Following the same average rules as in the previous
se tions, we get the following relationship:
div( L L grad )+ L F N X i=1 z i C i =0 (92)
The models to al ulate the hemi al a tivity oe ients are numerous. The rst ones
developed are the Debye-Hü kel and extended Debye-Hü kel models [37℄. From purely
ele trostati onsiderations, they relate the hemi al a tivity oe ients of ioni spe ies
to the ioni strength of a solution. They are valid for ioni strengths up to 10 and 100
mmol/L respe tively.
In ement-based materials, the ioni strength is mu h higher. To suit this parti ular
situation, a hemi al a tivity relationship was developed re ently by Samson et al. [38℄
whi hgivesgoodresults for highly on entrated solutions:
ln i = Az 2 i p I 1+a i B p I + (0:2 4:1710 5 I)Az 2 i I p 1000 (93)
where I is the ioni strength of the solution, and A and B are temperature dependent
parameters. The parameter a
i
inequation (93) varies with the ioni spe ies onsidered.
3.5 Modeling of hemi al rea tions
Two terms appear in equation (84) to a ount for hemi al rea tions. The term r
i L
is
a sink/sour e term that models homogeneous hemi al rea tions [39℄, i.e. rea tions that
solelyinvolvethe aqueousphase, as for instan e:
Ca 2+ (aq) +OH (aq) !CaOH + (aq) (94)
The other term related to hemi al rea tions is (
s C
is
)=t. As mentioned before, it
a ountsforioni ex hangesbetweenthe aqueousandsolidphases. This typeof hemi al
rea tion is alled heterogeneous [39℄. It in ludes dissolution/pre ipitation and surfa e
exhange phenomenons. The formation of portlandite is an example of heterogeneous
rea tion: Ca 2+ (aq) +2OH (aq) !Ca(OH) 2(s) (95)
In most ases, hemi al rea tions are modeled by assuming that they are faster than
tious materials. Under LEA, hemi alrea tions are modeled by algebrai mathemati al
relationships [39℄. Following a paper published in 1989 by Yeh and Tripathi [41℄, the
urrent trendfor solving ioni transportproblems inrea tivematerialsis toseparate the
transport and hemi alrea tion parts. The partial dierential equationsdes ribing ioni
transportare solved withthe nitedieren eorniteelementmethodwhereas aNewton
algorithmisused tosolve the nonlinear algeabri system of equation asso iated with the
hemi al rea tions. Depending on the type of hemi al rea tions involved in a problem,
dierentalgorithms anbeusedtosplittransportand hemistry,asreviewedinreferen es
[42, 43℄.
When the lo al equilibrium assumption is not valid, hemi al rea tions are modeled
with kineti expressions [39℄ involving rea tion rate. This ase arises for problems in
groundwater ioni transport where large pressure head gradients an be at the origin of
high uid velo ity. The topi of kineti ally ontrolled rea tion modeling is dis ussed in
referen es [44, 45, 46℄.
3.6 Evaluation of the ioni transport properties
Two dierent transport parameters appear in equation (88). There is the diusion
o-e ient D
i
asso iated with the diusion pro ess and the liquid water diusivity D
L to
hara terizethe ee t ofthe uidvelo ity onthe ioni transport. Adis ussion ofD
L was
already given inse tion 2.5.
The diusion oe ient is evaluated with the migration experiment test. It onsists
of a elerating hloride ionswith anapplied externalpotentialthrough adisk of
ement-based materials glued between two ells lled with ioni solutions. The analysis of the
results yieldsthe diusion oe ients. Dierent analysis methodsare found in the
liter-ature. One is based onsteady state measurements of hloride having rossed the sample
[47,48℄. Another[49℄isbasedonmeasuringnon-steady state hlorideprolesbygrinding
the sample after a short exposure. A re ent paper by Samson et al. [50℄ des ribes a
method based on urrent measurements during the migration test. The measurements
All these methods are performed insaturated onditions. As shown in equation(76),
the diusion oe ientD
i
depends, through
L
,onthe saturation ondition. Nomethod
ould be found in the literature to evaluate this parameter for unsaturated onditions.
However, it is possible that D
i
might not be ae ted by the saturation state of the
material above a given saturation level, the latter being dened as s =
L
=, where
is the porosity. Revil [32℄ showed that for shalysand, the diusion of the ions is almost
unae ted for a water saturation above 0.6. We thus infer that for on rete stru tures
exposed tohigh relativehumidityenvironment, the diusion oe ient isindependentof
the water ontent.
