Professional Documents
Culture Documents
Estimation of Hydraulic Conductivity From Grain Size Analyses
Estimation of Hydraulic Conductivity From Grain Size Analyses
ANDREA SVENSSON
Department of Civil and Environmental Engineering
Division of GeoEngineering
Engineering Geology Research Group
CHALMERS UNIVERSITY OF TECHNOLOGY
Göteborg, Sweden 2014
Master’s Thesis 2014:1
MASTER’S THESIS 2014:1
Cover:
Grain size curve from borehole OC4008-1 at Skansen Lejonet. The results are
interpreted in chapter 8.
ABSTRACT
The purpose of this thesis is to evaluate different methods to calculate hydraulic
conductivity, as well as to compare the methods to obtain soil samples used in the
project Västlänken, with sampling methods used in this thesis. Soil samples from two
locations, Skansen Lejonet and Korsvägen, were taken. Laboratory work such as grain
size analyses and different porosity measurements were used on these samples. By
using a version of the Kozeny-Carman method to calculate hydraulic conductivity, the
results were compared to the most common methods, Hazen and Gustafson. The
values from the previous sampling in the project were also used to calculate hydraulic
conductivity with Kozeny-Carman. The three methods were compared to each other
and the results from the previous soil sampling were compared with the results from
the current sampling performed in this thesis. The results from grain size analyses
were also compared with slug tests and pumping tests performed in the project
Västlänken. The results showed that the Kozeny-Carman equation generally gives a
lower conductivity than the Hazen and Gustafson equation, but may be more in line
with the results from the hydraulic tests. The results also showed that there were very
small differences between the more limited sampling method used previously and the
sampling methods used in this thesis. The conclusion drawn from these results was
that for test sites with fairly homogenous soil like Korsvägen and especially Skansen
Lejonet the limited sampling method is accurate enough. The more elaborate
laboratory work needed to use the Kozeny-Carman method may discourage the use of
this method. However, if some work were performed on classifying the degree of
compaction of a soil sample easily, the Kozeny-Carman method would be easier to
use. The conclusion is that the Kozeny-Carman method could be useful to evaluate
the hydraulic conductivity from grain size analyses with more accuracy.
KEY WORDS: hydraulic conductivity, grain size analysis, Kozeny-Carman equation,
soil sampling, porosity, Västlänken.
I
II
Contents
ABSTRACT ................................................................................................................... I
CONTENTS ................................................................................................................. III
HANDLEDARENS FÖRORD .................................................................................. VII
PREFACE ................................................................................................................. VIII
NOTATIONS .............................................................................................................. IX
1 INTRODUCTION ................................................................................................. 1
1.1 Background .............................................................................................................. 1
1.2 Aim and objections .................................................................................................. 2
1.3 Delimitations ............................................................................................................ 2
2 HYDROGEOLOGICAL CONCEPTS .................................................................. 3
2.1 Darcy’s law .............................................................................................................. 3
2.2 Soil properties .......................................................................................................... 3
2.2.1 Grain size ............................................................................................................. 3
2.2.2 Porosity ................................................................................................................ 4
2.2.3 Degree of compaction .......................................................................................... 4
2.2.4 Grain shape .......................................................................................................... 4
2.3 Hydraulic conductivity............................................................................................. 4
3 LITERATURE REVIEW ...................................................................................... 5
3.1 Comparison between field methods and grain size analyses ................................... 5
3.2 Evaluation of the Kozeny-Carman equation ............................................................ 6
3.3 Comparison between laboratory methods and grain size analyses .......................... 7
3.4 Porosity .................................................................................................................... 8
4 HYDROGEOLOGICAL ENVIRONMENT ....................................................... 10
4.1 Skansen Lejonet ..................................................................................................... 11
4.2 Korsvägen .............................................................................................................. 11
5 METHODS FOR MEASURING HYDRAULIC CONDUCTIVITY ................. 15
5.1 Indirect methods..................................................................................................... 15
5.1.1 Hazen ................................................................................................................. 15
5.1.2 Gustafson ........................................................................................................... 15
5.1.3 Kozeny-Carman ................................................................................................. 16
5.2 Hydraulic tests........................................................................................................ 19
5.2.1 Pumping test ...................................................................................................... 19
5.2.2 Slug test ............................................................................................................. 19
5.3 Laboratory methods ............................................................................................... 19
Extern handledare
Bergab – Berggeologiska Undersökningar AB
Annika Nilsson
Greek letters
γ =Specific weight [N/m3]
ε = Void ratio [-]
μ = Viscosity [(Ns)/m2]
μG =Geometric mean [mm]
ρ = Bulk density of soil sample [g/cm3]
ρs = Compact density of soil sample [g/cm3]
ρ1 = Density of mineral turpentine [g/cm3]
σ =Geometric standard deviation [mm]
τ = Ratio of kinematic viscosity at 10°C and ground water temperature [-]
Figure 1 The stretch of the railway tunnel through central Gothenburg (Trafikverket, 2013)
𝑑ℎ
𝑄 = −𝐾𝐴 2.1.
