You are on page 1of 18

Pow&, W. & Preene, M. (1994). Gotechnique 44, No.

1, 83-100

Time-drawdown behaviour of construction dewatering


systems in fine soils

W. POWRIE* and M. PREENEt

In the design of a construction dewatering system Lors de la conception d’un systeme drainant dans
in a fine-grained soil it is necessary to consider not un sol granulaire fin, il faut a la fois prendre en
only the steady state flow rate that must be compte le debit en regime permanent qui doit Etre
pumped, hut also the time it will take to achieve a pomp& et les temps mis pour atteindre un certain
given drawdown. This may be weeks because the rabattement. Ce temps peut&re de plusieurs
soil will usually remain saturated, and the mecha- semaines puisque le sol reste genirallement sat&
nism of pore-water pressure reduction is consoli- et que le mbanisme gouvernant la diminution de la
dation rather than dewatering per se. If the pression interstitielle est la consolidation plus que
response time is not taken into account in pro- le drainage. Si le temps de response n’est pas pris
gramming an excavation, the cost implications en compte lors de la conception d’une excavation,
could be considerable. In this Paper approximate les cotits peuvent augmenter considerablement.
methods based on the assumption that the iso- L’article presente des methodes approach&s, fon-
thrones are parabolic are developed for the estima- d& sur l’hypothese que les isochrones sont para-
tion of time-drawdown relationships for boliques, qui permettent d’approcher les relations
dewatering systems in fine soils. The parabolic iso- temps-rabattement pour des systemes drainant
throne analyses are compared with exact solutions dans des sols fins. Les ri?stdtats des analyses des
where these exist, and with field data from five isochrones paraboliques sont compares a des solu-
case studies. The parabolic isochrone approx- tions exactes, lorsqu’elles existent, et a des don&s
imation is shown to be suitable for flow to lines of in-situ provenant de 5 cas d’etude. L’approxi-
wells, and to rings of wells, provided that the dis- mation des isochrones par des paraboles semble
tance of influence remains small compared with adaptee aux lignes de puits et aux cercles de puits
the radius of the ring. si le rayon d’action des puits est faible devant le
rayon du cercle de puits.
KEYWORDS: case history; consolidation; excavation;
groundwater; permeability; pore pressures.

INTRODUCTION (Roberts, 1988); the time to full drawdown is


Construction dewatering systems in fine soils are rarely a critical parameter in design. In fine soils,
usually concerned primarily with the temporary such as silts or silty sands, for the same aquifer
reduction of pore pressures to improve the stabil- boundary conditions pore-pressure reduction will
ity of excavations, rather than the prevention of occur more slowly and its rate will control the
flooding. Pumped flow rates are likely to be low, rate at which excavation can proceed. Some esti-
and conventional well systems ineffective due to mate of the time necessary to achieve a given
losses at the soil-well interface. To overcome drawdown is required so that programming deci-
these problems, vacuum wellpoint or ejector sions can be made. In this Paper, an approximate
systems are often used. approach to the prediction of timeedrawdown
In coarse soils, such as gravels or coarse sands, relations for dewatering systems in fine soils is
drawdown around the dewatering system nor- developed, and its limitations are explored with
mally occurs rapidly. Typically, the required reference to field data from a number of case
drawdown is achieved within a few hours or days studies.

Manuscript received 1 September 1992; revised manu- CONVENTIONAL METHODS OF ANALYSIS


script accepted 7 June 1993.
When the phreatic surface is lowered in a
Discussion on this Paper closes 1 July 1994; for further
coarse soil, water drains freely from the pores to
details see p. ii.
* Queen Mary and Westfield College, University of be replaced by air, and the soil is literally
London. dewatered. In contrast, in fine soils surface
t Formerly Queen Mary and Westfield College, now tension resists the entry of air into the pores, with
WJ Engineering Resources Ltd. the result that water cannot drain freely by air

