).
E
st
was calculated as the slope of the straight line
which links the origin of the stressstrain curve with the
corresponding point at 70% of ultimate strength (secant
modulus) following ASTM D 314869 (1996).
Weathering tests
A sample of each of the ten rock types was subjected to a
salt crystallization test and a further sample of each to the
thermal shock test.
Salt crystallization test: Dry samples of each rock were
introduced vertically into a vacuum vessel and covered
with a 31.7% w/w Na
2
SO
4
solution. The immersion was
carried out at a temperature of 20C and a pressure of
30 mbar and maintained for 24 h. The samples were then
put into a climatic chamber at 5C and 90% RH for 10 h.
Finally, the samples were dried in a heating cabinet at
105C for 14 h. They were then cleaned with pure water
until the salt was eliminated and dried until they reached a
constant weight.
Thermal shock test: the conditions for this test were
identical to those used for the salt crystallization test except
that the samples were saturated in the vacuum vessel with
pure water instead of the Na
2
SO
4
solution.
Results and discussion
Both the unweathered and weathered samples were tested
using ultrasound and uniaxial compression tests and the
dynamic (E
d
) and static (E
st
) elastic moduli calculated for
each sample. The ten selected varieties of rocks show
different degrees of heterogeneity: from homogeneous
limestones to brecciated dolostones. After the weathering
tests, new ssures began to appear whilst pre-existing
discontinuities became larger. Thus, the similarities
between E
d
and E
st
in rocks with ssures and a high degree
of deterioration could be studied.
Figure 2 shows the relationship between E
d
and E
st
in
the studied samples. The straight red line (line with slope
equal to 1) represents the ideal correlation E
d
= E
st
, but as
seen in the gure, there was a considerable spread in the
results.
From these results, it is obvious that a correction factor
must be applied to Eq. 1 in order to render it suitable for
highly heterogeneous rocks. The correction of this equation
is carried out by a dimensionless coefcient, K:
K
E
d
E
st
; 3
Equation 1, corrected after K, will be
E
st
1
K
q
bulk
V
2
P
1 2m 1 m
1 m
4
The ideal K value is 1. In this case (K = 1), E
d
is perfectly
calculated by means of Eq. 1 and no correction is neces-
sary. When K[1, E
d
(obtained by means of Eq. 1) is
overestimated. When K\1, the calculus of E
d
underesti-
mates the real value.
According to Ciccotti and Mulargia (2004), K acquires
values between 1 and 1.3 although Al-Shayea (2004) points
out that the proportion between E
d
and E
st
(K value) varies
from 0.85 to 1.86. Table 1 shows the K values obtained in
the samples examined in this study (weathered and
unweathered samples). The dynamic to static modulus ratio
varies between 0.5 (the calculated E
d
is just half of the
measured E
st
) and 2.1 (E
d
is practically twice E
st
).
Al-Shayea (2004) observed that the variation of K is
directly proportional to the obtained value of E
d
: the higher
the E
d
, the higher the K value. In order to check this
assertion, the results obtained in the present study were
plotted in Fig. 3.
It can be seen that a relatively good relationship can be
established between E
d
and K when E
d
values are higher
Fig. 2 Relationship between E
d
and E
st
in the studied samples. The
straight red line corresponds to the ideal ratio E
d
/E
st
= 1
266 J. Mart nez-Mart nez et al.
1 3
than 50 GPa. This relationship is dened by the following
equation:
K 0:03 E
d
0:5 5
When the calculated value of E
d
is lower than 50 GPa,
however, the values show an extremely wide scatter such
that it is not possible to dene a reliable relationship
between these parameters.
Samples with an E
d
value lower than 50 GPa correspond
to highly ssured or highly decayed samples. The scatter-
ing observed in these samples is due to the effect of the
fractures on the ultrasonic wave propagation or on the
spreading of mechanical stress. Previous studies have
demonstrated that wave velocity (V
p
) has a low sensitivity
in terms of detecting differences in fracture orientation,
fracture apertures or fractures with a different relative
position in the sample (Mart nez-Mart nez et al. 2011).
However, rock elasticity (quantied by E
st
) is highly
dependent on these structural aspects (orientation, aperture
and position of fractures).
For example, take two theoretical rock samples A and B
in which there is a fracture in the central section of each
one. In sample A, the fracture is perpendicular to the wave
propagation direction and has a large aperture while the
fracture in sample B has a moderate aperture and forms an
angle of 20