Permeability Determination in An Alluvial Dam Foundation: C&technique

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

B. C. (1986). C&technique 36, No.

1,95S108
SOLMAR, Z. V. & ILOABACHIE,

Permeability determination in an
alluvial dam foundation

Z. V. SOLYMAR* and B. C. ILOABACHIEt

A series of in situ and analytical tests was performed to observatioq piezometers, Lefranc-Mandel tests at
evaluate the coefficient of permeability of the river-bed a number of depths in cased boreholes, computer
sediments at the Jebba Dam site. The study involved models of the simulation of aquifer response to
falling head tests, pumping tests on screened wells,
river fluctuation and coefficient of permeability
Lefranc-Mandel tests, computer models of the simula-
tion of aquifer response to river fluctuation and coefli- calculations based on grain size distributions of
cient of permeability calculations based on grain size sand samples.
distributions of sand samples. Test results are compared The determination of the foundation per-
with the coefficients obtained from piezometric data meability at Jebba was a key element in the
and flow from pressure relief wells during and after design of the coffer-dams and also of the main
impounding. dam, particularly since the permeability of granu-
lar foundation materials is subject to a much
Une skie d’essais analytiques et in situ a ktk r&alike wider range of variations than are the other
pour kvaluer le coefficient de permkabilitk des skdiments physical properties of such materials. A computer
en lit de la rivikre au site du barrage Jebba. L’btude a
program developed by Verruijt (1970) was used to
comporti: des essais sous gradient variable, essais de
pompage sur puits crtpinis, essais Lefranc-Mandel, analyse the seepage through the alluvial founda-
modkle sur ordinateur de la simulation de la rtponse de tion. Since the possibility of cracks developing in
la nappe phrkatique aux fluctuations de la rivitre, et the impervious blanket during the life of the
calculs du coefficient de permtabilitt basks sur la granu- project cannot be precluded, its effect on under-
lomitrie d’tchantillons de sable. Les rCsultats des essais seepage was also determined (Solymar, MacTav-
sont cornparks aux coefficients obtenus B partir de don- ish & Matthews, 1985). The Contractor had the
n&es pikom&riques et de debits dans des puits filtrants responsibility for the design of the coffer-dams.
durant et apr6s la mise en eau.
The dam has been completed and subjected to
its hydraulic loading. The instrumentation incor-
KEYWORDS: dams; field tests; groundwater; per-
meability; sands; seepage. porated in the foundation makes it possible to
compute the actual coefficient of permeability of
The Jebba main dam across the Niger River, the foundation alluvium and to compare it with
Nigeria, is a zoned earth-rockfill embankment the coefficients obtained from the series of tests
42 m high founded on river alluvium up to 70 m and analyses that were performed previously.
in depth. The foundation conditions encountered
at Jebba are typical for many dam sites, especially RIVER ALLUVIUM
‘in Africa. The control of seepage through the The river-bed alluvium is fine-coarse-grained,
alluvial foundation of the dam is accomplished by mostly uniformly graded, clean quartzitic sands
the provision of an impervious blanket extending with traces of fine gravel. The gravel that has
upstream from the base of the impervious core. been encountered has always been the coarser
De-watering of the dam foundation area required portion of a fine-medium-coarse sand mixture
the construction of three coffer-dams, two with and never in the form of gravel beds. Grain size
cut-off walls to bedrock. envelopes of 338 samples are shown in Fig. 1.
A series of in situ and analytical tests was per- Laboratory density tests of the alluvium gave
formed to evaluate the coefficient of permeability an average minimum dry density of 1620 kg/m3
of the river-bed sediments. The study involved and an average maximum dry density of 1920
falling head tests in cased boreholes and in kg/m3. The average maximum and minimum
piezometers, pumping tests on screened wells with voids ratios were found to be 0.65 and 0.39
respectively. The unit weight of the saturated
Discussion on this Paper closes on 1 July 1986. For sand varies between 19.3 kN/m3 and 21.1 kN/m3,
further details see inside back cover. with an average of 20.1 kN/m3.
* Monenco Consultants Ltd, St Catharines. The uniformity coefficient varies between 1.52
t National Electric Power Authority, Lagos. and 8.83, with an average value of 2.94, while the
95