83
84 POWRIE AND PREENE

127
kN/m3. Fig. 1 shows the range of u,/y, evaluated
from equation (1) for e = 0.4-0.6 as a function of
permeability k estimated using Hazen’s rule (k =
O~01D,,2; k is in m/s and D,, is in mm: Hazen,
1892). For fine soils with permeability k < 10m5
m/s, u.Jr, may be of the order of several metres,
and for typical construction dewatering draw-
downs, little or no drainage by air entry will
occur. Since, if the soil remains saturated, pore-
water pressure reduction at constant total stress
requires the effective stress to increase, the prin-
cipal mechanism of drainage will be consoli-
Permeability k: m/s dation.
Theis (1935) presented a solution that is often
Fig. 1. Variation of air entry head with permeability k
used to model the consolidation of a confined
aquifer due to pumping from a single well. This
solution is of limited use for the modelling of
replacement, and fine soils can remain saturated
drawdowns around construction dewatering
with negative pore pressures. In general, a soil
systems in fine soils, for the following reasons:
will begin to desaturate only when the pore pres-
sure falls below the air entry value u,. The likely (a) The Theis solution assumes that the flow rate
range of u, can be related to particle size by con- from the well is constant, while the drawdown
sideration of surface tension effects (e.g. Bolton, in the well varies with time. In practice, in fine
1991), or by use of the empirical formula given by soils the capacity of the pumping equipment
Terzaghi & Peck (1967) will normally be much greater than that of the
well, so that after the first few minutes of
o.o5y, pumping, the drawdown in the wells will
- - < U, < - > (kPa)
eD1, 10 remain constant while the flow rate decreases.
(b) In a construction dewatering system, the
where e is the void ratlo, D,, is the 10% particle
size in mm, and y, is the unit weight of water in

. . . . . . . .
. .
. . . .
. . . .
. . . . . . . . .
Circle Rectangle

. . . . . . .
. .
. .

. .
. .
. . . . . .
Open circle Open rectangle

(a)

. . . . . . . . Initial excess
. . . . . . . .
pore pressure
. . . . . . . .
6
Single line Parallel lines
(b)

. . . Horizontal distance from smk


. . .
. . . . .. . . . Boundary
. . . conditions
. .. . . . . . . At face of the ank: ir = 0
. . .
. . . At outer boundary: U = &
At outer boundary: air/&x = 0
Top hat Double top hat

(c) (b)

Fig. 2. Simplified layouts of construction dewatering Fig. 3. Flow through a thin layer to a pumped sink: (a)
systems: (a) ring systems; (b) linear systems; (c) other pumped sink and soil layer; (b) distribution of excess
systems pore pressure at time t (t > 0)
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 85

drawdown results from the action of a group trolled by the soil permeability and compress-
of wells, usually in a line or a ring. The most ibility. If flow is predominantly horizontal, the
common idealized cases for analysis are the equipotentials are vertical and the excess head
drawdowns outside a ring of wells, inside a within the consolidating soil will vary with dis-
ring of wells, alongside a line of wells, and tance from the sink, but will be independent of
between two parallel lines of wells (Fig. 2). depth. Fig. 4 shows isochrones representing the
The Theis solution and its variants can model variation of excess pore pressure with distance
only the first of these cases, with the ring of from the sink at various times t, < t, < t, after
wells idealized as an equivalent single well. the start of pumping. For application to construc-
tion dewatering systems, it is useful to convert the
Other solutions are available for the geometries
excess pore pressure into drawdown of total head
outlined above, and for drainage conditions cor-
(Fig. 5).
responding to a constant drawdown in the well.
In plane flow to a pumped slot, the assumption
Appendix 1 summarizes the analytical solutions
that the isochrones are parabolic, following Scho-
for plane flow to a linear slot (Carslaw & Jaegar,
field & Wroth (1968), was used by Powrie &
1948), radial inward flow to a well (Richardt,
Roberts (1990) to obtain an expression for the
1957; Jacob & Lohman, 1952), and radial
variation of drawdown 6 with horizontal distance
outward flow to a ring slot (Carslaw & Jaegar,
x from the pumped slot at a time t after the start
1948). The principal drawback of these solutions,
of pumping
which are generally in the form of Fourier series
for plane flow or Bessel functions for radial flow,
is that the distribution of drawdown is difficult to M&$+1) O<x<L (2)
evaluate. A simpler approximate solution for
horizontal flow to construction dewatering where h, is the drawdown at the face of the slot
systems idealized as pumped rings or slots, in and L is the distance from the slot to the furthest
which the isochrones of excess pore water pres- point at which drawdown occurs. The distance of
sure are assumed to be parabolic, is now devel- influence L is a function of time and can be
oped. expressed as

