Professional Documents
Culture Documents
1990 - Groves e Zhang - A Dilatation Model For The Expansion of
1990 - Groves e Zhang - A Dilatation Model For The Expansion of
ABSTRACT A model for the expansion of a silica glass/alkaline OPC mortar has
been developed which is based on the observation that the main
reaction product of the glass is a layer of CSH gel formed at its
surface. In the model, which takes into account the mechanical
- properties of the system, the increase in voltm~e of the glass plus
its reaction product layer above that of the original glass leads to
a predicted mortar expansion which is slightly less than that
observed, but of the correct order of magnitude.
i. Introduction
453
454 G.W. Groves and X.Z.,hang Vol. 20, No. 3
Creep of the matrix at this stress level would reduce the effective modulus
over a period of months, by roughly one-half (8). A value of ~ on the order
FIG. 1
T~M micrograph of 6 month old alkaline OPC/silica glass mortar showing
reaction layer at the surface of a glass particle.
V~,I. 20, N~. 3 Dilatation Model. $iO 2 Glass, Cement, Mortars 455
FIG. 2
Optical micrograph of 3 month old OPC/silica glass mortar showing angular
glass particles.
456 G.W. Groves and X. Zh~tg Vo]. 20, No. 3
the observed volume of CSH reaction product layers on the surfaces of the
glass particles represents an increase in volume of regions within the
mortar. Although the value of A is an estimate, it should be noted that a
variation of A in the range say, 3 to 4, would produce a variation of only
from 67% to 75% in the percentage of reaction product volume representing a
volume increase. The uncertainty in the calculation of the volune increase
of the mortar introduced by the uncertainty in the exact composition and
porosity of the gel is probably smaller than the uncertainty of other
factors in the calculation, as discussed below.
2.2. The volume of CSH gel reaction product at the qlass particle surfaces
V = St (2)
where S is the total surface area of the particles. In our case, the glass
particles were prepared in a certain size range by grinding and sieving and S
was not measured directly. The estimation of S gives the greatest source of
uncertainty in our calculation. If the particles were spherical, tb~ value
of S could readily be calculated from the average particle size. However the
glass particles in our case are extremely angular as shown in fig.2. This
can be expected to introduce a much larger value of S. Thus for exanple
Shacklock and Walker, as quoted by Neville (ii) found that crushed granite
particles sieved to the same size range as rounded gravel particles had a
surface area between 1.6 and 1.9 times as great, over a wide size range from
300-600~m to 17.7 - 19.05ram. (Our particles are somewhat smaller than this
range, but there is no indication in Shacklock and Walkers' results of a
systematic change in the factor with particle size).
6~tv a
V = d (3)
where d is the average paricle size and V a is the total volLm~e of glass
aggregate. The factor 6Va/d arises from the surface : voltm~ ratio for a
Vol. 20, No. 3 DilatationModel, $iO2Glass, Cement,Mortars 457
AV = 0.71V (4)
4.26 ~ t V
or AV = d a (5)
3K + 4~ (8)
Y= 3K
Using a value of 0.25 for the Poisson's ratio V of the c~ment paste
matrix (12) we calculate y to be 1.8. The bulk modulus of the glass
particle, modelled as a mis-fitting sphere should be corrected for the
reduced modulus of the reacted surface layer, although for the layer
thicknesses -i~ with which we are concerned the effective modulus is reduced
by only 3%, which is a relatively small factor. An effective bulk moch/lus
for the glass particle of 36(~a and a shear modulus for the matrix of 8GPa
gives a value of ~' of 1.3 and
Thus in the elastic misfitting sphere model the expansion of the mortar is
40% greater than the effective increase in volume of a glass aggregate
458 G.W. Groves and X. Zhang \%1.20, No. 3
6~tV a
AV = (Ii)
d
tV a
AVm -- ii.4 d (12)
4
P =]~v (13)
s
i Arm V
-3 ( - T ) : (14)
a
m m
Vol. 20, No. 3 D~a~ationModel, SiO 2 Glass,Cement, Mortars 459
in favour of the widely accepted explanation in other cases, the authors are
not aware that a detailed quantitative prediction of an expansion has ever
been compared with a measured value, using this explanation as a basis. In
view of the success of our alternative model in this respect, albeit, in a
particular case, it may be worth keeping an open mind about the mechanism of
expansion in some other instances.
Acknowledqement
References