In the ux equation (75), the re urring quantity
L
L
, whi h ould also be written
L
s,is analogous toa saturation-dependent formation fa tor for the liquidphase of the
poresystem. ThesaturationsresultsfromtheaveragingovertheREV.Thetortuosity
L
is alsoa fun tion of the saturation and ree ts the onne tedness of the moisturephase.
At a riti al moisture ontent s
, the liquid phase is no longer onne ted, the tortuosity
L
goes tozero, the transport withinthe liquidphase eases.
The remaining question is the dependen e of the tortuosity
L
on the saturation.
Althoughnopre isedataexistfor ementitioussystems,thereexistsqualitativedatafrom
whi hinferen es an bemade. Thesedata typi ally expressthe relative total ondu tion
as a fun tionof the saturation s. The relative total ondu tion =
o
is analogous to the
produ t of the saturation and the relative tortuosity:
(s) (s=1) = s L (s) L (s =1) = s L Lo (96)
Therefore, dividing these results by s will yieldthe relative hange inthe tortuosity.
TheworkofMartys[19℄suggeststhatforapreferentiallywettingliquidbeingdispla ed
by a non-wetting one, the limiting behavior of
L
near saturation an be approximated
by the diluteee tive mediumtheory result:
s L Lo =1 3 2 (1 s)+ 1 2 (1 s) 2 s!1 (97)
ap- L Lo = s 2 + 1 2 (98)
Therefore, a de rease to 80 % saturation will result in a 10 % hange in the tortuosity.
Given that transport oe ients an routinely hange by orders of magnitude, a 10 %
hange in the tortuosity is relatively quite minor. Sin e this result is only approximate
near saturation, further redu tions insaturation would have a fargreater ee t.
4 Con lusion
Themathemati almodeldevelopedinthispaperisrstsummarized. Formaterialswhere
the water transport o urs as a result of apillary su tion, the water ontent prole an
be al ulatedwith Ri hards'equation(51). The ions willmove inthe materialunderthe
ombinedee tofdiusion(in ludingele tri alanda tivityee ts)andwatermovement
a ording to equation (88). The ele tri al potential, arising from the ele tri al oupling
between the ions in order to maintain a neutral solution, is al ulated with Poisson's
equation (92). The hemi al a tivity oe ients, for the highly harged pore solution
of ement-based materials, an be evaluated with expression (93). Finally, several
refer-en es were given to address the modeling of hemi al rea tions o uring in ementitous
materials.
The useoftheaveragingte hnique learly helpsto larifythe meaningofsome
impor-tantparametersinthe model. A ordingtothiste hnique,thewater ontentinRi hards'
model orrespondstoavolumetri water ontent. Thewaterdiusivitywas learlyshown
to be a ontribution of both liquid water and vapor transport. The diusion oe ient,
the parameter that hara terize the ioni diusion pro ess, isdire tly related tothe
geo-metri alproperties of the materialthrough aparameter alled the tortuosity.
The averaging te hnique proved to be a powerful mathemati al tool to lay the
[1℄ A. Andersen, HETEK Investigation of hloride penetration into bridge olumns
exposed todei ing salts, TheDanish Road Dire torate, Reportno. 82,Copenhagen,
1997.
[2℄ G. Fagerlund, Predi ting the servi e-life of on rete exposed to frosta tion through
a modeling of the water absorption pro ess in the air pore system, In: H. Jenning,
J. Kroppand K,S rivener (Eds.),The Modeling ofMi rostru ture and itsPotential
forStudyingTransportPropertiesandDurability,NatoASISeries,SeriesE:Applied
S ien e, vol.304,Kluwers A ademi Publishers,TheNetherlands, 1996,pp.503-539.
[3℄ B.F. Johannesson, Non-linear transient phenomena in porous media with spe ial
regard to on rete durability, Advan ed Cement Based Materials6 (1997)71-75.
[4℄ E. Samson, J. Mar hand, J.J. Beaudoin, Des ribing ion diusion me hanisms in
ement-based materials using the homogenization te hnique, Cement and Con rete
Resear h 29(1999) 1341-1345.
[5℄ L.A. Ri hards, Capillary ondu tion of liquids through porous mediums, Physi s 1
(1931) 318-333.
[6℄ H.A. Dinules u, E.R.G. E kert, Analysis of the one-dimensional moisturemigration
aused by temperature gradients in a porous media, International Journal of Heat
and Mass Transfer 23 (1980)1069-1078.
[7℄ G.Dhatt, M.Ja quemier, C. Kadje,Modellingof drying refra tory on rete, In: M.
Mujumdar (Ed.), Drying '86, Vol. 1, Hemisphere Pub. Corp., New-York, 1986, pp.
94-104.