L = J( 12c,, t) (3)
SOLUTION BY PARABOLIC ISOCHRONES
Consider the stratum of soil shown in Fig. 3. If, where cm = k, &‘/y, is the coefficient of vertical
at time t = 0, the pore pressure at one place in consolidation due to horizontal flow, k, is the
the layer is reduced by pumping from a well or a horizontal permeability of the soil, Ev’ is the
slot, and maintained at a constant level, the layer modulus of one-dimensional vertical compres-
will begin to consolidate as water flows towards sion, and y, is the unit weight of water. A para-
the sink. Pore pressures can be considered as bolic isochrone solution can also be developed for
excess relative to the reduced piezometric level at the drawdown in a soil layer between two parallel
the face of the sink, and will dissipate as the soil slots (Appendix 2).
skeleton consolidates and the effective stresses A similar solution can be developed to the
increase. Consolidation cannot occur instantane- problem of consolidation due to inward radial
ously, because the rate at which water can flow flow to a fully penetrating well in which the draw-
through the soil is controlled by the permeability. down is constant, although the suitability of the
Excess pore pressures will dissipate at a rate con- parabolic isochrones assumption is less certain

LL
Horizontal distance from snk

Fig. 5. Relation of draw-down and excess pore pressure:


Horizontal dstance fromsmk h = iW_- (rS/yJ, where h, is drawdown at the face of the
sink, h is drawdown at distance x from the sink, U is
Fig. 4. Isochrones representing the distribution of excess excess pore pressure at distance x from the sink, and y,
pore pressure is unit weight of water
86 POWRIE AND PREENE

r Parabolic isochrone model

/
I I
0
, , , I I O-01 0.1 1
0 ,

IO' 10' 103 104 105 106 Time factorJ,


cH"tIx2
Fig. 6. Plane flow: comparison of parabolic isochrone Fig. 8. Outward radial flow to a ring slot: comparison of
and analytical solutions parabolic isochrone and analytical solutions

and the algebra is more complex than for plane solutions agree very closely over the whole time
flow. Powrie & Roberts (1990) derived an expres- range.
sion for the drawdown h at a radius r from the Figure 7 compares the flowrates according to
centre of the well the parabolic isochrone solution for inward radial
flow with the analytical solution of Jacob &
i Lohman (1952) (Appendix 1). The time since the
(r* - 2Rr + R’)
’- =w(R - r,)* start of pumping is expressed in terms of a non-
dimensional time factor T,
rW < r < R (4)
T, = cHv tjr,*
where &, is the drawdown at the face of the well,
rw is the radius of the well and R is the longest Fig. 7 shows that for small time factors, when
distance from the centre of the well at which flow is essentially plane towards the face of the
drawdown occurs (i.e. the distance of influence). well, there is reasonable agreement between the
An explicit expression for R is derived in Appen- parabolic isochrone and analytical solutions, but
dix 3, while Appendix 4 details the parabolic iso- thereafter the solutions diverge. For T, > 0.1,
throne analysis for outward radial flow to a ring radial effects become more significant, and the
slot. Expressions for drawdown and distance of shape of the drawdown curve becomes different
influence for the parabolic isochrone solutions are from the parabola assumed in the analysis. In
given in Appendix 2. practice in fine soils (lo-’ m/s < k < 10m5 m/s),
Figure 6 compares the drawdowns calculated the divergence of the parabolic isochrone and the
by using the plane flow parabolic isochrone solu- analytical solutions may occur after a period of a
tion (equations (2) and (3)) with those from the few minutes to a few weeks, depending on the size
exact analytical solution given in Appendix 1: the of the dewatering system. Unfortunately, the true
shape of the isochrones cannot be determined
from the Jacob & Lohman solution: however, a
numerical solution to equation (9) is given by
103/- Rao (1973).
Figure 8 compares the drawdown in the centre
of a ring slot calculated from the parabolic iso-
throne (equations (18)-(20)) and analytical
(Appendix 1) solutions. At T, = 0.1, the parabolic
isochrone solution gives a drawdown -50%
greater than the analytical solution, but for T, >
0.3 the two methods agree to within 20%. In this
case, since the drawdown under consideration is
actually within the area to be excavated, the
higher values of T, are likely to be of greater prac-
, I I I J tical importance.
10-21
10-d 10-Z 1 102 104 106
Time factorJr
BACK-ANALYSIS OF FIELD DATA
Fig. 7. Inward radial flow to a well: comparison of para- The applicability of the parabolic isochrone
bolic isocbrone and analytical solutions solutions is now examined with reference to the
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 87

