In the

rate 01
asswnir of ste. paste

"diffuf
calcu1c coeffic

This mE as foll

to unr'e obtain contair

In the spec ime cond i ti i a solu profilE is app.
(6) :

This e, condit: c, . and th, c,

Obtain
11)

mechanisms, are

724

CHLORIDE DlFFUSrON. AASHTOTEST. RESISTIVITY

725

enetration

-J(X)

dC dt

dC(x) [ 1)

dx

This methodology presents some limitations which may be sumarized as follows: a) cement paste and not concrete is used which can lead to unrealistic results, b) it is time-consuming and takes weeks to obtain one result, c) a constant concentration in the chamber containing chlorides from the beginning should be maintained.
-

~~
) " . .,. ~~ In the case of non-stationary conditions concrete blocks or specimens can be used which results in much more realistic conditions. These concrete blocks are maintained in contact with a solution of constant chloride concentration and the chloride profile along the time is measured. In this case second Fick's law is applied to calculate the, an Apparent Diffusion Coefficient D
I

(f.
. ''/

(6):

oC(x)
-J(x)

at
solved

D.
applying the

(2)

This equation conditions:

is usually

following

boundary

c,=

C'I

x = 0,
condition:

> 0
(3)

and the initial cs= 0,

> 0,

o
which is the most widely used (7-

Obtaining the following

11)

solution

C.

------

C,

x
1 - erf -------2 (D.t) 112 [4 )

, This type of test also results very time-consuming and maintains several uncertainties on the rigorous application of Fick's law. Some authors (12) rather prefer to compare chloride profiles than calculate D. In addit' . hot lon, ln both steady and non-steady test conditions, it is 'ce- USually calculated the reaction or ads tion of chlorides y ,.me~Phases. Thisci c ns ance considered 0 mlnor lnfluence,
/'

726

C. Andrade

although few researchers (13-14) do take into account. ThuB Atkinson (3) refers to it by defining: a) a D.: Apparent Diffusio~' Coefficient in a porous medium, which considers the avera concentration gradients of the diffusing substance, therefore t~e. adsorption phenomena, and b) a D,: Intrinsic Diffusion coefficien~ which tries to take into account the average flux per unit of area and therefore, the volume fraction of porosity. ' There are also a variety of improvement proposals basic approaches (15-18) which make much more calculation of the chloride ~. to these complex

along 6 ho higher amoi concrete t(

Electrical Because of consuming, resistivity penetration (1) (20-29).

methods testing "natural" chloride penetration results timeattempts have been made to calculate the D. from' measurementes (3) (19) or to accelerate the rate of of chloride ions by applying an-electrical field

When the ( electrical mainly of f shOWS. Thi different: In additior ions occur: '. only chlor: spent in commented simul t.aneoi on the re L. Then the pr field is aj

,\,

..,:. ~ ... ...<:/

Both kind of test types will be commented in present paper in order to analyse their possibilities and limitations. First a critical review will be done on the test known as "AASHTO TEST" (30) explaining Why its mode of operation leads to erroneous conclusions. Secondly a brief summary will be made on the basic knowledge needed to understand migration phenomena in electrolytes. I Finally, a proposal will be presented on how to calculate Diffusion ,. Coefficient from electrical (migration) measurements and which are . the theoretical limitations. Numerical examples will be given. Extensive experimental trials will be needed to verify whether the assumptions taken in the numerical examples are reliable or not.

