CEMENT and CONCRETE RESEARCH. Vol. 14, pp. 622-630, 198~. Prlcted in the USA.

0008-8846/SL $3.00+00. Copyrigh: (c) 1984 Pergamon Press, Ltd.

TI~I DISPERSION HODEL FOR HYDRAIIOK OF PORTLAND GEt-lENT

I, GEXERA[ '.r)\.EfPlt;

l o r b e n Kr~udsen

]he l e c h n i c a ] U n i v e r s i t y o f Denmark,
DK-28OO l y n g b y , Denmark

(Communicated by D.M. Roy)

(Received Sept. 7, t983)

ABS1 RA(:I

This paper introduc'er~ an a r ~ a l y t i c a l m a t h e m a t i c a l model f o r d e v e l o p -

ment o f the h y d r a t i o n o f P o r t l a n d cement:
t o + t l A + t2A2 = t

t i s t i m e , and t o , t. 1 ar,d t 2 are t i m e - c o n s t a n t s depending or~ temperat.ur~,

cement t y p e , a d m i x t u r e s e t c . A i s the h y d r a t i o n r a t i o , i . e . the r a t t o
o f h y d r a t e d t o u n h y d r a t e d cement. The d i s p e r s i o n model has been d e r i v e d
on b a s i s o f the t y p i c a l part icJe s i z e d i s t r i b u t i o n for t u b e - m i l l e d pr~-
duct.~;. E a r l i e r f i n d i r , g s by us have proven thu p a r t i c l e s i z e d i s t r i b u t i o r .
to be a dominant f a c t o r irl the c o r r e c t m o d e l l i n g o f cement h y d r a t i o n .
The p a r a m e t e r i z a t J o n o f h y d r a t i o n c u r v e s , which i s ea.'~i]>, performed w i t h
the d i s p e r s i o n model, g i v e s a s u f f i c i e n t c h a r a c t e r i z a t i o n o f these f o r
purposes o f p r a c t i c a l i n t e r p r e t a t i o n s o f f.he effect'.s o f f a c t o r s d e c i s i v e
f o r the development o f h y d r a t i o n .

i i,t.roduc t 1on
I h i s paper d e s c r i b e s a m a t h e m a t i c a l model f o r the development o f the hy-
d r a t i o n o f P o r t l a n d cements, lhe c o n t e n t o f the model d i f f e r s i n many r e s p e c t s
from e a r l i e r a t t e m p t s t o modc'] cement h y d r a t i o n , but a compaTiSOr, to o t h e r r.,c;--
d e l s i s o u t s i d e the scope o f t t l i s paper'. The mode] o f h y d r a t i o n p r e s e n t e d here
i s the r e s u l t o f our p r e v i o u s work w i t h cement h y d r a t i o n ,,1,2). These works
were i r l s p i r e d by the works o f l a p ] i n t 3 ) . He was the f i r s t t o p o i n t t o the im-
p o r t a n c e o f the p a r t i c l e s i z e d i s t r i b u t i o n (psd) i n the m o d e l l i n g o f cement hy-
d r a t i o r ~ . The i n f l u e n c e o f psd or~ the shape o f the h y d r a t i o n curve has i n our
work been b r o u g h t irate e x p l i c i t form i n the d i s p e r s i o n model o f h y d r a t i o r , , ft~e
model appears i n a m a t h e m a t i c a l a n a l y t i c a l form, which can e a s i l y be used f o r
p a r a m e t e r i z i n g h y d r a t i o n cur~es.
l h e d i s p e r s i o n i [i p a r t i c l e s i z e s in a m a t e r i a l l i k e cement i s such, t h a t the
s t a l e o f r e a c t i o n o f the s m a l l e s t p a r t i c l e s i s s i g r ~ i f i c a n t l y d i f f e r e f l t from the
s t a l e o f r e a c t i o n o f the l a r g e s t p a r t i c l e s . While the s m a l l e s t p a r t i c l e s may be
f u l l y h y d r a t e d , the l a r g e s t p a r t i c l e s may have a low degree o f h y d r a t i o n . ]r~
fac?t, at any time o f h y d r a t i o n t h e r e e x i s t s a d i s t r i b u t i o n o f degrees o f h > d r a -
622
Vo]. 14, No. 5 623
DISPERSION MODEL, CEMENT HYDRATION, CONCEPTS, PARTICLE SiZE