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
96 SOLYMAR AND ILOABACHIE

GLWfS Sand

Slit Clay
CLWKX Fine CCWSe Medium Flfl62
100

Fig. 1. Grain size envelope

effective grain size (d,,) varies from 0.06 mm to FIELD AND LABORATORY TEST
2.3 mm, with an average of 0.31 mm. PROGRAMME
In situ relative densities of the river sand, which Feasibility study testing
vary from 35% to 85%, were determined from Falling head tests in cased boreholes on the
dynamic and static cone and standard penetrat- river alluvium were conducted as part of the geo-
ion test data. Before construction of the dam, the technical investigation for the feasibility study of
upper 20 m below the original river-bed had been the project. Eleven holes were drilled from barges
densified to 70%, the layer between 20 m and in well-distributed locations throughout the
30 m depth to 60% and 50% relative density blanketdam-foundation area (Fig. 2). Per-
below this level, using vibrocompaction and deep meability testing was conducted in 82.5 mm
blasting techniques (Solymar, 1984; Solymar, inside diameter cased holes with flush bottoms.
Iloabachie, Gupta & Williams, 1984). Of the 29 separate tests performed in six selected

test I” borehole

Scale of metres

Fig. 2. Location of boreholes and falling head permeability tests

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
98 SOLYMAR AND ILOABACHIE

nstrumented sectIon

Fig. 4. Location of pumping tests, drill holes and observation wells

TEST AND ANALYTICAL PROCEDURES where K is the intrinsic permeability in darcys, d,


Grain size is the main diameter in millimetres and u+ is the
A variety of empirical and theoretical formulae standard deviation of d, in 4 units.
have been devised for calculating the coefficient of The coefficient of permeability k is related to
permeability of porous media from grain size dis- the intrinsic permeability K by the equation
tributions and the hydraulic properties of the
fluid. After a review of different methods devel- k=!%!
oped by several authors, the formulae proposed P
by Hazen (1892, 1908) and by Krumbein & Monk
(1943) were selected for application to the grain where p is the density of water, g is the gravita-
size data obtained during this study. The Hazen tional constant and p is the viscosity of water.
equation can be expressed as Hence, for water at 28 “C the Krumbein-Monk
formula reduces to
k = 10-2Cd,,2(0~70 + 0.03T)
k = 8.8 x 10-3d,Z exp (-1.31) m/s
where k (m/s) is the coefficient of permeability, C
is a constant, d,, (mm) is the effective grain size Falling head tests
and T (“C) is the temperature of the water. In the The falling head method in cased boreholes
literature (Cedegren, 1968), the Hazen equation and in piezometers was the simplest test pro-
for sands of high uniformity is often expressed cedure employed at the site. In boreholes, the drill
simply as casing, extending OG1.0 m above the river, was
filled to the top with water and measurements
k = Cdlo2 cm/s were made of the rate of fall of the water level
(Fig. 5). For determination of the mean coefficient
in which d,, is in centimetres and C varies from
of permeability the following formula compiled
about 90 to 120. With C = 100 and d,, in milli-
by Hvorslev (1951) was used
metres, the formula reduces to
nd2
k = 10-Zd,,2 m/s k, =
llD(t, - tJ
Krumbein and Monk performed a series of
laboratory experiments in which they evaluated where d (m) is the diameter of the casing, D (m) is
the permeability of artificially mixed sands in a the diameter of the sample, and h,’ and h,’ (m)
constant head permeameter. The resulting are piezometric heads for times t, and t, (s).
formula was Falling head tests were performed in nearly all
piezometers installed for the pumping tests (Fig.
K = 760d,’ exp (- 1.31aJ 5). Most tests were performed twice to check the

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
PERMEABILITY DETERMINATIONS 99