𝑑𝑙
Q = Flow [m3/s]
K = Hydraulic conductivity [m/s]
A = Area [m2]
dh/dl = Hydraulic gradient [-]
The difference in height and length is represented by the hydraulic gradient, and is the
reason for the negative sign as the water flows from higher to lower level. When
describing ground water flow, Darcy’s law is altered to describe all three dimensions.
Often, this is simplified to describe 2-dimensional flow. Depending on type of
aquifer, different partial differential equations are used to describe the flow, like the
Laplace equation. To solve these equations analytically, idealized cases are used and
the boundary conditions of the aquifer must be known. By using a conceptual model
of a ground water system, reality can be translated and simplified into a manageable
model.
2.2 Soil properties
The hydraulic conductivity varies in different soils. A sorted soil like an esker deposit
has a high conductivity, and is therefore often used for ground water extraction. In
Sweden, the most common soil type is glacial till, which is an unsorted soil (Knutsson
& Morfeldt, 2002). It is often difficult to determine hydraulic conductivity in an
unsorted soil due to the great variation in the composition of the soil. In a glacial till,
the grain size varies a lot as well as the degree of compaction. Unsorted soils often
show a high degree of anisotropy; the hydraulic conductivity depends on the direction
of measurement. This means that the hydraulic conductivity is different in a
horizontal orientation than in a vertical direction.
2.2.1 Grain size
The grain size distribution of a soil is one of the soil mechanic properties that affect
the hydrogeological conductivity. A sorted soil with larger grains has a high hydraulic
conductivity. If a sediment contains a mixture of grain sizes, a more multi-graded soil,
the porosity will be lowered, and thus the hydraulic conductivity (Fetter, 2001). This
is because the void between the larger grains is filled up with smaller grains.
Figure 2 Different grain shape makes the water travel in different paths (Fagerström & Wiesel,
1972)(modified)
𝜏 𝜀3
𝐾 = ∙ 405 ∙ ∙ 𝑑𝑒2 3.1.
𝑅 (1 + 𝜀)
3.4 Porosity
Frings et al (2011) studied the accuracy of porosity predictors for fluvial sand-gravel
deposits. They looked at several ways to measure porosity based on different
parameters such as effective grain size, standard deviation, deviation of the grain size
distribution from a type curve and theoretical predictors that take mixing processes
into account and calculate porosity computationally. They used laboratory methods
and in-situ experiments to calculate the porosity and compared this to different
methods based on the parameters mentioned above. The uncertainty of the laboratory
methods could lead to an overestimation of the porosity, mainly due to disturbance of
the packing near the container walls, but the authors believed that this overestimation
was probably small. In addition to the experimentally obtained results, the authors
also used two porosity data sets from literature to compare to.