Elect:
a.1)

CRITICAL

REVIEW

OF THE RAPID

CHLORIDE

PERMEABILITY

TEST a.2)

That chlorides move quicker troughout the concrete when an electrical field is applied arose from earlier experiments, (31) (32) on that known at present as chloride removal (33~ (34). Actually, this fact on chloride migration was already experlenced. by many researchers using electrochemical techniques (as cathodic. protection (12) for instance). However, it was Whiting who, (2021) (30) proposed a "Rapid Chloride Permeability Test" in order tQ obtain in few hours an appraisal on concrete permeabili~y. Th! standard test has promoted a strong controversy (28) (29), wlth m~'~, heat than light in clarifying the meaning of the test an~ .. ability to predict concrete resistance to permeation of chlorldes. The real fact is that the test is increasingly being used althoU9
j

everybody

recognizes

some

still

unknown

uncertainties. cm) concrete ~i~~ in an arrangem& chloride (3t Pt of about 0.1 H ~ recommended to I coulombs record
I

Summing up, this test uses a thick (usually 5 between two electrodes (usually copper meshes) similar to that of the diffusion cell. Sodium weight) is added to one of the chambers and NaOH the other. Then, an electrical field of 60V is applied between electrodes and the amount of

der cal
30) -ous

sic cs.

.t .

an
.t s , <1)

Iced' ,die' 20to 'his

!IOre

its
les.

lugh

128

C.Andnde

As well as reaction,

oxygen

reduction

following

the

negative electrode: 202 + H20 + 4e- -> 40HAll these reactions tends to maintain the electroneutrality of the experiment which is one fundamentals of electrochemical reactions .

'!
OH

0) DIFFUSION

Mi

e
No--__ OH--C,No+--

.I TJ

,~
',-"

b) MIGRATION

~I

e
No --OH

Concrete

0
--CI-

No+-

c) DIFFUSION + MIGRATION

Figure 1. Mass transport processes in concrete.

.
-.--

e
'0 o
'0

'. o

o o' o

OH-

No+-

'. o

'.

2H20-01+2~
CI-CI2'

2H20-H2t.20HO2+ 2HtO 40H

CU-CU1

Figure 2. Processes occurring when an electrical field is applied in a diffusion cell: Joule effect, anode dissolution, electrolysis of the electrolite (gas evolution at electrodes and reduction reactions) and ionic migration and diffusion. b) MigrationThe third process happening in a cell is the movement of ions in the electrolyte in order to carry the electricity passing through the cell. Therefore, migration is developed and diffusion may appear if this migration leads to concentration differences. As it was mentioned, not only chlorides move but all ions take part in migration in a proportion what is known as their "transport or transference number". Transference number Let us -try to explain something here about this parameter. The transference number of an ion moving under the action of an external electrical field is defined by the "proportion of the current carried by this ion in relation to the current carried by the rest of the ions" (36). It is formulated as:
ij
ZjCj~

t;=

------ = ------------i

~ A

{5]

1: ZCh

Hence, the transference number is a function of the ionic mobility or the equivalent conductivity. This means that OH" ions will carry much more proportion of current than Cl' ones due to the ionic conductivity of 011 is 198,5 ohm" lcm1ecrl and that of Cl" is 76,34 ohm-I'cm1'ecrl(36). This fact is ve.ry important in the case of concrete because

C.Andnde

VaL 23. No. 3

it means that the main proportion of the current would be taken by OH' ions and not by the Cl- and therefore, hydroxydes might behave as a "supporting electrolyte". Therefore, only if chloride transference numbers are calculated is possible to specifically deduce chloride transport feasability, which is not taken into account by the Rapid Chloride permeability test which only records the total amount of current (that corresponding to the movement of all ions). In addition when flowing through the concrete the chlorides may react with the ~A and therefore a stationary flow cannot be achieved until all reactive sites are saturated. Movement of cations An additional aspect to be stressed now is related to the "anomalous" Diffusion Coefficient that is measured in the case of the cations of small ionic radius, as Na+ and K+ (1). This behaviour is very well described by Bockris (36) and Glasstone (37) considering that these ions migrate solvated, that is, due to its small ionic radius, Na + and K+ diffuse or migrate surrounded by water molecules, as they normally are in solution. That makes their movement more difficult and therefore, 0 values smaller than those of chloride ions are reported (1). This fact also explains why water may concentrate at the cathode, as was sometimes noticed in the case of cathodic protection. Na+ and K+ migration means that 'a net flow of water (electroosmosis) is also simultaneously produced. This fact can be also applied to explain the basic process of electrochemical realkalization: there, hydroxydes are produced at the rebar acting as cathode, and solvated Na+ ions move from the external carbonate solution in order to balance the electrical charges and, finally they support the reconstruction of a NaOH solution around the rebars. Anyway, as water is simultaneously reduced at the cathode together with oxygen, the dilution effect may be balanced. In the case of the migration test, having two chambers with solutions, the effect of increasing water around the rebar is not noticeable, but in the case of concrete (cathodic protection, realkalization or chloride removal) the effect will be dependent of the potential applied or the lasting of the treatment. A consequence of this effect in the case of cathodic protection, is that the ~ resistivity will increase at the anolyte (and therefore decreasing the efficiency of the anodes) .and a dilution of the solution around the cathode, and therefore a "bUffering" of the increase in pH value on the catholyte, may happen.
:

SUDU1

(db sidE

depi

a) ') at J to ele4 of ' Rap

1)

2)

3)

Thf

trc
"pI

AI'

Ch

mu
ne Wh

No "p

me
cc
01

tc
mf Pl
bl

CHLORIDe DIFFUSION. AASIITOTEST. RESrsnvnY

731

up what has been said up to now, when an electrical field eot current) is applied between two electrodes placed both SS of a concrete block, several phenomena develop as figure 2

iets:
The anodic material, if:l,ossiblei dissolves and gases may evolve ~bOth electrodes, b) aIr' ~ons :of the electrolyte move in order :oarry the current passing' through the cell and to maintain 'eotroneutrality c) in addition'heat is produced as a consequence the current flow. ' ", At the sight of these co~ents, it can be deduced pid Chloride Permeability test contains the following It accounts the total the chloride flow. current that, the errors: to

and not that corresponding

When integrating the total current from the beginning of the experiment it does not distinguish between chloride flow plus reaction and simple flow. The high voltage drop used (60v) induces in turn changes the flow speed. heat (23) (27) which on or

Therefore a migration test of this type cannot transport of chlorides (38) and much less "permeability" of the concrete specimen.
~he
J

at all inform on porosity

ic of

COEFFICIENTS

FROM MIGRATION

MEASUREMENTS

.d

;lc 'e,
le I Ite

Id,

,OH
is

th rs
he

te 1)

-How an attempt is presented on how to calculate, not he "permeability", but the Diffusion Coefficient from an electrical measurement similar to that described in the AASHTO test. Diffusion coefficient is the parameter which may characterize a concrete in order to predict its long term performance, that is, its resistance to the penetration of ions. The calculation of 0 from electrical measurements has to be based in the fundamental of transport processes in electrolytes, very well established in the traditional books of Electrochemistry Science (36)(37) (39-43). There, it appears that the general equation for transport processes in solution is that named Nernst-Planck (36) equation which can be written as:

or
is
he

r:-e

on a

oCj (x) = ~ ----------

oE(x)

oX

RT

--------- +

o(x)

GV(x)

(6J

D)

C)

, li
CHLORIDB DIFPUSION. AASfrrO

00. REsIS11V1TY

733

stationary flow - in order to apply equation [6] in its present formulation, a steady-state flow has to be established as figure 3 depicts. If non-stationary flow is produced, then the variation with the distance of the chloride concentration, should be also taken into account aiming to an equation in partial derivatives of second order similar to second Fick's law. Reaction - the first chloride ions traversing the concrete disc will react with A~ and therefore an erroneous D may be calculated as has been detected in the case of pure diffusion cells. In order to neglect this fact, the calculation of D has to be made when a linear increase of chlorides is recorded in the chamber not containing them at the beginning, that is to record the chloride flow when the reactive ACl was saturated with the first migrating Cl- (figure 4). Ionic strength - In order to take into account the high ionic strength of the concrete pore solution, two main aspects have to be considered: a) that activities instead of concentration must be considered and therefore, either a selective ion electrode for chloride is used or activities must be calculated, for instance as is suggessted in (44), and b) the influence of the ionic strength in the transport number, ~' and in the value itself of D~ should be considered. Let us again try to analyze this aspect of the influence of the ionic strength on D with more detail. It has been established that 0 is not a constant, but a function of the concentration of the solution (36) and therefore, high ionic strength influences D value. In a recent paper (44) a simple way to calculate activity factors from conductivity measurements has been offered, and therefore a trial will be presented in the numerical examples, on how Dcff variation with concentration is calculated.