the water level response may be that of an over-


damped, critically damped or underdamped oscil-
lator (Van Der Kamp, 1976). During the field
tests all three types of response were observed.
Fig. 6 shows samples of data from falling head
tests in piezometers. Numerous analytical pro-
cedures have been developed to analyse data
-_ ;’ from the overdamped type of response. For this
. .=--Gravel ! I study, the methods of Hvorslev (1951) and
Bouwer & Rice (1976) were used. An approximate
method has been developed by Van Der Kamp
(1976) to estimate the coefficient of permeability
Fig. 5. Falling head test in cased borehole, Lefranc- on the basis of the results of falling head tests
Mandel teat and falling head test in piezometer with underdamped response. To date, no analyti-
cal methods have been developed to handie the
results. Water levels were measured using quick critically damped response.
response pneumatic transducers with a settlement
gauge read-out. The read-out instrument could be Lejiianc-Mandel test
read to the closest 12.5 mm. At the start of each Several pumping-in and pumping-out tests
test, the transducer was placed in the piezometer were performed in a selected borehole. After drill-
with a submergence of at least 1 m to ensure that ing to bedrock a clay plug 1 m long was com-
the response with depth was linear, and water pacted with a heavy rammer at the bottom of the
was then added until the water level was at the hole. Rounded gravel was placed through the
top of the piezometer pipe. The falling water PQ size drilling rods (117.5 mm outside diameter,
levels were recorded at 2 s intervals for the first 103.2 mm inside diameter) to fill a length about
20 s and then at 5 s intervals up to 1 min. When 3 m high, while the rods were slowly withdrawn.
a slug of water is added to a well or piezometer, A constant water level tank was used to supply a

3000
Overdamped response Crh~cally damped
Ptezometer P2-4C response
1000 (lest 2) Pierometer OP-1 C
(lest 1)
A
E 500 \
E t\

15oog
Underdamped response

El000 R Piezometer
(test 1)
PZ-3C

Fig. 6. Examples of falling head test responses

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
100 SOLYMAR AND ILOABACHIE

,y;:..._ ..,.

of mllllmetres

o- M _i._._’
_’
Fig. 7. Pumping well details

steady water flow for the pumping-in tests and a the Neuman (1975) model and applicable Dupuit
100 mm submersible pump for the pumping-out (Cedegren, 1968), Babbitt & Caldwell (1948) and
tests. For the pumping-out tests the overflow was Boreli (1955) equations. The Neuman model rep-
measured in graduated tanks and the drawdown resents a well pumping from an areally infinite
with an electrical dip meter. aquifer. The sources of water assumed by the
Between two successive test lengths, the bore- model are from storage at the water-table and, to
hole was filled with fine sand covered with an a lesser extent, from elastic expansion of water.
impervious clay plug (Fig. 5). The following However, test pumping from the sand pad
formula compiled by Hvorslev (1951) was used involved rapid induction of recharge from the
for calculating the horizontal coefficient of per- river which prevented a further decline of the
meability water-table as the river replaced the water-table
as the source of water.
For this reason, coefficient of permeabilities
obtained using the Neuman model were looked
on as approximations. In addition, river-induced
where q (m3/s) is the volume of water, k, (m/s) = fluctuations were sometimes of greater magnitude
(k k )I’* is the mean coefficient of permeability, L than those imposed by the pumping test.
(m) i’, the length of tested area, d (m) is the diam- The most commonly used Dupuit well formula
eter of the casing, D (m) is the diameter of the
sample and h,’ (m) is the constant piezometric k= ’ In ‘z
head. r@,* - hi*) 0 r1
is based on the assumption that a steady state
Pumping test flow condition exists and that the water-bearing
Two pumps were used during the pumping test formation is homogeneous, isotropic and extends
programme at sites PW 1, PW 2 and PW 3 (Fig. an infinite distance in all directions (for symbols
4): a submersible pump capable of producing see Fig. 8). The reliability of a well pumping test
about 3.3 m3/min and a suction lift pump depends on how accurate the above assumptions
capable of producing up to 5.2 m3/min. The dis- are fulfilled. At Jebba, owing to space restrictions
charge from the pumped wells was measured and densification work in progress, the wells were
using orifice weirs. The pumping tests were located relatively closely to the bank of the river
undertaken in stages, starting with the well fully and the infinite distance assumption became
penetrating the saturated thickness of the river- invalid. In addition, to facilitate the distinction
bed alluvium. Following the initial tests at sites between response to test pumping and response
PW 2 and PW 3, grout curtains 3-6 m deep were to river level fluctuations, the observation
installed around each pumping well to isolate the piezometers were located as closely as possible to
sand pad hydraulically from the well as much as the pumped wells. If the distance between the well
possible (Fig. 7). Following completion of these and the observation piezometer is less than one
pumping tests, the wells were grouted in stages and a half times the thickness of the aquifer, the
from the bottom and the pumping tests repeated coefficient of permeability can no longer be calcu-
at each stage. All depth-to-water measurements lated by means of Dupuit equations, because the
were made with standard dip meters. real drawdown curve lies considerably above the
The pumping test results were analysed using Dupuit curve on which the equation is based.