tion, corresponding to one point on the hydration curve. Thus it may be under-
stood, that the development of hydration of a single particle can not directly
be discovered by the hydration curve. For materials with a broad psd like an
cements it can be shown by numerical means, that only gross features in the ki-
netics governing the hydration of a single particle, may be discovered by the
hydration curve. A change for instance from constant rate kinetics to surface
reaction controlled kinetics, both being linear kinetics, makes no significant
change in the hydration curve obtained by calculation. ]his important observa-
tion wi]i be designated the "dispersion rule"
Thus the inf]uence of psd may be considered a c:omp]exity in the study of
hydration kinetics per se. On the other hand, in ~Jttemptsto mode] the hydration
development of cements it constitutes a simplifying factor. According to the
dispersion rule differences within a size class in morphorlogy, t:hemica] com-
position etc. can hardly be recognized in the hydration curve. The hydra-
tion curve is sensitive mainly to the psd (vith the exception mentioned later),
and the psd-curve for tube-milled products is smooth and simple in form.
Now what is left of kinetic Jr,formation in the hydration curves?
[4ainly two features:
i) Uhen hydration curves are plotted as function of log-t, the position
of the curve on this axis determines the value of the rate-constant
(The situation is more complicated when 2 rate-constants govern the
hydration development).
2) The two important groups: linear and parabolic kinetics, can be recog-
nized in the hydration curves.
In order to appreciate the last point one should note that a11 kinetic equations
leaning on the shrinking-core model contain the variables r,t and k, being the
particle size, time and the rate-constant respectively, in either of two com-
binations: 1

linear: kt ; parabolic: (kt) (i)

r r

This fundumentaI difference between iinear and parabolic kinetJc~ equations

"overrides" the dispersion rule since it constitutes a difference in the way
r and t enters the kinetic equations. The fact that r and t enter the kinetic
equation in ratios is not a mere coincidence. It is a result of shrinking-core
kinetics, and it guarantees that reaction curves belonging to different partic-
le sizes never cross. While the dispersion rule may be said to be prir,arily
responsible for the simplicity in modelling hydration with the dispersion model,
the simplicity is also the result of a deliberate ]imitation in its range of va-
lidity. As can easily be verified by noting the form of the dispersion mode],
this model does not give a correct description of the early acce]ieratory part
of the hydration curve, i.e. the interval 0-15% is outside the reach of the mo-
del, and should be left out in numerical work with it. The model in its pre-
sented form models this part by shifting the zero-point of time to t O . A closer
agreement in the range 0-15% can be obtained by using a more correct formula
for psd (see eq.2 in ref. 1) than used in the derivation presented in this pa-
per. But the new parameter entering the model, as a result of this modification
of the formula for psd, depends only on the correct description of the finest
5% of psd, and contains no further kinetic information. Therefore we feel, that
the loss of close a9reement in the first 15% of the hydration curve is not a
serious loss, and the primary'goal with the dispersion model is a general ~Jnd use
ful characterization of hydration curves, and not the optimal accuracy in mo-
de]ling all parts of the hydration curve.
624 Vol. 14, No. 5
T. Knudsen

Outlines of the derivation of the dispersion model of hydration.

Due to the earlier mentioned effect of the
crucial point in modelling cement hydration is
particle kinetics to the hydration of the total
By assuming independent reactions of particles
presented by:
co
dispersion of particle sizes the
the transformation from single
ensemble of particle sizes.
this transformation may be re-

cz(t) = ~ w(r) ~(r,t)dr (2)

./
0

where ~(t) is degree of hydration of the ensemble of particle sizes, u(r) is

t!,e frequency t~';~ and ~(r,t) the degree of hydration of a particle of size r.