opriate

:e flow

state
-re the
,kes

it

.asuzea

ths

of

enough

CI- in the cotholitt

TIME
f the

Figure 4. Flux (J) of chlorides leaving the cathodic chamber along time.

734

C.Andrade

Vol. 23. No. 3

Anyway, it is important to stress that being the concrete pore) solution a very concentrated one, the influence of ionici strength cannot be neglected in the calculation. . D) J~ule .effect - The po~ential difference applied to drive m~grat~o~ should be as h~gh to promote a quick enough movement of c~lor~des, and as small as;to avoid a waste in heating. Ten to f~fteen volts could be a sensible compromise.
4.

ell

of
di:

ef
Th

st

re
ch

Solving Nernst - Plank equation Really a rigourous solution for equation (6) cannot be achieved in solutions as concentrated as concrete pore solution (36)(43). In polielectrolytes (more than binary solutions) a rigourous application of flux equation (6) fails, an even more if the solution is concentrated, because 0 bas to take into account interaction of all ionic species. Therefore at least two main ' difficulties arise when facing our particular problem: 1) first that of the high ionic strength previously commented and 2) how to apply the equation to a particular ion and not to the solution as a whole (19). A semirigourous calculation might be undertaken using two possible aproaches: a) To consider phenomenological onsager's equations (36), or b) To use a Mean 0 for the whole electrolyte (43). Both approaches lead to unsatisfactory solutions for the sake of practical purposes. The first because needs many and sophisticated mathematical equations and the second, because does not allow to differentiate between the different ions. ThUS, with the aim of looking for a simple and practical solution, a simplified approach should be tried. This will be based in;; several assumptions able to overcome previous difficulties.} Simplified calculation of D~ The several assumptions which have to be taken into account order to solve equation [6J are (36) (39-43):

a
4}

cc
5. Tt m1

.~

iJ
TJ
0:

c' Once a expres 'I

in wh
exper alone
flux'

Calc1

only what happens inside the concrete disc is influencing measurements. This assumption may be accepted from the taot that ionic mobilities in solution are 3 or 4 orders 0 :' magnitude higher than in the concrete and therefore, tor t~~: sake of the measurement, the slowest process is the on . considered relevant.
2.

th: .

equa' Anot duri

43)

the

The term dealing with convection in equation (6) neglected. This seems not difficult to be accepted, what happens inside the concrete disc is considered. The diffusion component of equation (6) is consider. negligible in comparison to that due to migration. As well

3.

", I1
loRIDe DIFPUSION.

I
I

AASHTO TEST. RESIS1TVITY-

i
I' I'

electroneutr~lity in this experiment is maintained by'means' of the electrodic processes (gas evolution or ~et~~ dissolution) and therefore counter diffusion or membrane effects are not considered. The concrete disc is thin enough to allow to reach a steadystate condition is few hours, which in turns means that all reactive A~ in the disc is saturated with the first diffusing chlorides and therefore, after the transient initial period, a linear flux of chlorides along time, is established (figure 4), This allows to make the term E/a 1 = aE/l, being 1 the concrete disc thickness, and 6E the potential applied.