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
PERMEABILITY DETERMINATIONS 101

Duput when it is drained by gravity. It is expressed as a


surface Actual drawdown volume percentage. Specific retention is defined as
the volume of water retained by a unit volume of
aquifer material after complete gravity drainage.
The sum of the specific yield and the specific
retention is equal to the porosity of the aquifer
material. A schematic cross-section of the appar-
atus used for the laboratory determination of
these volumes is shown in Fig. 9. Two samples of
sand were air dried and tested in the apparatus.
For tests on uncompacted sand, the sand was
poured slowly into the cylinder and without any
compaction. For the tests on compacted sand, the
sand was tamped as the cylinder was filled. A
water reservoir was suspended about 1 m above
the test apparatus with a hose running to the
Fig. 8. Typical arrangement for determining soil per- valve at the base of the cylinder. The cylinder was
meability by the well pumping test (after Cedegren, filled with water through the bottom valve to
1968) minimize trapping of air bubbles in pore spaces.
The volume of water required to saturate the
Equations devised by Boreli (1955) from more sand was measured and the porosity was calcu-
advanced theories are available. lated. The specific yield tests were done both by
filling and draining the apparatus. Several filling
qP In (R/H) and draining tests were performed on each sand
k=
7tH(H - h)[ 1 - O%(s/H)‘+J sample at different densities.

for 0.3 < r/H < 1.5 Computer model


Water level measurements between tests were
P=O.l3ln R made to enable assessments of background water
0r level changes in response to river level fluctua-
for rjH < 0.3 tions, water injections associated with ongoing
P=C,fAC vibrocompaction of the dam foundation and
where floodings associated with sand pad construction
by hydraulic emplacement of dredged sand. As
C,=O.l3ln R -0~01231n2 i these responses were of similar and sometimes
0 ( ) greater magnitude than the responses to the

AC=$$3ln~~)(l.2~-04g)

+O~ll3ln(~)ln($)]
Plezomete
Plastic
tubing,
cylinder
To use these equations, however, the observa-
, * .. ,. : . ,
tion piezometer must not extend appreciably ..*. ._.
.e, ., . . , .:. :: ‘.
below the lowered position of the water-table . . .: :. ‘..:. ‘:. -.:
. ,’ . . _ . . .‘.. . . .
because, at distances from the well less than .. . . . ..,: .’ ‘-
. -. * . . ..
about the thickness of the aquifer, the piezometric ..s .,,..‘.‘:‘..:, . . . ;.
,. . . . . .. . ..-..
levels are not the same along a given vertical line. I . ... .. Sand, . .-* *”
The Neuman graphical method was used to esti-
-.**8.,, .. . : ,... ,: ‘_‘....,
‘...’
mate coefficient of permeability values for all test . . :.
‘. ’ x: :._: ; .,. ‘.‘..,.
,.:::.. :“’ ..
sites and Boreli’s equation for test site PW 2,
‘.: . ‘. , ‘:.- :,..,;, ‘,
where several shallow piezometers were installed , *. .‘.‘,.
. . ..._..a,.,:,
in addition to the deep piezometers. *.,.-‘.-
. . ~ . ..,-.‘.!,, , . .

Spec$c yield test


Coarse sand’
Specific yield is defined as the quantity of water lalve

that a unit volume of aquifer material gives up Fig. 9. Specific yield test apparatus