e(r,t)
Much mathematical modelling of hydration in the literature has been concer-
ned with theories about ~(r,t). Comparison to experimental data has often been
done by putting r=#, where r is the mean particle size of the material in
question. This procedure may be very misleading, if the psd of the samples
are not sufficiently narrow (2). Theories involving specific assumptions as
to the form of ~(r,t) must be compared with hydration data from narrow-si-
zed samples. The frequently referenced data presented by Kondo and Ueda in
Tokyo symposium (4) is not narrow-sized as far as modelling with r=# is con-
cerned. This was shown in ref. (2).
faplin seems to have been the only author who earlier suggested a solu-
tion of eq.2, which was not purely numerical (5). He used a log-normal distri-
bution for psd, and his very elegant index-of-reaction equation for ~(r,t),
but his solution is in form of a series-development, which can hardly be cal-
led analytical in the sLricter sense of Lhe word.
Some of the reason for the difficulty in suggesting analytical solutions
to eq.2 lies in the form of ~(r,t). All suggestions, so far, as to the form
of ~(r,t) has a time of terminaLion. From Lhat time on the ~(r,t)-curve is a
straight line with a=l. Such a curve, composed of an ascending part and a con-
slant part, can hardly be given a simple analytical form, suitfull for the so-
lution of eq.2.
In our case we have chosen an analytical form for ~(r,t) with the property
~(r,t)<l for all values of t:

] for linear
e(r,t) : exp(- r .) ; i : (3)
"'~ktj2 ~ for parabolic
2

whrre k is a rate-constant. This form resembles quite closely the "classical"

forms up to ca. 70% of hydration, but thereafter ascends more slowly towards
the asymptote ~=I. It is a c:onvenient mathematical form for the solution of
eq.2, but so far no direct physical interpretation can be given of it. A choice
like this, of an ~(r,t) which disregards a physical content, is relying on the
dispersion rule, which says that the hydration curve is insensitive to the form
of ~(r,t). All, which is important, is that the independent variables should
appear in the combination as given by eq.l. This is not the whole truth thoughZ
The use of the dispersion rule should be seen in the light of the convolution
principle as derived in (]), but a closer evaluation of this point is outside
the scope of this introductory paper.
Vol. 14, No. 5 625
DISPERSION MODEL, CEMENT HYDRATION, CONCEPTS, PARTICLE SIZE

w(r)
As to the form of w(r) appearing in the integral of eq.2 i~ is a lucky co-
incidence, that the psd of tube-milled product like cement may be modelled in
95% of its range by a form as simple as:

w(r) : -1 e x p ( - _r )
r0 r0 (4

uhere r 0 is a constant in the equation depending on fineness. This form is sui-

ted for the integration (eq.2) with ~(r,t) given by eq.3. Its major draw-back
is the lack of agreement for the finest 5% of the material. The error is such
that it leads to infinite specific surfaces. This is of no serious practical
consequence though, s!nce experiments show i/r 0 to be directly proportional to
B1aines measure of fineness (6).

Solving the integral

By insertion of eq.3 and eq.4 into integral of eq.2 ue arrive at the
solution:

t :~
for linear: ~(t) = t + t I ; tl kI (Sa)

for parabolic: ~(t) = ~/t ; t 2 = r_~ (Sb)

+ v~ 2 k2

Both eq. 5a and eq.Sb are simple hyperbolas in t and ~/-t respectively. The time-
dependency in eq.Sa is well-known in drying shrinkage and creep, and has been
suggested a few times for modelling the strenght development of concrete (7,8,9).
Why it has never found wide-spread application in practice within hydration mo-
delling may be due to two reasons. Firstly, eq.Sa covers only one group of hydra-
tion curves encountered in practice. The other group obeying eq.5b is equally
important. We find that hydration curves for normal Portland cement are modelled
by eq.5a, whereas eq.5b should be used when modelling the hydration development
of for instance white cement. Secondly, the importance of introducing the correct
zero-point for time has not been recognized in much mathematical modelling of
hydration (lO), (Carino in ref. (9) correctly introduces in). The proper use of
eqs.5 depends on the correct choice of zero-point. Thus th~ time appearing in
eqs.5 is "reaction time", and should be understood as t -t~, where t is the time
. . . . m . m
from mixing. The reason for the introduction of t O in modelling of hydration is
the obvious one, that the dormant period is not included in the modelling, ex-
cept as a delay in the start of the hydration development. (In the equations to
follow t n will appear explicitly and t takes on the meaning of time from mixing).
In m6st practical cases the use of eqs.5 is dependent on another adjustable
parameter besides t I (or t 2) and tO . This parameter appears in connection with
the definition of ~. Since very few data in the literature contain direct mea-
surement of ~ (as obtained with XRD), properties like heat evolution, non-eva-
porable water, chemical shrinkage, etc. have to be relied on. The use of such
data for the modelling of hydration require a knovledge of P (the property mea-
sured after infinitely long time) or the estimation of this parameter by adju-
626 Vol. 14, No. 5
T. Knudsen