In;' rourou. if the;;'

!ved 3).

in:

LCCOunt;t;
o mahi' first '"
how to . Lon as

The concentration of chlorides in one chamber of the cell is much higher than in the other. That is, chloride concentration in the catholite should be high and that in the anolite, zero. This allows to accept that the concentration (activity better) of chlorides in one side of the cell remains comparatively constant. these assumptions are in the following way: considered the
6E
[7)

equation

can

be

)ssible

mol
Total where ions.
sake

ZF
= - ----

flux=

-J(------) sg'cm2 thickness

RT

1 of chloride

,-

_,.t

1= disc

and Ca is the activity

ot
The equation may be also writen:
J R T 1 ---------z F Ca aE
J !

:icated
llow to

..J

li

lution,
sed in
J

Our

[8)

in which all parameters are known and J can be calculated from an experimental test in which the amount of chlorides is monitored along the time. Thus, from a plot similar to that of figure 4, the flux J can be calculated from the slope of the linear part. Calculation equation of
Ddi'

runt; in from the value of intensity. Nernst-Binstein ing the .1e fact
lers of

I,

for the le only can be if only sidered swell,

Another way to calculate D~ is from the recording of the intensity during the experiment, because it is also well established (36) (3943) that the flux of a migrating species is also proportional to the total intensity: J= nF being the transference number (of the chloride in this particular 'tease)the proportionality factor. Therefore, on the contrary from he value of J obtained in an experimental test, it is possible to
[9)

736

C.AndrIde

Vol. 23. No. 3

calcu~ate th~ transference number ~, and by substitution equat10n [9) 1n equation [7), the following is obtained:

of
rJlst-PJ

-----nF

ZF
[10)
RT

Figure 5. catholytf

This expression results similar to Nernst-Einstein (32) equation but applied for a single ionic species through the use of the transference number:
RT RT

fol1owin~ cross-sel
D

1 A

1 [ 11)

R
F

10-8

being A = cross section area of the concrete disc. This equation opens the door to the possibility of calculation of Diffusion Coefficients from a simple measurement of resistivity or conductivity providing that tj of the particular ion could be accurately calculated (see numerical example later on). This approach would represent a very promising 'simple way for the future, if the theoretical difficulties of calculating accurate transference numbers of chloride ions in concrete, could be solved. Even, as it will be presented at the numerical example, this equation serves for an approximate calculation of the order of magnitude of D from water saturated concrete resistivity values, once proper account of the influence of the ionic strength of pore solution is considered (3)(44)(45)(46). Finally it has to be mentioned that, if the calculation is made from resistivity measurements the reaction with the ACJ is also' misled and therefore, the values obtained are those of D~ and not~
of ~. .

= =

1.98'
2306

Applied Concrete Consider produced

Cl

This is in this testing

In figu:

with the is giveJ can Lncr a poter diffusi,

As an
E

order
0.1 M NoOH

0
I

w/c , 0
0.5 M Noel

by the result

Calcul

O.~m

Figure 5. Simple representation of the cell for testing migration.

.
'I"

;'

... ..:..

,,'i",;'

,,'

.~rnst-PlaDk

equation

Figure 5 depicts an example in which NaCl 0.5 M is added to the catholyte of'a igration cell and NaOHO. 1 M to tl\e;iiil.n-d~yte. The tollowinq parameters are assumed: cross-section area: 30 cm' 1.987.2cal mol-I.<1\1 F c 2306 calvoltleq-1 Z ::1
R"::

o ==.

0,

cm2'Sg-1
'#' .... ''''.

ion ion or
be his

:J
12 V
~,1

, >J''''

Applied potential between electrodes dE concrete~thickness I = 0,5 cm.

'nt-

,n .

the
ate ed.

considering the activity equal to the concentration, the flux produced for a concrete having a Dctf of 10" cm2.s1wo~ld be: (1)(23063) (l0")(0.5XIO)(12)
J

his
of
.es ,

lore
iade
1150

= --------------------------------= (1.9872) (293)(0.5)

mol 0,47 x 10" ------s'cm2

This is the order of magnitude of the flux that will be recorded in this kind of experiments, and a period of several hours of tes~ing a~e needed in order to minimize errors in monitoring it.