[8℄ J.Selih,A.C.M.Sousa,T.W. Bremner,Moisturetransportininitiallyfullysaturated
on rete during drying, Transport inPorous Media 24(1996) 81-106.
[9℄ T.A. Carpenter, E.S. Davies, C. Hall,L.D. Hall,W.D. Ho, M.A. Wilson, Capillary
watermigrationinro k: pro essandmaterialpropertiesexaminedbyNMRimaging,
sizedistribution,equilibriumwater ondensationandimbibition,TransportinPorous
Media 3 (1988)563-589.
[11℄ D.A. De Vries, Simultaneous transfer of heat and moisturein porous media, Trans.
Ameri an Geophysi alUnion 39 (5)(1958) 909-916.
[12℄ Y. Ba hmat, J.Bear, Ma ros opi modellingof transportphenomena inporous
me-dia. 1: The ontinuum approa h, TransportinPorous Media 1 (1986)213-240.
[13℄ J. Bear, Y. Ba hmat, Introdu tion toModeling of Transport Phenomena in Porous
Media, Kluwer A ademi Publishers, The Netherlands, 1991.
[14℄ D.L. Landau and I.M. Lifshitz, Fluid Me hani s, PergamonPress, Oxford, 1987.
[15℄ S. Whitaker, Simultaneous heat, mass, and momentum transfer in porous media: a
theory of drying, Advan es inHeat Transfer 13(1977) 119-203.
[16℄ B.R. Munson, D.F. Young, T.H. Okiishi, Fundamentals of uids me hani s, John
Wiley &Sons, Canada, 1990.
[17℄ F.A.L. Dullien,Porousmedia : Fluidtransport andpore stru ture,A ademi Press,
San Diego, 1992.
[18℄ F. Heleri h, Ionex hange,M Graw-Hill,New-York, 1961.
[19℄ N.S. Martys, Diusionin partiallysaturated porous materials,Materialsand
Stru -tures 32(1999) 555-562.
[20℄ P. Crausse, G.Ba on, S. Bories,Etude fondamentale des transferts ouplés
haleur-masse en milieu poreux, Int. J.Heat MassTransfer 24(6)(1981) 991-1004.
[21℄ L Pel, Moisture transport in porous building materials, Ph.D. Thesis, Eindhoven
University of Te hnology, The Netherlands, 1995.
[22℄ H.M. Kunzel, Simultaneous heat and moisture transport in building omponents,
Ph.D. Thesis, FraunhoferInstitute of BuildingPhysi s,Germany, 1995.
[23℄ A. Saetta, R. S otta, R. Vitaliani, Analysisof hloride diusion into partially
minationof diusion oe ients, ACI Materials Journal 95(2)(1998) 113-120.
[25℄ P.N. Gospodinov, R.F. Kazandjiev, T.A. Partalin, M.K. Mironova, Diusion of
sul-fate ions into ement stone regarding simultaneous hemi al rea tions and resulting
ee ts, Cement and Con rete Resear h 29 (1999)1591-1596.
[26℄ K.A. Snyder, The relationship between the formation fa tor and the diusion
oef- ient of porous materials saturated with on entrated ele trolytes: theoreti al and
experimental onsiderations,Con reteS ien eandEngineering3(12)(2001)216-224.
[27℄ K.A. Snyder, J. Mar hand, Ee t of spe iationon the apparent diusion oe ient
innonrea tive poroussystems, Cement and Con reteResear h31(2001)1837-1845.
[28℄ M. Masi, D. Colella, G.Radaelli, L. Bertolini, Simulationof hloridepenetration in
ement-based materials,Cement and Con rete Resear h 27(10) (1997)1591-1601.
[29℄ O.Tru , J.P. Ollivier, L.O. Nilsson, Numeri alsimulationof multi-spe ies diusion,
Materials and Stru tures 33(2000) 566-573.
[30℄ Mills and Lobo,Self-DiusionCoe ients, Elsevier, New-York, 1989.
[31℄ J.O.M.Bo kris,B.E. Conway,E.Yeager,Comprehensivetreatiseofele tro hemistry
- Volume1: The doublelayer, Plenum Press, New-York, 1980.
[32℄ Revil, A. Ioni diusivity,ele tri al ondu tivity, membrane and thermoele tri
po-tentialsin olloidsand granular porous media : A unied model, Journal of Colloid
and Interfa e S ien e 212 (1999)503-522.
[33℄ Y.Xi,Z.Bazant,L.Molina,H.M.Jennings,Moisturediusionin ementitious
mate-rials-Moisture apa ity anddiusivity,Advan edCementBasedMaterials1 (1994)
258-266.
