400
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
the k factor would be sufficient, but even that is beyond the
scope of laboratory tests."
The handbook of groundwork prefers pumping tests
performed on site, but fails to mention what to do in layers
above the ground water tables. Kovács (1972) takes a different
view and recommends using the grain distribution curve.
"Based on a brief description and a critical analysis of
laboratory and on site tests, it is underlined once more that
formula based calculation should normally be recommended as
the method for determining the coefficient of permeability, not
only because this is the simplest technique but also because its
reliability reaches and in most cases even surpasses that of other
methods. In-laboratory and on-site measurements are justified
only in case it is our intention to describe a unique stratification
property of a layer. That would require laboratory analysis of
undisturbed drill cores, advanced percolation tests or pumping
tests using several observation wells."
The coefficient of permeability of a rather large basket of
soil types, i.e. ones not characterised by grain size distribution,
is left undetermined this way. The following remarks allow us
to conclude that there the industry lacks consensus on how to
determine the coefficient of permeability value.
Section S3 “Evaluation of test results” of the annex to the
EUROCODE 7 standard specifies the following requirements
for evaluating test findings:
„There are four widely used methods to determine the
coefficient of permeability (hydraulic conductivity):
• field tests, such as pumping and borehole permeability tests;
• empirical correlations with grain size distribution;
• evaluation from an oedometer test;
• permeability tests on soil specimens in the laboratory.”
So it can be concluded that there are many laboratory and in
situ methods to obtain the coefficient of permeability. However
there is no universally applicable method; each method is valid
within certain limits, which we need to identify to render
investigations easy to target and plan.
Measurements have tended to take the form of site
investigation as soil mechanics have developed in the past 15-
20 years. That way multiple damage to samples can be avoided
and results will better reflect local conditions. It is
internationally accepted that local investigation provides more
accurate site specific values. The question arises whether or not
this statement also holds for the determination of coefficient of
permeability values.
1.2
Aims of study
We set out to determine the coefficient of permeability of
transitional and fine grained soils (ranging from fine sand
through sand meal and miry sand to silt). The following
boundary parameters were assumed for the purposes of our test
series:
• We selected methods whose range of validity matched in
principle the type soil selected for the tests.
• Homogeneous isotropic strata were assumed for the
purposes of the test despite the likelihood of periodic
sedimentation of coarser and finer grains during layer
formation, and an apparently homogeneous layer may
be composed of a network of more conductive and more
watertight lenses seams.
• Potential filtration anomalies at layer boundaries are
ignored.
• The increased conductivity due to atmospheric effects and
human intervention of a layer of top soil, which can be
up to 0.6-0.8 m thick, is also neglected.
• Most tests determine the coefficient of permeability on a
relatively small sample of soil. It would, however, be a
mistake to generalise the value achieved that way for
the whole layer represented by the sample.
2 TESTING METHODS
The following methods were used to measure the coefficient
of permeability of fine grained and transitional soils on site:
• Horizontal permeability can be measured with a Menard
probe inserted into a vertical bore hole. The radial
infiltration of water into the soil is facilitated by packers
and by the injection of water below and above the
measurement section.
• Water absorption test across a trickling head lowered
through a Khafagi probe to determine the coefficient of
permeability. Soil conditions are taken into account for
the purposes of dimensioning the trickling head to be
used and the calibrated container.
• Depending on ground water level, one or more boreholes
may be lowered for pumping or water absorption. Soil
conditions must be taken into account for determining
the layout and dimension of the boreholes. Serious
errors may occur if the liner fails to connect properly to
the hole bottom, as water will not only trickle into the
soil across the bottom but along a sleeve of unknown
length.
Equipment of constant or falling water head may be used in
laboratory measurements depending on the coefficient of
permeability.
We have also determined the value of coefficient of
permeability indirectly (by empirical correlation based on grain
size distribution) to compare and verify local measurements.
Different authors have identified different relationships to be
used in the indirect method of calculation and have partially
combined these methods with a variety of status descriptors. A
shared feature of these methods involves plotting a grain size
distribution curve typically identifying the grain diameter (d
10
)
associated with ten mass percentages passing and this value is
normally on the power of two. This paper presents the results
received from calculations using formula (see Figure 2).
3 TEST LOCATION, SOIL TYPES
Tests were performed at five locations, but this paper only
covers the findings testing section 54+260 of the left bank of the
Danube near Ráckeve. The tests were performed on the
protected side 10 meters from the toe of the flood control dyke.
Exploratory drilling identified the following order of layers:
• the upper layer from 0.0 to 3.4 m contains yellow and
yellowish grey silt with silty sand of low water content
(7% < w < 14%) and with moist density at around
=
1.76 g/cm
3
. The grain size distribution curve shows that
the fine content makes up 80-90% of soil particles. The
coefficient of uniformity vary between C
u
= 8.6-12.3
(see figure 1).
• the layer from 3.4 to 5.0 m contains sand with grey silt. The
water content of this well graded layer is 20% on
average. Wet bulk density is around
= 1.86 g/cm
3
. The
layer is understood to be much looser than the one
above. The examination of grain size distribution
suggests that the sand fraction makes up 70-75% with
silt at 25-30% (see figure 1). The coefficient of
uniformity is at C
u
= 30-33.