The results showed that empirical predictors based on the relation between median
grain size and the porosity did not correspond very well to the laboratory results. In
part, this is because the finer sediments of a grain size distribution often represents a
small percentage of the entire distribution, and does not affect the median grain size a
lot even though this has a great impact on the porosity. The conclusion was that there
is no unique relation between grain size and porosity. The empirical methods could be
useful in cases when the geological conditions mirrored the original conditions in
which the methods were developed, but were not generally applicable. When looking
at methods based on the other parameters the correlation was better, especially when
using theoretical predictions, but the methods based on these parameters were still not
able to produce highly accurate porosity predictions. The authors developed a tailor-
made equation for the studied area with multivariate regression analysis, which used
two independent parameters: the sediment standard deviation and the number of
grains smaller than 0.5 mm. They were still not able to produce accurate predictors
but could see trends such as downstream decreasing porosities, and concluded that
porosity predictors are useful to provide insights in the spatial variation in porosity.
Sakata and Ikeda (2013) studied how hydraulic conductivity varied by depth in
alluvial gravel deposits. The dependence of hydraulic conductivity on depth in
𝑑20 1.9
𝐾 = 6.89 ( ) ∙ (𝐿1 + 𝐿2 ) + 0.0167 ∙ 𝐿3 3 3.2.
1000
The L-values gives a form of visual way of measuring the porosity. The d 20-value was
chosen because it produced the highest correlation to the slug tests. The authors
concluded that there was a clear depth dependency, where an increase in depth of 1
meters corresponds to an approximately 10% decrease in hydraulic conductivity.
However, the relations between the slug tests and the core properties were not
sufficiently verified, and were only valid for this particular site. The slug tests showed
a slightly lower hydraulic conductivity than the grain size analysis.
FILL
CLAY/
5 SILT
CLAY
SAND/
10 GRAV.
BED-
ROCK
15
Figure 3 Example of typical layer sequence. The numbers marks depth in meters. Modified from
(SWECO, 2013c).
Figure 4 View over Skansen Lejonet. Possible tunnel stretches is marked in black (SWECO, 2013c).
4.2 Korsvägen
Korsvägen is situated in central Gothenburg, and the tunnel will be constructed from
east to west through the area, see Figure 5.
JOHANNEBERGS
LANDERI
SÖDRA
VÄGEN
CARLANDERS-
PLATSEN
Figure 5 Map over the area around Gothenburg, in the south central part of the town. The lines show
the approximate corridor where the tunnel will be constructed (Google Maps, 2014)(Modified).
The area around Korsvägen is characterised by valleys and dips in the north-north-
west direction. South of Korsvägen, along Södra vägen, there is a valley with a flat
ground in this direction. The soil layers in the valley consist of glacial clay, which in
the north is overlayed by postglacial clay. In the south, the thickness of this layer is
less than ten meters, but increases to the north to over twenty meters. Below the clay
there is a layer of friction soil, on top of bedrock. The bedrock in the area consists of
schisted gneiss with a north-south strike.
To the west, parallel to this valley, there is another dip. The layers are similar to the
valley. The layer of friction soil in this valley is approximately two meters thick in the
south and approximately six meters thick in the north, around Carlandersplatsen.
Many construction projects have been executed around the area. Several of the urban
areas are sensitive to settlements, for example the gardens in Johannebergs Landeri.
Previous investigations have shown that the area south of Carlandersplatsen and
around Södra vägen can tolerate a lowering of the ground water table up to one meter,
but the area north of Carlandersplatsen cannot tolerate any lowering of the ground
water Table.
The soil depths at Korsvägen can be seen in Figure 6, and means that the foundation
of the tunnel will be constructed in part bedrock and part soil.
There is in some places an upper groundwater zone in the fill on top of the clay.
Where the thickness of this layer is less than 2 meters there is probably no water.
Some friction material has also been found inside the clay, which may constitute a
middle zone, but the thickness and range of this is quite uncertain and probably
limited in its extension.
The lower ground water zone in the friction material is characterized as fine sand to
sand on top of glacial till. The thickness of this zone varies between approximately 1-
10 meters, see Figure 7.
This formula was developed for designing sand filters for water purification but is
very commonly used to estimate the permeability of soil.