sling asP/P= until the best fit with eqs.5 is obtained. How this is done in prac
Lice will be described in the following chapter.

The final form of the dispersion model

For the purpose of arriving aL the final form of the dispersion model we pre
fer to rewrite eqs.5:

linear: t 0 + LIA : t (6a)

parabolic: t0 + t2A2 : L (6b)

where Ase/(1-e) = P/(P -p), and t in contrast to eqs.5 take on the meaning of
Lime from mixing, since t0 is nov explicitly entering the equations. Written in
this way the designation linear and parabolic Lake on a direct formal meaning,
besides refering to the underlying kinetics. A is called by us the hydration
ratio as opposed to e the hydration degree.
Experiments in this laboratory using chemical shrinkage of Portland cement
paste, with the admixture CaC12, triggered the fusion of eq.6a and eq.6b to the
final form (11):
t 0 + L1A + L2A2 : t (7)

Eqs.6a and 6b are still contained as limiting cases of eq.7, when either t I o~
t 2 Lakes on the value O.
The hydration experiment with additions of CaC]~ showed, that by adding from
2 to 6~ by weight of cement, kinetics changed from~linear (eq.Sa) to parabolic
(eq.6b). In between none of these limiting cases could produce satisfactory
fits. The combinatory eq.7 gave excellent fits for all additions: 2,4 and 6%
(see fig. 1). The parameters tI and t 2 changed in this interval in such a way,
that the ratio tl/tl+t ~ changed its value from approximately 100% to approxi-
mately 0%. We defined ~his ratio as degree of linearity:DL.
The derivation of eq.7 from first principles is not complicated, but will
not be given in this paper. In brief, it considers two separate resistances de-
pending on the linear and parabolic part, and makes series-coupling as with re-
sistances in an electrical curcuit. This is done for kinetics of the individual
particles, whereafter the transformation by means of the integral in eq.2 is
performed. The fact that eq.7 comes out as a linear combination of eqs.6a and
6b is a result of the particular form assumed for linear and parabolic kinetics
in eqs.3.

Concluding this chapter we find it important to stress, that although

the form for eq.7 is such, that the parabolic element L2A2 wi]] be governing
for sufficiently long times, numerous hydration curves ~odelled by us could
in practice be considered 100% linear. A change from linear to parabolic ki-
netics as pictured by eq.7 may come so late (small va]ues of L 2) as to have
no practical implication.

Parameterizing hydration curves with the model

In the following a brief description will be given of the use of the dis-
persion model for paramelerizing hydration curves. Examples of the success of
the dispersion model in model]ing hydration curves for cement will not be given
in this paper, except for the example presented in fig.l. Reference will be
made to earlier presentations concerning the applicability of the dispersion
model in the form of eqs.6 (1,2,]2), and to a recent Ph.D. thesis for a more
thorough documentation including the final form eq. 7 (11).
Vol. 14, No. 5 627
DISPERSION MODEL, CEMENT HYDRATION, CONCEPTS, PARTICLE SIZE

Chemical Shrinkage Chemical Shrinkage

RPC 20"C 2%CoCIz/g cement RPC 20"C 4,~CCoCI~/g cement
P P,:5.47, to:2, t,: 17.7, t~: 1.1 P P. 5.41, toO.5, t,: 15.9. t 2 11

i- : . ~lplr lion
&O
,o I ........... :77
-....,.,

0£.: 94% DL: 5 9 % 2/

SO
,I.0 /,

40
}

I I0
O0 1 i
'°i
10

O0
!
~ . . . . . r ..... I . . . . . I--
, ,o %° ,~ ,~ I I0 100 1000 10000

t (log scale) t (log sca'e)