Table 1. Summary of case studies*

Case Pore- Penetration Locations where Piezometer


study pressure of aquifer drawdowns were type
control depth: % monitored
method

WPl VWP 30 m line of wellpoints 100 Alongside the 19 mm dia.


line of wellpoints standpipe
WP3 VWP 12 m x 12 m square ring <35t Inside and 50 mm dia.
of wellpoints outside the ring standpipe
(rr = 7.6 m) of wellpoints
WP7 WP 9 m line of wellpoints 1.0 50 Alongside the 38 mm dia.
line of wellpoints standpipe
EJ12 EJ 135 m line of ejector wells 15 100 Alongside the Pneumatic
(rlr = 0.1 m for an individual well) line of piezometer
ejector wells
DW3 DW Well pumping tests 100 Around the well 19 mm dia.
(rx = 0.125 m) standpipe

* VWP = vacuum wellpoints; WP = conventional wellpoints; EJ = ejector wells; DW = deep wells; ru, = radius of
equivalent well used in radial flow analyses.
t Boreholes did not prove aquifer thickness.

Table 2. Parameters used in back-analysis of case studies


Permeability k

Case study Permeability selected Back-calculated


according to permeability* : m/s
Appendix 5: m/s

WPl 2.3 x 1O-5 3.2 x 1O-5


WP3 4.1 x 1o-6 1.9 x 1om5
WP7 6.4 x lo-’ 1.6 x 1O-5
EJ12 1.8 x lo-6t 2.5 x 10-G
DW3 7.0 x 1o-6 6.9 x 10-6t

Modulus of one-dimensional compression E’

Case study Mean SPT E’: kPa


blowcount N

WPl 62 31 x 1032
WP3 46 23 x 103$
WP7 45 23 x 103$
EJ12 9.4 x 1035
DW3 1~11 50 x 103$

Drawdown at the well or slot

Case study Drawdown:


m

WPl 3.0
WP3 3.9
WP7 3.0
EJ12 -7
DW3 2.2

* Back-calculated permeability was estimated from the


steady-state performance of the dewatering system in each
case.
t Estimated from well pumping tests (after Cooper & Jacob,
1946).
~91;mated from E’ = 5OON, where E’ is in kPa (from Parry,

5 Estimated from oedometer consolidation tests.


1)N estimated from SPTs terminated at 50 blows for less than
300 mm penetration.
88 POWRIE AND PREENE

Pore pressures I” the aquifer


(Head, 1980) over an appropriate stress range.
FlkTll Reduction However, it can be difficult to recover suitable
samples, particularly if the soil has a layered
structure, and E,’ must often be estimated from