5.1.2 Gustafson
Gustafson introduced a way to calculate hydraulic conductivity from grain size
analyses that is often used in Sweden today. The formula was calculated by using a
large number of samples where results from grain size analyses were compared with
results from pumping tests (Andersson, et al., 1984). The hydraulic conductivity is
calculated as follows:
𝑑10 2
𝐾 = 𝐸(𝐶𝑈 ) ∙ ( ) 5.2.
1000
𝐷60
𝐶𝑈 = 5.3.
𝐷10
𝐸3 1
𝐸(𝐶𝑈 ) = 10,2 ∙ 106 ∙ ∙ 5.4.
1 + 𝐸 𝑔(𝐶𝑈 )2
1 1
𝐸 = 0,8 ∙ − 2 5.5.
2 ∙ 𝑙𝑛(𝐶𝑈 ) 𝐶𝑈 − 1
1,3 𝐶𝑈 2 − 1
𝑔(𝐶𝑈 ) = ∙ 5.6.
log10 𝐶𝑈 𝐶𝑈 1,8
5.1.3 Kozeny-Carman
The Kozeny-Carman equation was proposed by Kozeny in 1927 and modified by
Carman in 1937 and 1956. It is a semi-empirical, semi-theoretic formula, and will be
explained a bit more in detail below (Carrier, 2003).
𝐴 5.8.
𝑚=
𝑃
m = hydraulic radius [-]
AC = cross-sectional area of channel [m2]
P = wetted perimeter [m2]
𝑛 5.9.
𝑚=
𝑆
m = hydraulic radius [-]
n =Porosity [-]
S = Particle surface for unit volume of the porous media [-]
S is the particle surface for unit volume of the porous media. The specific surface S0,
relates to the particle surface as
𝑆 = 𝑆0 (1 − 𝑛) 5.10.
𝑑50 2 𝜀 3 2 𝛾 5.11.
𝐾= ∙ ∙ e−0,48∙𝜎 −0,9∙𝜎 ∙ ∙ 1000
180 1 + 𝜀 𝜇
5.4 Porosity
Åberg (1992a & 1992b) studied the porosity function in the Kozeny-Carman
equation. By studying the solid volume of grains of the granular material and the
fraction of the solid volume that passes through a specific grain size, he set up
integrals to describe grain size distribution. The void ratio and thus the porosity can
then be calculated on the basis of the grain size distribution. These integrals will not
be explained in this thesis, but is based on the effective grain size and the geometric
standard deviation. For a thorough explanation, see the paper by Åberg (1992a &
1992b). This gives the parameters A0 and B0 by the following equations:
2𝑑50 5.13.
𝐴0 = ∙ (3,0523 − 1,1549𝜎 + 0,6497𝜎 2 − 0,1521𝜎 3 + 0,0281𝜎 4 )
2∙𝜋
2 ln 2⁄2
𝐵0 = 2𝑑50 ∙ 2𝜎 5.14.
The geometric is mean used to calculate the effective grain size d50:
ln (1⁄𝑑 ) 5.15.
60
𝜇𝐺 = + 0,25 ∙ 𝜎
ln 2
μG = Geometric mean [mm]
d60 = The particle size for which 60% of the material is finer [mm]
σ = Geometric standard deviation [mm]
1 5.16.
𝑑50 = ∙ 0,001
2𝜇
d50 = The particle size for which 60% of the material is finer [mm]
μ = Geometric mean [mm]
𝐴0 5.17.
𝜀 = 2𝑐 + 2𝑑
𝐵0
ε = Void ratio
c = Constant, in this thesis set to 0.73
d = degree of compaction
The constant d depends on the degree of compaction, and is determined by Proctor
compaction. d varies between 0.18 and 0, for very loose to very compact packing. The
different values can be seen in Table 1.
Table 1 Classification of degree of compaction
0.045 3 Loose
0.03 6 Medium
0.015 12 Compact
0 25 Very compact
Figure 8 Nomogram for the classification of mineral soil after grain size distribution (Larsson 2008)
𝑉∙𝑙
𝐾= ∙ 10−2 6.1.