Se' I¢,. S t l
f4~i O,,i,.~.~;o~ w,,d,w. , ~ w o , PSO
8.~-c41- I I

........ Chemical Shrlnkog~-. . . . . . . . . . . ~]

RPC 20"C 6%CoCl2/g cemen| L
P P. 5.15, to2.S, f,: 1.5, t2:34 [
Fig.1 Measurements of chemical
SO
shrinkage of rapid hardening
Portland cement (w/c=0.5), fit-
40 ted by the dispersion mode].
The degree of ]inearity DL is
30 seen to change from 94% to 4%
by addition of from 2-6% CaC] 2
by weight of cement. Time-
20
constants are in hours and

l~-~'-,o chemical shrinkage P in ml/loo g

cement.

O0
1 tO 100 loOP
t (log score)
o,,~i ~ w, s11l

Regression calculation
The dispersion model of hydration, as given in eq.7, presents itself as a
parabola in A and t. The fitting to experimental data may therefore be performed
as a simple polynomial regression, higher order than 2 being discarded. The pa-
rameter P=, entering the definition of A, can be determined by stepwise, re-
peated regression calculations until a value of P= produces a minimum in the
residual variation calculated at each step. As mentioned in the introduction,
points belonging to the interval O-ca. 15~ in degree of hydration should not
be inc]uded in the modelling.
Our experience with regression analysis using eq.7 has shown, that the cor-
rect assignment of dependent and independent variables is a crucial point in
the analysis. We found, that if t is used as dependent and A as independent va-
riables, too much emphasize is put on points with large t's on the expense of
the lower part of the hydration curve. Thus for purposes of regression calcu-
]ation ve rewrite eq.7:
628 Vol. lq, No. 5
T. Knudsen

1 = to(l/t) + tl,iA, t ) + t2(A2,,'t~ 8)

When this form is used in regression, approximately equal weight is put on all
points.
Sometimes a visual inspection of points in the late part of the cur~e may
prove useful. Points deviating too much from a smoothly dravn curve may be
d e t r i m e n t a l to r e g r e s s i o n i n t h i s p a r t o f the curve, and are b e t t e r l e f t out
off the a n a l y s i s .

A g r a p h i c a l method
A g r a p h i c a l method may be a p p l i e d f u r the d e t e r m i n a t i o n o f the parameters
L o , t ,t~ and P in the d i s p e r s i o n model ] h i s method uses a d o u b l e - l o g paper
and g t~ansparent s h e e t . Place the sheet on top o f the l o g - p a p e r , and draw a
smooth curve through the data p o i n t s . Nov t r a n s l a t e the sheet along the l o o -
P a x i s (P i s the neasLred p r o p e r t y ) ui~tJl the f o l l o w i n g e q u a t i o n i s s a t i s f i e d :

t80 : 16t20 27t40 + 12t60 (9)

where t ? n , t / , n , t 4 n and tRn are times ( as read from the c u r v e ) c o r r e s p o n d i n g to

20,40,60 an~ 80.0 h y d r a t i o n . Now P~o i s the v a l u e on the sheet f a l l i n g at 100%
on the d o u b l e - l o g paper, and t o , t 1 and t 2 may be c a l c u l a t e d by:

t O : 1.4t20 .44t50 + .022t80

t 1 = -1.8t20 + 1.9t50 - .]ltso (10)