[34℄ J. Bear,Y. Ba hmat, Ma ros opi modellingof transportphenomena inporous
me-dia. 2: Appli ations tomass, momentum and energy transport, Transportin Porous
under salinity: 1. Amathemati almodel,Transportin porous media11 (1993)
101-116.
[36℄ J.C. Kotz, K.F. Pur ell, Chemistry and hemi al rea tivity, Saunders College
Pub-lishing, New-York,1987.
[37℄ J.F. Pankow, Aquati Chemistry Con epts, Lewis Publishers, Chelsea, 1994.
[38℄ E. Samson, G. Lemaire, J. Mar hand, J.J Beaudoin, Modeling hemi al a tivity
ee ts instrong ioni solutions,ComputationalMaterialsS ien e 15 (3)(1999)
285-294.
[39℄ J.Rubin,Transportofrea tingsolutesinporousmedia: relationbetween
mathemat-i alnatureofproblemformulationand hemi alnatureofrea tions,WaterResour es
Resear h 19(5) (1983)1231-1252.
[40℄ R. Barbarulo, J. Mar hand, K.A. Snyder, S. Prené, Dimensional analysis of ioni
transportproblemsinhydrated ementsystems Part1.Theoreti al onsiderations,
Cement and Con rete Resear h 30 (2000)1955-1960.
[41℄ G.T. Yeh, V.S. Tripathi, A riti al evaluation of re ent developments in
hydrogeo- hemi al transport models of rea tive multi hemi al omponents, Water Resour es
Resear h 25(1) (1989)93-108.
[42℄ D.J. Kirkner, H. Reeves, Multi omponent mass transport with homogeneous and
heterogeneous hemi alrea tions: ee t of the hemistry onthe hoi e of numeri al
algorithm- 1.Theory, Water Resour es Resear h 24(10) (1988) 1719-1729.
[43℄ H. Reeves, D.J. Kirkner, Multi omponent mass transport with homogeneous and
heterogeneous hemi alrea tions: ee t of the hemistry onthe hoi e of numeri al
algorithm- 2.Numeri al results,Water Resour es Resear h 24 (10) 1730-1739.
[44℄ J.C. Friedly, J. Rubin, Solute transport with multiple equilibrium- ontrolled or
kineti ally- ontrolled hemi alrea tions, Water Resour es Resear h 28 (1992)
porous media,In: P.C. Li htner, C.I.Steefel,E.H.Oelkers (Eds.),Reviewsin
Miner-alogyVol.34: Rea tiveTransportinPorousMedia,Mineralogi alSo ietyofAmeri a,
Washington D.C.,1996, pp. 83-129.
[46℄ T. Xu, K. Pruess, G. Brimhall, An improved equilibrium-kineti s spe iation
algo-rithm for redox rea tions in variably saturated subsurfa e ow systems, Computers
& Geos ien es 25(1999)655-666.
[47℄ C. Andrade, Cal ulation of hloride diusion oe ients in on rete from ioni
mi-grationmeasurements,Cement and Con rete Resear h23 (1993)724-742.
[48℄ T. Zhang,O.E. Gjorv,An ele tro hemi almethodfora elerated testingof hloride
diusivity in on rete, Cement and Con rete Resear h 24(8) (1994)1534-1548.
[49℄ L.Tang,L.O. Nilsson, Rapid determinationof the hloridediusivity in on rete by
applying anele tri aleld, ACI MaterialsJournal 89(1) (1992)49-53.
[50℄ E. Samson, J. Mar hand, K.A.Snyder, Cal ulation of ioni diusion oe ients on
Solid
Solid
Solid
Solid
Liquid
Gas
V
o
V
o
L
V
o
G
Toward the center of the pore
Surface of the material
Stern
layer
Diffuse
layer
Free
water
Solid phase
Aqueous phase
Shear plane
Concentration
Potential
n
LG
Gaseous phase
n
LS
grad c
i
Figure2: Con entrationandpotentialprolea rossaporenear thesolid/liquidinterfa e
0
0.5
1
1.5
2
2.5
3
0
5
10
15
20
25
30
35
40
45
50
Concentration (mmol/L)
Distance from the surface (nm)
c
+
c
−
c(
∞
)
κ
−1
(a) + (1)= (1) =1.0mmol/L0
20
40
60
80
100
120
140
0
5
10
15
20
25
30
35
40
45
50
Concentration (mmol/L)
Distance from the surface (nm)
c
+
c
−
c(
∞
)
κ
−1
(b) + (1)= (1)=50.0mmol/LFigure 3: Con entration proles of a 1-1 ele trolyte near a harged surfa e al ulated
with the Gouy-Chapmandoublelayer model. The al ulations were made with asurfa e
potentialof 25 mV. The Debye length
1