𝐴𝑠 ∙ 𝑡 ∙ ℎ
𝑚3 − 𝑚1
𝜌= ∙𝜌 6.3.
𝑚3 − 𝑚4 − 𝑚1 + 𝑚2 1
The mineral turpentine is used since it does not penetrate the saturated soil sample
and will evaporate when the soil sample is taken out of the liquid.
6.5 Porosity
The measurements to calculate the porosity has been done in two different ways,
depending on the methods used for the soil sampling. The porosity of the samples
from Korsvägen has been calculated by using specific gravity and water content and
the porosity of the samples from Skansen Lejonet has been calculated by colon tests.
𝑚8 − 𝑚7
𝑛= ∙ 100 6.4.
𝑚6 − 𝑚5
n = Porosity [-]
m5 = Mass of cylinder [g]
m6 = Mass of water up to mark and cylinder [g]
m7 = Mass of soil up to mark and cylinder [g]
m8 = Mass of water and soil up to mark and cylinder [g]
𝜌
𝑛 = (1 − ) 6.5.
𝜌𝑠 (𝑤 + 1)
n = Porosity [-]
ρ = Bulk density of soil sample [g/cm3]
ρs = Compact density of soil sample [g/cm3]
w = Water content [%]
𝑚𝑤
𝑤= ⁄𝑚𝑠 ∙ 100 6.6.
Figure 9 Map over Skansen Lejonet with boreholes marked in red (Google Maps, 2013b). Modified.
At Skansen Lejonet, the soil material was flushed up with compressed air and water,
see Figure 11. The tip of the drill were perforated with oval holes approximately
20x40 millimetres, see Figure 12. This sample method provides more disturbed
samples than when using a moraine sampler. As the material is flushed with water,
the original water content of the soil cannot be calculated.
KK4016
KK4003
KK5040 KK4001H
KK4008 KK4009
Figure 13 Map showing the boreholes at Korsvägen. Current sampling is marked in red and previous
sampling in green. The pumping well is marked in blue (Google Maps, 2013a) (Modified).
Air or
water
After pipe is
driven into the
ground
Samples
Figure 14 Principal sketch of the sampling method used in previous Figure 15 Drawing of a perforated
sampling (Gustavsson, 2005) pipe (Andersson, et al., 1984)
Figure 16 From left to right, permeameter test, pycnometer with soil and water, pycnometer
with soil and mineral turpentine
All samples were sieved to perform a grain size analysis, see description in chapter
6.1. Many of the samples from Korsvägen had a large amount of fine material, but
these samples were prematurely destroyed before a sedimentation analysis could be
done.
The porosity measurements were done in two different ways, described in chapter 6.5.
The methods used for samples were the water content cannot be determined requires
the samples to be of a certain volume, which is why experimental porosity values are
missing for some of the samples from Skansen Lejonet. This method is also more
likely to have uncertainties since it is less precise. The porosity values from
Korsvägen are therefore probably more reliable than the values from Skansen
Lejonet. For the Korsvägen samples, pycnometers were used to determine compact
and bulk density, see Figure 16.
7.4 Calculations of hydraulic conductivity
The three different empirical methods that are described in chapter 3 have all been
used to calculate the hydraulic conductivity. Gustafson and Hazen both depend on
grain size, more particularly the parameters d60 and d10 and the relationship between
these. In the cases where d10 were smaller than 0,063 millimetres the values have been
extrapolated with a log-normal distribution using the Excel add-on @risk.
The Kozeny-Carman method also uses d60 and d10, but in addition to this, porosity is
also used.
Skansen V. Com- V. Cu
Lejonet loose Loose Medium pact compact Exp Soil type
Gravelly 23.8
OC4008-1 0.31 0.16 0.14 0.11 0.09 0.28 sand
Gravelly 9.8
OC4008-2 0.36 0.23 0.21 0.19 0.17 sand
Gravelly 15.0
OC4008-3 0.34 0.19 0.17 0.15 0.13 sand
For the porosities from Skansen Lejonet the theoretical porosities closest to the
experimentally varies from Very Loose to Medium.