t 2 = .36t20 .44t50 + .089t80

I n a c c u r a t e or incomplete data sets

The success of" m o d e l l i n g w i t h eq.7 depends on the accuracy and complete-
,less of the data s e t . Data s e t s may o f t e n be encountered, which c o n s i s t o f
few p o i n t s or do not cover a s u f f i c i e n t range i n ~ [o be u s e f u l i n the above
mentioned methods off p a r a m e t e r i z i n g . In t h a t case i t i s w i s e r to r e s o r t to the
forms given i n eqs.6a and 6b. With these forms a graduated c h a r a c t e r i z a t i o n
o f h y d r a t i o n curves i n degree o f l i n e a r i t y has been a b o l i s h e d , and s u b s t i t u t e d
f o r an e s t i m a t i o n i n e i t h e r l i n e a r or p a r a b o l i c k i n e t i c s . A simple g r a p h i c a l
procedure i s to p l o t ] / P a g a i n s t 1 / ( t - t O) in the l i n e a r case, and a g a i n s t
1/7~-~n, in the p a r a b o l i c case. In both cases a s t r a i g h t l i n e i s produced g i -
v i n g l~Po~by e x t r a p o l a t i o n to i n f i n i t e t i m e . Since t O i s e n t e r i n g the v a r i a b -
l e s p l o t t e d , t h i s a d j u s t a b l e parameter must be determined by t r i a l and e r r o r
u n t i l the best s t r a i g h t l i n e i s being produced.
For data s e t s from h y d r a t i o n e x p e r i m e n t s which are not c a r r i e d f a r enough,
i t i s o f t e n found t h a t p l o t t i n g a g a i n s t 1 / ( t - t O) or l/tVT~t-t n produces e q u a l l y
good s t r a i g h t l i n e s . In t h i s case l i n e a r or p a r a b o l i c k i n e t i c s may be s e l e c -
ted by c o n s i d e r i n g the two d i f f e r e n t v a l u e s of Poe o b t a i n e d . One may be jr| c l o s e r
agreement w i t h e x p e c t a t i o n than the o t h e r . I f an e x p e c t a t i o n v a l u e o£ Poe i s not
known, i t must be concluded t h a t the data set at hand i s i n s u f f i c i e n t f o r pa-
rameterizJng.

Chemical shrinkage as a source of hydration data

Heasurements of chemical shrinkage have been found by us to be an extreme-
ly valuable source of information on cements hydration. Huch of our success
w i t h the use of eq.7 has depended on Lhe completeness of da~a sets o b t a i n e d
w i t h t h i s method. Measurements of chemical s h r i n k a g e can e a s i l y be o b t a i n e d as
c o n t i n o u s curves and can be c a r r i e d out f o r as ]ong as neccesary to p r o v i d e the
 Vol. 14, No. 5 629
DISPERSION MODEL, CEMENT HYDRATION, CONCEPTS, PARTICLE SIZE

completeness required for successful modelling. The measurement is simple, and

the number of samples studied is not limited by cost or labour. By adjusting
the sample-size depending on the water/cement-ratio we have found no trouble in
securing unhindered suction of water into the samples and thus securing accu-
rate measurements of hydration. A full documentation of the use of chemical
shrinkage in the study of cement hydration can be found in Mette Geiker~ Ph.D.
thesis (11).