1
in situ tests. Parry (1971) has suggested that E.,’
can be estimated approximately from standard
penetration tests (SPTs) by use of the relation
E,’ = 0.5N (6)
where E,’ is in MPa and N is the SPT blowcount
for a 300 mm penetration. It is now well known
that the apparent stiffness of a soil depends on
the shear strain following a change in the direc-
tion of the strain path. Correlations between E,’
and N presented by Stroud (1989) demonstrate
this effect quite clearly, and suggest that at the
strain levels which might be expected to result
from the operation of dewatering systems, values
of E,‘IN in the range 0.552 MPa could be appro-
priate. In the absence of more rigorous data,
equation (6) has been used to estimate values of
E,’ at the low end of the likely range. This
approach is conservative in that it will tend to
(b)
overpredict both consolidation times and settle-
Fig. 9. Drawdown at the face of a well or slot: well face ments. In two of the case studies, the calculations
(a) is, (b) is not, an eqoipotential; pore-pressure redaction have also been carried out using E,‘jN = 1 MPa
(a) is, (b) is not, constant with depth; water level in well and E,‘/N = 2 MPa in order to investigate the
(a) is not, (b) is, drawn down below the top of the aquifer influence of the assumed soil stiffness.
Equations (2)-(5) and (12)-(20) show that the
drawdown in the surrounding_soil depends on the
drawdowns monitored during the transient phase magnitude of the drawdown h, at the face of the
around a number of construction dewatering sink (well or slot). The parabolic isochrone solu-
systems in fine soils. The excavation geometries tions assume that the face of the well is an equi-
and permeability data for each of the case studies potential, and hence that the drawdown here is
are summarized in Table 1. In order to calculate independent of depth (Fig.. 9(a)). However, in
the rates of change of drawdown with time, it is wells drained to just above their base, the bound-
necessary to estimate the horizontal permeability ary between the well and the soil is not an equi-
k, and the modulus of one-dimensional compres- potential, and the drawdown along this interface
sion E,’ of the soil, and the effective drawdown varies with depth (Fig. 9(b)). The distribution of
h, at the face of the equivalent well or slot. The drawdown at the face of the well is difficult to
parameter values used in the analyses are given in predict precisely. It depends on the vacuum devel-
Table 2: these were selected using the procedures oped inside the well by the pumping equipment,
discussed below. the head losses at the soil-well interface, and the
Typically in fine soils, the permeability data surface tension forces that may act at the soil-
available at the design stage may range over well interface to resist the entry of water into the
several orders of magnitude. In general, the well. The values of h, used in the back-analysis of
analyses presented in this Paper show drawdowns the field data are given in Table 2: except where
calculated by use of both the permeability back- otherwise stated, these values were estimated
from the variation of drawdown with distance
calculated from the steady-state performance of
under steady-state conditions, as indicated by
the dewatering system and the permeability esti-
nearby piezometers.
mated according to the guidelines given in
The case studies were in three main categories,
Appendix 5. These guidelines are considered to
give the optimum permeability from data avail- according to the nature of the idealized flow
able at the design stage, and are discussed in (a) plane flow to lines of wells, modelled as equiv-
detail by Preene (1992). alent slots (cases WPl, WP7, EJ12)
The modulus of one-dimensional compression (b) radial flow to a ring of wells, modelled as an
Ev’ can in principle be determined from labor- equivalent well (case WP3)
atory tests by use of an oedometer or a Rowe cell (c) radial flow to an individual well (case DW3).
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 89

30 m line of

= Initial
piezometnc
level

-% -- )L
(a)

9 m line of wellpoints
v lnltlal
piezometric
level

.
Observation well
,- ..
4- .,
I. ., ;’
5- ‘_7 --;-x _x _ x _y
--Y-
>- x Silty clay -S _ x --r -*-
(b)

Well Piezometer

-. -x
F/ llitial
plezometric
level
J
: .,

_ --Y-
-~--Y--r-x-_x_-x_x
Cc)

2 Bedrock

(e)

Fig. 10. Geometry of case studies: (a) WPl; (b) WW; (c) DW3; (d) WP3;
(e) EJ12
90 POWRIE AND PREENE

studies WPl, WP7 and EJ12 respectively.


Recorded and calculated drawdowns and flow
rates are compared in Figs 14-16 for a ring of
D Recorded drawdown wellpoints (case study WP3), and in Figs 17 and
---- k = 3-2 x 10 5 m/s
- k = 2.3 x 10 ‘m/s
18 for an individual well (case study DW3).