When comparing the experimental values from Korsvägen with the closest
corresponding theoretical value, the degree of compaction varies a lot. The chosen
degree of compaction for the porosities is Compact, however the porosity at KK5040-
1 is so different from the experimentally obtained one that Loose is chosen instead.
This borehole level is in this case seen as an anomaly.
Korsvägen
V. V.
Soil type Cu
Borehole loose Loose Medium Compact compact Exp.
Gravelly
KK5040-1 0.30 0.14 0.12 0.10 0.07 0.20 sand 31.1
Gravelly
KK5040-2 0.41 0.30 0.29 0.27 0.26 0.27 silty sand 4.5
Sandy
KK5040-3 0.40 0.29 0.27 0.26 0.24 0.19 silt/clay 5.2
Clayey silty
KK5038-1 0.42 0.31 0.29 0.28 0.26 0.27 sand 4.2
Clayey silty
KK5038-2 0.33 0.18 0.16 0.14 0.11 0.18 sand 17.5
Two factors can be seen that influences the porosity, soil type and uniformity
coefficient. When calculating the theoretical porosity, the uniformity coefficient
influences the value quite a lot. A multi-graded soil gives a low porosity as smaller
grains fits into the voids left by larger grains. For soil with a high content of fine
material, the material is probably more compacted whereas for coarser material the
packing is looser. This influences what grade in the compaction index the porosity
will lie in. A graph of how the different porosities vary can be seen in appendix 4. The
impact of the degree of compaction for the hydraulic conductivity can be seen in
appendix 3.
8.2 Hydraulic conductivity
The hydraulic conductivity has been calculated with four different methods, a short
description of these can be seen in Table 4. There are two different calculations based
on the Kozeny-Carman method, as the porosity is calculated in two different ways,
see chapter 6.5.
The hydraulic conductivity results are shown in Table 5. The permeameter test has
only been performed on four samples, and only three of these gave a result. The
Kozeny-Carman hydraulic conductivity calculated from experimental porosity, KKC2,
have not been calculated in three cases due to the lack of porosity calculations in these
cases. To show the difference between the results based on Gustafson and the results
based on Kozeny-Carman, the ratio, KG/KKC1, between these two methods are also
shown.
Gravelly
OC4008-1 4.6E-05 2.2E-05 2.7E-06 1.6E-06 2.4E-07 8.20 sand
Gravelly
OC4008-2 8.8E-05 7.3E-05 3.4E-05 - 12.65 sand
Gravelly
OC4008-3 1.2E-04 7.5E-05 1.9E-05 - 31.09 sand
Korsvägen
Gravelly
KK5040-1 2.9E-05 1.3E-05 8.30E-08 2.3E-07 - 203.00 sand
Gravelly
KK5040-2 1.9E-05 2.3E-05 4.4E-06 4.3E-06 - 5.24 silty sand
Sandy
KK5040-3 2.4E-06 1.9E-06 6.0E-08 8.0E-08 - 8.73 silt/clay
Clayey
KK5038-1 1.6E-05 1.9E-05 3.0E-06 3.7E-06 - 5.98 silty sand
Clayey
KK5038-2 4.6E-06 2.7E-06 2.4E-08 2.7E-07 - 45.17 silty sand
It is clear that Gustafson and Hazen gives the highest values, sometimes with a
difference of several orders of magnitude between the highest and the lowest values.
The lowest values are the ones experimentally calculated from permeameter tests,
Kexp, and the Kozeny-Carman equation with porosities calculated from grain size,
KKC1.
Skansen Lejonet shows higher hydraulic conductivity than Korsvägen. The values at
Skansen Lejonet spans between 10-4 and 10-7 m/s and at Korsvägen between 10-5 and
Just as in previous results, Hazen and Gustafson show fairly similar results where
Kozeny-Carman gives lower hydraulic conductivity than the other two. The hydraulic
conductivity at this location is rather high, with maximum values in orders of 10-2 and
10-3 m/s and ranging down to around 10-5 or 10-6 m/s in the case of KKC1.