not

In fi~e 6 a qraphic representation of the variation of flux J with'those of D~, C and dE being the rest of parameters constant, is given. It is apparent that an increase of ~E from 2 to 20 volts, can increase one order of magnitude the flux. As was refered in (1) a potential of 2 volts almost does not influences the pure diffusional flux. As an ~ample too, in (47) chloride fluxes are reported of the order of ~bout 10 mmol/day for a concrete cured at 20C and with a w/c 0,5. Thus taken into account the experimental conditions used by.the aut~ors and assuming an stationary flow, the Dctf values would reult the sensible value of:
-e .f'4,~;, J ;, ,;,1 i lItfi.'!lt r:

- ~- t: .,. G 1; intensity values

t. 1L

~t E -,~ ;:, ;/

calculation;f~om

1",:~ ': '",

t~~ theoretical
of:

A~lculation of chloride transfer~nce ~umber in a s,oluFionO.2M NaOH to. 5M NaCI from equation [5]~will q1. v~ a value
. . ."', ;
, (0.5) (76.34)

--(~~~)(~~~~)~~~~~)(;~~;~)~(~~;)(~;;r~(b~i 0.338

Fiqt

SUlI1lII

ohlo reco

Alte
of' c

" - stat is a

DCI~cmi'sg )

t:>

o=O.lmol/cm3 0 =0-35 " 0 =0.5 "

NERNST - EINSTE IN EQUAT ION

Figure 7. Graphic representation of Nernst-Einstein equation (10) in function of chloride activity values. summing up, Nernst-Plank equation has to be used when values of chloride flux along time in a Migration Cell, are accurately recorded, providing steady-state conditions are established. Alternatively, Nernst-Einstein equation may be used when instead of chloride flux, intensity values are accurately recorded, steadystate conditions are operating and the chloride transference number is also accurately calculated.

CONCLUSIONS

More than conclusions, the following paragraphs are a summary of comments which can be drawn up from a thorough study of basic books on electrochemistry and a careful meditation on their application to the particular case of concrete. The Rapid Chloride Permeability Test (AASHTO) in its present formulation cannot inform on concrete permeability to chlorides. The recording of the total current passing across the cell is a function of the amount and type of ions, but not of the chloride flux or chloride mobility. The calculation of ionic migration can be only rigorously ressolved in homogeneous, binary and dilute solutions. Concrete and concrete pore solution is a polielectrolyte with high ionic strength and therefore, a rigourous calculation cannot be performed or results very difficult.

740

C.Andnde

Vol. 23. No. 3

Vol. 23;'No.

However, as approximate v~lues may be enough for practical purposes, simplified way~" of calculation of Diffusion Coefficient of chlorides' may be tried. This supposes the acceptance of some assumptions and uncertainties. Thus, assuming some simplifications, Nernst-Plank and NernstEinstein equations can be used in a disposition similar to a diffusion cell (migration cell): the main being to accept that convection does not operate inside the concrete and that diffusion is negligible compar-edto migration when electrical fields higher than 10v are operating. From equations, (8) and (11), Effective Diffusion Coefficient, DefT' can be calculated in an experiment of few days. NernstPlank equation can be used.when only chloride flux along time is recorded and Nernst-Einstein equation when intensity values and chloride transference numbers are accurately applied and measured respectively. An extensive experimental program, which at present is being carried out by the author, is needed to check whether the assumptions considered are correct or not. Finally it has to be mentioned that the same equations can be applied to calculate ionic movements in the case of cathodic protection, chloride removal or realkalization, although in these cases a non-stationary process is established which makes more sophisticated the solution of equation (6) and (11). specially seems very attractive the ressolution in the case of chloride removal, because it may give the theoretical time needed to decrease the amount of chlorides below a certain threshold.
ACKNOWLEDGEMENTS

(5) (6) (7) ~: (8)

.:
(

I
}

'1
~

(9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27)

I 1 I
] ] ] ]

The author is grateful to several researchers for their comments and experimental trials. First she is grateful to Dr. J. Galvele of Argentina for the discussion on the preliminary states of the paper and the comments introduced in the final version. Thanks are also given for the discussions to her colleagues at the Institute: Dr. S. GoBi and Dr. C. Alonso~ and finally to Mr. M.A. sanju~n for his experimental trials.