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
102 SOLYMAR AND ILOABACHIE

RESULTS OF LABORATORY AND FIELD


TESTS
Table 1 presents a summary of all calculated
coefficients of permeability from laboratory and
in situ tests performed at Jebba.
Laboratory test results for samples obtained
from drill holes gave low values for the coefic-
ient of permeability (k = 9 x lo-’ m/s to k =
8 x 10m5 m/s) and indicated a slight increase with
increasing relative density (Table 2).
The Hazen formula generally gave lower coef-
ficient of permeability values from the grain size
analysis than the Krumbein and Monk formula.
A comparison of the results suggests that the
Hazen values continue to be valid for a uni-
formity coefficient (c,) greater than 3 (Table 3). A
frequency distribution of the Hazen permeability
Fig. 10. Model representation of the river-sand pad results has been plotted in Fig. 12. The over-
aquifer system (for section A-A see Fig. 11) all average permeabilities ranged from
8.67 x 10e4 m/s to 944 x 1O-4 m/s depending
pumping tests, it was considered appropriate to on whether or not results larger than
develop a mathematical model to simulate the 25 x 10m4 m/s were included in the averages.
transmission of the river level fluctuations Because the number of anomalously high coe.f-
through the alluvial sand deposit to the piezom- ficient of permeability values is small and no con-
eters. By trying various coefficient of permeability tinuous layers of coarse sand/gravel were
values until the simulated responses coincided detected, it is felt that the average coefficient of
with the actual response observed at each permeability based on grain size analysis is closer
piezometer, a range of bulk coefficient of per- to the lower limit.
meability values was determined from the graph Results from falling head testing of the river
of water levels at each piezometer. Prickett & sands show a considerable scatter of values, con-
Lonnquist (1971) developed a finite difference firming anisotropic permeability conditions.
model code in FORTRAN digital language. The From Table 1, it can be seen that the range of
model simulates aquifer response for a wide range coefficient of permeability values determined from
of conditions, but it was designed as a two- the falling head test in piezometers falls within
dimensional model, to operate in a flat, horizon- quite a narrow range and the mean is just over
tal representation. For its application at Jebba, 1 x 10e4 m/s. There is very little difference in
the model was modified to represent a wedge as either range or the mean coefficient values
shown in Fig. 10. Fig. 11 shows the model rep- obtained from Lefranc-Mandel tests or other
resentation of the river-sand pad aquifer system borehole permeability tests.
and the finite difference grid geometry in vertical An attempt to analyse pumping test data from
section. Two parameters, the estimated coefficient all three sites using the Neuman model was made.
of permeability and the specific yield obtained In this analysis the aquifer replenishment by
from laboratory tests of the sand pad materials, recharge from the river was disregarded. The
were embodied in the mathematical model. application of Boreli’s formula to the data from

Column
Piezometers
Rwer sediments her
/
Sand pad I

I
0 50 100 150 200 250 300 350
Bedrock
Distance m

Finite element grid geometry in vertical section (A-A)

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
PERMEABILITY DETERMINATIONS 103

Table 1. Coefficient of permeability determination of foundation sands

Type of test and/or analysis


T Coefficient of permeability:

Range
x lo-‘+ m/s

Mean
Laboratory 0~009-0~8 0.5
Grain size
Hazen’s formula 0.4-84.0 9.4
Krumbein and Monk’s formula 2.5-62.0 14.0
Falling head 0.6-9.2 3.9
Lefranc-Mandel 0.3-2.1 0.8
Slug 0.4-2.3 1.2
Pumping
PW I-Neuman’s analysis 0.1-1.6 0.6
PW2-Neuman’s analysis 2.1-15.2 8.4
Boreli’s formula 4.5-6.4 5.8
PW3-Neuman’s analysis 0.1-4.4 1.2
Simulation of aquifer response to
river fluctuations 8X)-31.0 18.0

Table 2. Coeffkient of permeability testing in the laboaratory

Sample Depth: m Coefficient of permeability : 10 - 5 m/s

gJ = 196 kPa c3 = 392 kPa (rs = 588 kPa


Relative density: %

50 70 90 50 70 90 50 70 90
1 6.1&7.01 6.15 5.60 5.80 5.87 560 560 5.77 5.40 540
2 9.14-10.06 6.90 7.10 7.20 6.40 6.90 7.10 6.20 6.70 5.80
3 13.72-14.63 5.64 5.80 7.90 5.55 560 7.90 5.19 560 7.80