In two of the four boreholes, the filters were clogged and no results were achieved.
For the two boreholes where results were achieved, Kslug in OC4002 were slightly
lower than both KG and KKC1. In borehole OC4003, Kslug were significantly lower
than both KG and KKC1.
At Korsvägen, slug tests were performed in 14 of the boreholes and measurements of
the recovery of the ground water Table was made in the three percussion boreholes
(SWECO, 2013a). The results are presented in Table 9, similar to Table 8. In some of
the boreholes, no grain size analysis has been done and KG and KKC1 are left blank.
1,E-01
1,E+00
Figure 17 Comparison of the results from Skansen Lejonet of the hydraulic conductivity with different
methods. The highest bars show the lowest conductivities.
The soil at Skansen Lejonet is generally more one-graded, which can be seen in these
results. The exceptions for this are partly in borehole OC4008 and partly in the
permeameter results. The Kozeny-Carman equation gives up to one and a half order
of magnitude difference in this borehole, which differ from the other boreholes, were
the results are almost similar.
1,E-07 K-Kozeny-Carman 2
K-Hazen
1,E-06 K-Gustafson
1,E-05
1,E-04
1,E-03
1,E-02
1,E-01
1,E+00
KK5040-1 KK5040-2 KK5040-3 KK5038-1 KK5038-2
Figure 18 Comparison of the results from Korsvägen of the hydraulic conductivity with different
methods. The highest bars show the lowest conductivities
The results at Korsvägen show some differences between the different methods.
Hazen and Gustafson are consistently fairly similar, which also corresponds to the
statistical analysis that showed no significant differences between the methods. They
are also consistently higher than Kozeny-Carman. The more multi-graded soil at
Korsvägen is evident as the results from Korsvägen show the largest differences
between calculation methods, despite using the same sampling method. The
differences vary between one up to almost three orders of magnitude. A discussion of
the permeameter result will follow.
9.1 The effect of porosity on Kozeny-Carman results
When comparing the two different Kozeny-Carman results, KKC1 are sometimes lower
than KKC2. KKC1 is the hydraulic conductivity when using the theoretically calculated
porosities and KKC2 is the hydraulic conductivity when using the experimentally
obtained porosities. This difference is especially clear when the uniformity coefficient
is high. As can be seen in Table 7, chapter 8.3, there are some high CU-values that
lead to a difference between the two methods of many orders of magnitude. The most
extreme case is in borehole KK4001 where the uniformity coefficient is 125.
However, the soil from this borehole is classified as sandy silty clay, and the value
can be viewed as an anomaly. Even without this extreme value, there are results from
Korsvägen with uniformity coefficients up to 40, which leads to a difference of four
orders of magnitude between Kozeny-Carman and Gustafson, from 10-10 to 10-6 at for
example borehole KK4002 and KK4011. The hydraulic conductivity is thus almost
ten thousand times higher when using KG than KKC, obviously one of these values is
not realistic. An important thing to take into consideration is that some of the D10-
values at Korsvägen are extrapolated and thus uncertain.
Table 10 Example of different porosities based on various degree of compaction and experimental
values. The values marked in grey are the porosities that best correspond to the experimental values.
There are however only a few samples which makes the comparison somewhat
uncertain and arbitrary. Using experimentally calculated porosities as a key to study
how the soil type determines the degree of compaction would be most interesting, and
could if done thoroughly lead to a simplification when using the Kozeny-Carman
equation to calculate hydraulic conductivity.
As Vienken and Dietrich (2011) showed, the method to calculate porosity must be
chosen carefully as this may skew the results. The conclusion from Frings et al.
(2011) that there are no unique relationships between grain size and porosity means
that it might be difficult to use the Kozeny-Carman equation without performing more
laboratory tests to determine the porosity, at least when investigating a more multi-
graded soil such as glacial till. However, the authors that have studied porosities in
the articles mentioned above have not used degree of compaction as a factor. The
correlation between experimentally obtained porosities and theoretically calculated
porosities seem to be higher when using this method. There might of course be
uncertainties in the experimentally calculated porosities if the in-situ conditions were
disturbed during sampling and laboratory work, which might render very different
results than the actual in-situ porosity.