Discussion on the model

It is the ultimate goal of any mathematical model of hydration to be able
to predict the out-come of future hydration experiments. Io achieve this goal
the relations between the factors decisive for the development of hydration
and the parameters entering the mode] should be searched for. In the case of
the dispersion model two important relations are built into the model. Prima-
rily the effect of fineness is explicitly given by the model, as seen from
eqs.5, and has been experimentally verified in (6), at least for the linear
case. The effect of temperature on the parameters t I and t 2 is given in so far
as its effect on the rate constants k I and k? can be assessed. The applicability
of the Arrhenius eqation for this purpose ha~ been demonstrated in ref.(12).
Concerning the effect of chemical composition of the cement and the effect
of chemical admixtures we have been greatly concerned with the question of
what factors determine the underlying kinetics, linear or parabolic. Our obser-
vations so far seem to indicate, that high degree of "available" Ca++-ions in
the beginning of the hydration favors parabolic kinetics, lhese Ca++-ions may
come from sources such as gypsum in the cement or from added CaCi 2 etc. Ca ++
-ions may become"unavailable"for the hydration of the clinker particles by
precipitation with C~A in the well-known double salts of the form C~A(CB)
(H20)x, (where B sta6ds for an anion). Thus small ratios of Ca++-io6s to [he
amount of free C3A (C3A not bound in the interior of the clinker particle) pro-
duces linear kinetics-and for large ratios we get parabolic kinetics. Hany
experiments (to be published later) performed in our laboratory seem to reveal
this pattern. In some intermediate range of ratios we get a joint kinetics as
already explained with the additions of CaCl 2 to Portland cement pastes. An-
other example is white clinker which hydrates almost linear whereas white ce-
ment is parabolic, and with additions of gypsum we experience the same gradual
change from linear to parabolic kinetics with decreasing values of DL. It
should be mentioned that the factors influencing the "order of kinetics" (whe-
ther l i n e a r or p a r a b o l i c ) i f l f l u e r l c e the lenqth of the dormant period as w e l l , as
modelled by t O in eq.7. Linear k i n e t i c s g e n e r a l l y comes together w i t h long dor-
mant periods and v i c e versa.
lhese o b s e r v a t i o n s of the i n f l u e n c e of a v a i l a b l e Ca++-ions during the e a r l y
period of h y d r a t i o n have lead us to a t e n t a t i v e mechanistic i n t e r p r e t a t i o n .
According to these s p e c u l a t i o n s l i n e a r k i n e t i c s i s the r e s u l t of a membrane
h i g h l y r e s i s t a n t to t r a n s p o r t of water, and which maintains i t s s t a b i l i t y and
r e s i s t a n c e throughout h y d r a t i o n . When the membrane formed i s weak (or absent!)
d i f f u s i o n through the inner (and probably o u t e r ) zone of h y d r a t i o n products be-
comes r a t e - c o n t r o l l i n g and p a r a b o l i c k i n e t i c s r e s u l t s . The s t a t e of membrane,
we b e l i e v e , i s h i g h l y dependent on the a v a i l a b l e Ca++-ions ( a l k a l i - i o n s seem to
have the same e f f e c t , although to a l e s s e r e x t e n t ) in the p o r e - s o l u t i o n , ]arge
amounts of Ca++-ions f a v o r i n g a weak membrane. Since the membrane c o n s i s t s of
c o l l o i d a l p a r t i c l e bound together w i t h p o l a r a t t r a c t i v e forces, i t i s tempting
to speculate t h a t Ca++-ions ( i n combination w i t h a v a i l a b l e anions) s u b s t i t u t e s
the h i g h l y p o l a r s i l a n o l groups, thus l e a d i n g to weaker a t t r a c t i v e forces be-
tween the c o l ] o i d a l p a r t i c l e s . This l a s t p o i n t i s pure s p e c u l a t i o n , but might
act as a contemporary model to aid in proposing f u r t h e r experiments to e l u c i -
date this point.
630 Vol. 14, No. 5
T. Knudsen

An interesLi,lg article cori(:urnii~g tlle ~furementior~ud has appeared feces,fly

( 1 3 ) , N,L,Thomas e l a l performed e x p e r i m e n t s w i t h EDTA i n h y d r a t i o n o f cement,
T h e i r f i n d i n g s and c o n c l u s i o n s are i n good agreement wiLh our l i n e s o f t h i n k i n g
c o n c e r n i n g the i n f l u e n c e o f Ca++-ions on the l e n g t h o f the dormant p e r i o d and
on the C / S - r a t i o o f the membrane.

References
i. r. Knudsen, 7th International Congress on the Chemistry of Cement,
Paris 1980, 1-170.
2. f. Knudsen, Engineering Foundation Conference, Henniker, New Hampshire 1982.
3. J.H. Taplin, 5th International Symposium on the Chemistry of Cement,
Tokio 1968, Vol If, p. 249, p. 337 and p. 421.
4. R. Kondo and S. Ueda, ibid Vol II, p. 203.
5. J.H. Taplin, a.C.S.Faraday I, 76,2229(1980).
6. S. Bang Larsen, H.Sc. report, Danish Technical University, The Institute of
Mineral Industry, 1983 (in danish).
7. C.J. Bernhardt, Rilem Symposium on Winter Concreting, 1956.
8. Chin Fung Kee, ACI Journal, flarch 1971.
9. N.J. Carino, International Conference on Concrete at Early Ages,
Paris 1982, Vol I, p. 123.
lO. T. Knudsen, Cement and Concrete Research, _9,799(1979).
iI. H. Geiker, Ph.D. thesis, Danish Technical University, The Institute of
Mineral Industry, 1983.
12. M. Geiker and T. Knudsen, Cement and Concrete Research, 12,603(1982).
13. N.L. ]homas and D.D. Double, Cement and Concrete Research ]_~3,391(1983).