Table 3. Coefficient of permeability determination from grain size*

Location Hazen (with c, < 3) Hazen (with c, 3 3) Krumbein and Monk


I I
n R k o n R k 0 n R k D

PW 1 65 3.640 14.8 6.5 95 3.2-84 15.5 10.0 17 4-7-62 20 14.0


PW2 109 3.2-30 8.8 5.3 161 3G30 9.0 5.2 27 3.5-22 11 5.8
OP 1 3 4.2-16 10.6 5.9 6 4.2-l 6 8.9 4.2 7 5.618 11 4.7
OP2,OP3 22 2.5-27 9.8 7.0 24 2.5-27 9.8 7.0 24 2.5-27 13 8.1
BH l-5 82 1.2-11 4.8 2.6 121 1+14 4.9 2.9
DH and SP 17 2+26 9.8 5.7 31 0.446 11.2 9.0
holes
* n, number of samples analysed; R, range of coefficient of permeability values (x lo-“ m/s); k, average of coefficient
of permeability values ( x 10m4 m/s); (r, standard deviation.

site PW 2 yields coefficient of permeability values yield on two samples of sand pad material, the
that are slightly lower than those found by using mean specific yield was found to be 254% for an
the Neuman model (Table 4). Results obtained uncompacted sand and 18.1% for the compacted
using Boreli’s formula are not influenced by river sand (Table 5). For hydrograph modelling the
fluctuations; therefore they are considered to be average specific yield of 20% was taken. Table 6
more reliable under the Jebba conditions than presents coefficient of permeabilities obtained
those obtained by using Neuman’s model. from the fluctuating water-table simulation
On the basis of laboratory tests for aquifer method.

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
104 SOLYMAR AND ILOABACHIE

alluvial foundation material in three sections (Fig.


13). The aquifer response recorded at each
piezometer during operation is a function of all
With all values of c, the coefficient of permeability values along the
path of seepage between the reservoir and the
particular piezometer.
The coefficient of permeability can be calcu-
lated from piezometric data (Fig. 14) and flow
from pressure relief wells using seepage formulae
from Turnbull & Mansur (1960) (for symbols and
notation see Fig. 15)
k=-%_-
aH AM
where
A&&?!?_“_h,,
S x3

and from Casagrande (1961)

4, = 91- 4 P
where
Coefflclent oi permeablllly X 10 4 m/s
q, = kHi,a
Fig. 12. Frequency distribution of Hazen permeability
qp = kHi,a
results
then
OBSERVATIONS DURING RESERVOIR q, = kHa(i, - it)
OPERATION hence
To observe the performance of the impervious k= ‘,
blanket, 36 pore pressure cells were placed in the Ha(i, - i,)
Table 4. Coefficient of permeability determination from pumping tests

l-
Location Piezometer Coefficient of permeability: x 10m4 m/s

Designation Depth of
installation
below
water-table:
Distance from
,centre of well:
m
Test 1
T Test 2
T Test 3

m Neuman Boreli Neuman Boreli Neuman Boreli

PW 1 2A 14.4 2.7 0.19 0.65


2c 38.1 2.7 0.14 0.54
3A 15.0 7.8 1.16 0.59
3c 36.0 7.8 1.60 0.59
4A 14.0 2.7 0.78 0.29
4c 38.5 2.7 0.42 0.30
5A 14.3 7.8 0.27 0.61
5c 38.0 7.8 0.81 044
PW 2 2A 18.9 3.2 7.71 8.25 5.17
2D 1.5 2.8 6.07 6.07 7.63 6.38 6.72 6.12
3A 16.5 8.0 1040 15.25 12.42
3D 1.5 8.1 6.00 6.08 10.88 6.40 6.61 6.08
4A 19.2 3.0 2.05 12.42 12.42
4D 1.5 3.2 3.87 5.91 7.79 5.43 6.61 4.72
5A 18.8 7.7 2.95 13.31 14.46
5D 1.5 7.2 3.36 5.87 10.10 5.43 10.05 4.64
PW 3 1 57.1 10.2 1.40
2 37.6 10-O
3 10.6 10.2 4.40
4 42.0 5.4 0.69
5 25.2 5.0 0.40 0.15
6 8.6 5.4 0.39

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
PERMEABILITY DETERMINATIONS 105

Table 5. Specific yield test results

Sample Sample Sample Porosity: Test Specific


compaction density : % type yield :
kg/m3 %
No 1610 36.3 Draining 26.9
Draining 23.5
Filling 26.2
Filling 24.4
Yes 1770 32.3 Draining 21.4
Draining 18.0
Filling 19.5
No 1600 40.3 Draining 26.5
Filling 24.9
Yes 1740 37.1 Draining 19.6
Draining 18.4
Filling 17.6
Filling 15.0
Filling 15.3