9.2 Permeameter test
In the few cases were a permeameter were used to obtain an experimentally calculated
hydraulic conductivity, Kexp, these results are among the lowest in the tests. At
OC4012
OC4011
OC4010
OC4008
OC4009
Figure 19 Map of the boreholes at Skansen Lejonet (Google Maps, 2013b). Modified.
The results can be seen in Table 11. The previous and current sampling from the same
borehole is arranged between each other, with previous sampling marked with grey
cells. The results are arranged to facilitate easy comparison between the results where
the same method have been used.
The results show that the results between the different samplings are very small,
except for the borehole OC4008. Despite the fact that the two samplings were done at
the same time and with the same material, the results from the grain size analysis
differed a lot between the two samplings, which is quite inexplicable. For the rest of
the boreholes, the differences between the results are so small that the conclusion can
KK4007 KK4012
KK5038
KK4014
KK4003
KK5040 KK4001H
Figure 20. Map over Korsvägen. The two boreholes from current sampling are marked in black, and
the four boreholes from previous sampling closest to the current are marked in white. For a coloured
map, see Figure 13 in chapter 7.1.2 (Google Maps, 2013c)
The difference in depth between these two boreholes does however make the
comparison less direct. Borehole KK4014 is situated between KK5038 and KK5040,
and is also studied in this comparison. As can be seen in Table 12, the values are
generally around 10-6-10-5 for both previous and current sampling.
The comparison between borehole KK5038 and KK4007 and KK4012 show values in
approximately the same region between results calculated with the same method. KG
is as previously highest with values from 10-5 to 10-6. KKC1 and KKC2 vary more, from
10-5 to 10-8. The current sampling show somewhat lower results than the previous
sampling.
Table 13 Comparison between previous and recent sampling at Korsvägen. Grey cells show previous
sampling and white current.
Level 15.7-16.1 16.4-16.8 17.0-17.4 5-6 10-11 11-12 9-10 10-11 11-12
KKC1 8.3E-08 4.4E-06 6.0E-08 1.0E-06 1.7E-05 1.3E-05 1.2E-05 2.0E-05 1.9E-05
When comparing borehole KK5040 with KK4014 and KK4003, the trend from the
previous comparison continues. There are larger differences between the results based
on Kozeny-Carman than between the results based on Gustafson, which is not
unexpected as the Kozeny-Carman equation puts larger weight on uniformity
coefficients than Gustafson. As the sampling at Korsvägen are so scattered, it is
difficult to compare the many results from previous sampling with the few results
from current sampling. If more sampling had been performed, so that each borehole
from previous sampling would correspond to a borehole from current sampling, the
comparison would have been more relevant and accurate. As the soil at Korsvägen is
quite heterogeneous, especially compared to Skansen Lejonet, a more complete
comparison would have been preferred as the results are now difficult to put into a
more overall picture.
9.5 Comparison between grain size analysis and hydraulic
tests at Skansen Lejonet
Generally, the slug tests indicate a lower hydraulic conductivity than the grain size
analysis. With only two slug test results from Skansen Lejonet, confident conclusions
can’t be drawn, however when comparing slug test with KKC1, these results are more
Skansen
Korsvägen
1,00E-07
1,00E-06
1,00E-05
Very loose
Loose
1,00E-04
Medium
1,00E-03 Compact
Very compact
1,00E-02 Experimental
1,00E-01
1,00E+00
0,45
0,40
0,35
0,10 Experimental
0,05
0,00
SUMMARY
Groups Count Sum Average Variance
Column 1 14 -73,152 -5,22514 1,547315
Column 2 11 -57,9854 -5,2714 1,303981
Column 3 14 -61,6557 -4,40398 0,317735
Column 4 14 -61,849 -4,41778 0,434731
ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 9,214695 3 3,071565 3,505294 0,022073 2,793949
Within Groups 42,93697 49 0,876265
Total 52,15166 52