BIBLIOGRAPHY

Jfl)
(2)
(3)

(28) S. GOTO, D.M. ROY - Cement and Concrete Research, lA pp 751- I' 757 (1981). G2.ool3CoJ c,"33~ M. COLLEPARD1, A. MARC1ALIS, R. TURRIZ1ANI - 11 Cemento, ~, . pp 157-164 (1979). , A. ATKINSON, A.K. N1CKERSON - Journal of Materials Science .a,: pp 3068 - 1078 (1984). '. C.L. PAGE, N.R. SHORT, A. EL TARRAS - Cement ~, pp 395-406 (1981). (29) (30)
(31)

(32)

I
t'

oHLORJDE DIFFUSION. AASHTO TEST. RESISTIVITY

JJ

,t

.,. ',. ;;"v~:

i"'1'

:tic~l 'usion s the

rnst_
to a : that that rical ient,

rnsttime alues d and

t is ether

be

hodic these more

ially

oride
d

to

ments Ivele f the

s are
tute: n for
I

751r

67,

Res.

N~R~ BUENFELD~ J .B. NEWMAN - Materials and structur,es, (RILEM), ~, pp 3-10 (1987). ", J. CRANK - The Mathematics of Diffusion, Ed. Oxford University (1975) . ! J:li~;DIAB,. BENTUR, C. HEITNER-WIRGUIN, A L. BEN-DOR- Cement and );P9~erete Research, 18, pp 715-722 (1988). (8')' fR'.D>;,BROWNE - Perforance of Concrete in Marine Environment ACI SP-65, pp 169, 204 (1980). . (9) K. BYFORS - Chloride initiated reinforcement corrosion - CBI Report 1/90 - 121 pp (1990). (10) R.K. DHIR, M.R. JONES, H.E.H. ARMED - Maga~ine of Concrete Research, 43, pp 37-44 (1991). (11) K. TUUTTI - Corrosion of Steel in Concrete - Doctoral Thesis, Swedish Cement and Concrete Institute, Stockholm, 469 pp. (1982). ;:'~2) .O.E. GJORV, O. VENNESLAND - Cement and Concrete Research, i, pp 229-238 (1979). -'cem~nt and Concrete Research 19, pp 683 - 691 )6(13) J. TRITTHART (1989). ~rt" ~e:IIf\l\.k M.T. CHAIEB - Materials (14) L. HACHANI, E. THIKI, A. RAHARINAIVO, and Structures (RILEM), 24, pp 172-176 (1991). (15) P. SCHIESSL - Diffusionsmodell zur rechnerischen Erfassung der Chloridionendiffusion in Beton - IBS - Bo. 238 (1983). __ (16) R.K. DHIR, M.R. JONES, A.M. SENEVIRATNE - Cement and Concrete Res.,' 21,.pp 1092-1102 (1991). (17) Y. MASUDA - 1st. International RILEM Congress: Durability of Construction Materials, vol 3, Paris, pp 935 - 942 (1987). (18) A. RAHARINAIVO, J.M. GENIN - Bull Liaison LCPC, nQ 144, pp 5964 (1986). (19) N.R. BUENFELD, J.B. NEWMAN - Magazine of Concrete Res., 36, pp 67-80 (1984). (20) D. WHITING - Public Roads, 45, pp 101-112 (1981) (21) D. WHITING - American Concrete Institute SP-82-25, pp 501-524 (1981) ;-(22) S. LI, D.M. ROY - Cement and Concrete Res., 16, pp 749-759 (1986). J.G. CABRERA, P.A. CLAISSE - Cement arid Concrete Composites (23) 12, pp.157-161 (1990). (NACE), 41, pp 93-100 (1985). (24) Y. GAU, I. CORNET - Corrosion _ (25) T.C. HANSEN, H. JENSEN, T. aOHANNESSON - Cement and Concrete Research, 1, pp 782-784 (1986). . ~(' (26) C.M. HANSSON, B. SORENSEN - Corrosion Rates of Steel in Concrete. ASTM STP/1065, N.S. Berke, V. Chaker, D. Whiting Ed, pp 3-16 (1990). (27) M. GEIKER, N,' THAULOW, J. ANDERSEN - 5th International Conference on Durability of building Materials and Components, Brighton, U.K., pp. 493-502, Nov (1990).. . (28) R.J. DETWILER, K.O. KJELLSEN, O.E. GJOR - ACI Materials Journal, 88, pp 19-24 (1991). . (29) P. RECHBERGER - Zement-Kalk-Gips, 38, pp 679-684 (1985). (30) D. WHITING - Rapid Determination of the Chloride Permeability of Concrete. Report No. FHWA/RD-81/119, August 1981, NTIS DB No. 82140724. (31) A.A. HACHEMI, M. MURAT, J.C. CUBAUD - Revue des Mat~riaux de Construction, nQ 700, pp 149-155 (1976) (32) V.K. GOUDA, G.E. MONFORE - Journal PCA - Research and Dev~ Lab. L, pp 24-36 (1965).
~.