Table 6. Coeffkient of permeability determination from fluctu-


ating water-table simulation

Coefficient of permeability: x 10m4 m/s

6 Dec. 1980 18 Jan. 1981 8 Feb. 1981

OP IA 16 12 9
OP 1B 11 10 14
OP 1c 17 8 13
OP 2A 25 7 13
OP 2B 28 14 15
OP 2c 28 14 14
OP 3A 27 15 15
OP 3B 30 16 15
OP 3c 31 15 13

J 100 ~ 78.8
P P

Scale of metres

Fig. 13. Typical main dam cross-section and instrumentation

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
106 SOLYMAR AND ILOABACHIE

Scale of metres shown in Table 7. The weighted average is


54 x 10m4 m/s.

CONCLUSIONS
A summary of all coefficient of permeability
values obtained at Jebba is presented in graphical
Computed potenltal drop form in Fig. 16.
wlfh k, = 4 X 10 ’ m/s Comparing measured bulk coefficient of per-
meability values obtained from piezometer read-
ings and relief well discharges with values
obtained from laboratory and field tests it is con-
cluded that laboratory tests results are low by a
magnitude of 100 and Lefranc-Mandel and
falling head tests in piezometers by a magnitude
of 4-10. The upper range of the results from the
falling head tests in cased holes and the pump test
results, even with all the apparent inaccuracies in
boundary conditions, have been the most reliable.
The simulation of aquifer response to river fluc-
tuation gave values that were higher than those
found with the prototype during operation of the
reservoir. The analytically derived results using
the Hazen formula show a great scatter. More
than 430 samples were analysed and the average
varied from borehole to borehole. The overall
average (8.67 x 10e4 m/s) is remarkably close to
the weighted average (54 x 1O-4 m/s) obtained
during operation. The selection of the constant C
in the Hazen formula introduces a great
unknown. It is not suggested that the results
obtained using the Hazen formula are used for
design; however, if a large number of samples is
taken, the grain size is useful in indicating the
3 4 5 6 7
Time months wide range of permeabilities to be expected.
The most important factors influencing per-
Fig. 14. Piezometer readings
meability are
where
(a) viscosity and temperature of the fluid flowing
iw zz-
h, - h, through the porous media
Xl (b) size of the pore space
and (c) size of soil particles
i, = -h, - h, (d) sorting or stratification of the soil.
x2 At Jebba, variations in the size of pore space
Good correlation between the results calculated and soil particles had the greatest effect on per-
with the two different formulae was found, as meability, since the viscosity and temperature of
the water are nearly constant and no significant
stratification was detected. Density or pore
volume of the alluvium had very little effect on

j===JFgki&.& the permeability as indicated by the laboratory


test results. This observation is supported by the
results of the full-scale model, since no noticeable
reduction in the range of coefficient of per-
meability occurred in the clean, uniformly graded
9” quartzitic sands found at Jebba, after the exten-
sive in situ densification programme.
Fig. 15. Typical arrangement for determining the per- From these studies it can be concluded that it
meability of the foundation from piezometer readings is prudent to use the higher coefficient of per-
and flow from pressure relief wells meability values obtained from simple borehole

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
PERMEABILITY DETERMINATIONS 107

Table 7. Coefiicient of permeability from piezometer readings and relief well discharge*

Period Section Turnbull and Mansur’s formula Casagrande’s formula


I
n R k fs n R k g

Impounding MA 24 6.1-12.3 9.06 1.60 23 6.c12.4 8.79 1.78


MB 15 8.3-15.9 10.76 2.63 15 8.4-15.6 10.97 2.47
MC 12 2.3-6.2 4.59 1.21 12 2.7-7.2 5.39 1.42

Full MA 21 7.1-15.8 10.97 2.84 21 7.1-15.6 10.96 2.75


reservoir MB 20 3.8-13.6 7.11 244 20 3.5-13.9 7.07 2.52
MC 18 0.34.7 2.14 1.45 18 0.54.7 2.26 1.45

Weighted
average

* n, number of cases analysed; R, range of coefficient of permeability values (x 10m4 m/s); k, average of
coefficient of permeability values ( x 10e4 m/s); o, standard deviation.

Laboratory
*-
Grain SILO (Hazen. 1892.1908)

Grain size (Krumb*em &, Monk, 1943)

Fallmg head
*-
Letranc Mandel

Falling head in plezOmeier5


-*-
Pumping at PW ‘, (Neuman, 1975)

Pumping at PW 2 iNeuman. 19751

Pumping at P-W*; (&=955)

Pumping al PW 3 (Neuman. 1975)


*
Aquifer response model
-*-
Instrumented sectlons 01 completed maln dam
-w-.- *-.-I

L
0.1 05 1 5 10 50 100
Coefflclent of permeabllliy X 10 4 m/s

Fig. 16. Summary of coefficients of permeability

tests for the design of structures with a short REFERENCES


seepage path, e.g. coffer-dams, and an average Babbitt, H. E. & Caldwell, D. H. (1948). The free surface
value for structures with longer seepage paths. It ground, and interference between grauity wells. Bulle-
tin Series No. 374, Engineering Experiment Station,
is also concluded that at least two pump tests at
University of Illinois.
separate locations should be performed to obtain
Boreli, M. (1955). Free-surface flow toward partially
more representative results. penetrating wells. Trans. Am. Geophys. Un. 36, No.
4,664672.
ACKNOWLEDGEMENTS Bouwer, H. & Rice, R. C. (1976). A slug test for deter-
Consulting engineering services for the Jebba mining hydraulic conductivity of unconfined aquif-
Hydroelectric Development were provided by ers with completely or partially penetrating wells.
Montreal Engineering Company, Limited. The Wat. Resour. Res. 12,423-428.
pumping tests were performed by BERG Engin- Casagrande, A. (1961). Control of seepage through
foundations and abutments of dams. GPotechnique
eering and Foundation Engineering (Nigeria) Ltd.
11, No. 2, 161-181.
The Owner is the National Electric Power
Cedegren, H. R. (1968). Seepage, drainage, and flow nets.
Authority. S. K. Ho compiled the data and New York: Wiley.
assisted C. A. Bostock in modifying existing com- Hazen, A. (1892). Experiments upon the purification of
puter models to suit specific site conditions. sewage and water at the Lawrence Experiment

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.
108 SOLYMAR AND ILOABACHIE

Station. 23rd Annual Report, Massachusetts State Solymar, Z. V. (1984). Compaction of alluvial sand by
Board of Health. deep blasting. Can. Geotech. J. 21, No. 2, 305-321.
Hazen, A. (1908). Thefiltration of public water supplies. Solymar, Z. V.,-Iloabachie, B. C., Gupta, R. C. & Wil-
New York: Wiley. liams, L. R. (1984). Earth foundation treatment at
Hvorslev, M. J. (1951). Time lag and soil permeability in Jebba dam site. J. Geotech. Engng Div. Am. Sot. Ciu.
ground-water observations. Bulletin No. 36, Water- Engrs 110, No. 10, 1415-1430.
way Experiment Station, US Army Corps of Engi- Solymar, Z. V., MacTavish, G. C. & Matthews, W. G.
neers, Vicksburg. (1985). Some design aspects and performance of the
Krumbein. W. C. & Monk. C. D. (1943)./ I
Permeabihtv Jebba Main Dam. Trans. 15th ICOLD Congr., Lau-
as a function of the size parameters of unconsoli- Same 1, 199-221.
dated sand. Trans. Am. Inst. Min. Metall. Engrs 151, Turnbull, W. J. & Mansur, C. I. (1960). Underseepage
153-163. and its control, a symposium. Trans. Am. Sot. Civ.
Neuman, S. P. (1975). Analysis of pumping test data Engrs, Part I 126, 1427-1568.
from anisotropic unconfined aquifers considering Van Der Kamp, G. (1976). Determining aquifer trans-
delayed gravity response. Wat. Resour. Res. 11, No. missivity by means of well response tests, Wat.
2, 329-342. Resour. Res. 12, No. 1, 71-17.
Prickett, T. A. & Lonnquist, C. G. (1971). Selected Verruijt, A. (1970). Plane flow with sources and sinks.
digital computer techniques for groundwater Program LM4, Ottawa.
resource evaluation. Illinois St. Wat. Surv. Bull., No.
55.

Downloaded by [ UC San Diego Libraries] on [13/09/16]. Copyright © ICE Publishing, all rights reserved.

You might also like