742

C.Andnlde

VoJ. 23, No. 3

cJ!MBNT md CON (l(108-8846fl3. $6.0

j
'I ,

(33) G. GRIHALDI, J.C. LANGUEHARD - Bulletin de Liaison des Laboratoires des Ponts et Chauss6es, pp-79-84 May-Jun (1986). (34) J.E. BENNETT, T.J. SCHUE - Corrosion 90. NACE. Paper No. 316, (1990) . (35) J.A. BABOR, J. IBARZ - Qulmica General Moderna - Manual Marin Ed. Barcelona - Spain - pp 483-86 (1958). _ (36) J.O'M. BOCKRIS, A.K.N. REDDY - Modern Electrochemistry Plenum Press Ed. New York (1974). (37) S. GLASSTONE - Textbook of Physical Chemistry - Van Nostrand Ed. New York. (1947). of Concrete as a (38) H.K. HILSDORF, J. KROPP - Permeability criterion of its durability - Report of RILEM Technical Committee TC 116 (1992). L.R. FAULKNER Electrochemical Methods. (J9) A .J . BARD, Fundamentals and Applications - Jon Wiley & Sons Ed., (1980). COSTA Fundamentos de Electrodica. Cinetica (40) J.M. Electroqulmica y sus aplicaciones. Alhambra Universidad Ed. Spain (19B1). (41) W. FORKER - Cinetica Electroqulmica - Eudeba Ed. Buenos Aires - Argentina (1986). Group - Instrumental Methods in (42) Southampton Electrochemistry Electrochemistry - Ellis Horwood Series in Physical Chemistry Ed. (1990). Systems - Prentice Ha 11 Ed. (43) J. S .NEWMAN - Electrochemical Englewood Cliffs, New Jersey (1991). S. GONI, C. ANDRADE - Ceramic Transactions. (44) A. MORAGUES, Advances in Cementitious Materials. S. Mindess Ed., 16, pp 5765 (1991). E.J. GARBOCZI - Cement and Concrete Rese~rch, 20, pp 591-601, (1990). P. LAY, P.F. LAWRENCE, N.J.M. WILKINS, D.E. WILLIAMS - Journal of Applied Electrochemistry 15, pp 755 755-766 (1985). (47) R.J. DETWILER, K.O. KJELLSEN, O.GJORV - ACI Materials Journal, Jan-Feb, pp 19-24 (1991).

The aut cement pas sulphate so magnesium have been I

One ar

appears to was produc established seawall th

4MgO.SiO:

'r.

appeared t upon concr The M about corn during the secondary Bonen all( highMg v, own long t (9). From MSH in, Overs contributk Portland I reasons c the prese Sulphate,