Parabolic isochrone model COMPARISON OF FIELD DATA WITH


PARABOLIC ISOCHRONE ANALYSIS
Close agreement between the plane flow para-
bolic isochrones model and the analytical solu-
tion is shown above (Fig. 6). Using a drawdown
at the wellpoints estimated from the performance
of the pump, Fig. 11 shows that there is also close
agreement between the recorded drawdowns and
0

I
those calculated using the plane flow parabolic
isochrones model for case study WPl. In this case
study, the line of wellpoints penetrated to the full
depth of the aquifer, and nowhere did the pore-
water pressure fall below the air entry value u,:
thus the conditions in the field closely matched
the assumptions made in the parabolic isochrones
Parabolfc wxhrone model model.
/ The recorded and calculated drawdowns in
\
0 q o case study WP7 (Fig. 12) agree less closely. This is
-----__________ probably because the pore-water pressures in the
I I I I
0
I upper part of the aquifer were reduced to the air
3O 10 20 30 40 50 entry value u,, allowing local desaturation to
Time: h
occur. During the early stages of pumping, before
(b)
significant desaturation occurred, pore-pressure
Fig. 11. Case study WPl: comparison of calculated and reduction was achieved quite rapidly and the cal-
recorded drawdown: (a) 11 m from the line of wellpoints; culated and recorded drawdowns are reasonably
(b) 15 m from the line of wellpoints close. Once air begins to enter at the surface of
the aquifer, the water released from the desatu-
rating upper pores can percolate downwards and
Although case study EJ12 is categorized as type slow the rate of drawdown: under these condi-
(a), some drawdown data are available from a tions, the parabolic isochrones model over-
piezometer very close to one of the wells, where predicts the drawdown. The steady-state data
flow may be of type (c). The geometry of each indicated an effective drawdown in the soil to the
case study is shown in Fig. 10. top of the wellpoint screens; no seepage face
Drawdowns were measured by use of either losses were apparent.
pneumatic piezometers or 19-50 mm dia. stand- Figure 12(c) and (d) shows that in this case, an
pipe piezometers. Under transient conditions, the increase in the assumed value of Ev’ by a factor
pore-water pressure recorded in a piezometer will of four (from 23 MPa to 96 MPa) increases the
lag behind the pore-water pressure in the soil due drawdowns calculated at a distance of 3.5 m from
to the response time of the piezometer. This is the line of wellpoints only slightly. The validity of
related to the volume of water that must flow into the calculated drawdowns in the vicinity of the
or out of the piezometer for the change in pore wellpoint line is perhaps more likely to be
pressure to register, and the rate at which this can affected by the permeability used in the calcu-
occur. For each case study, the equalization times lation, which could easily be in error by an order
were calculated according to the methods of of magnitude or more due to the limitations of
Hvorslev (1951) for relatively stiff soils, and Brand permeability measurement in fine soils (e.g.
& Premchitt (1982) for more compressible soils: Preene & Powrie, 1993).
in no case were the recorded drawdowns signifi- Case studies WPl and WP7 involved lines of
cantly affected by the response times of the wellpoints at very close spacings (l-3 m centres).
piezometers used. In case study EJ12, the wells (which were pumped
Figures 11-13 compare the recorded draw- by ejectors) were much more widely spaced (15 m
downs with those calculated by use of the para- centres). Fig. 13(a) shows that in this case the
bolic isochrones model for plane flow for case plane flow parabolic isochrones model, using a
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 91

0 Recorded drawdown cl Recorded drawdown


---- k = 6.4 x 10-s m/s ---- k = 6.4 x 10 ‘m/s
- k= 1.6 x 10 ‘m/s - k=1.6xlO ‘m/s

Parabolic isochrone model

Parabok wchrone model

I I I I I
(b)

0 Recorded drawdown Cl Recorded drawdown


---- E,’ = 23 MPa ---- Ev’ = 23 MPa
- Ev’ = 46 MPa - Ev’ = 46 MPa
--- Ev’ = 92 MPa --- Ev’ = 92 MPa

I I 1 k I I I I I 1 I 1
3O 10 20 30 40 50 0 10 20 30 40 50
Time: men Time: men
(c) Cd)

Fig. 12. Case study WP7, comparison of calculated and recorded drawdown: (a) 35 m from the line of wellpoints; (b)
92 m from the line of wellpoints; (c) sensitivity of calculated drawdowns to the soil stiffness A?$ 35 m from the line of
wellpoints with k = l-6 x 10-s m/s; (d) sensitivity of calculated drawdowns to the soil stiffuess E,’ 35 m from the line
of the wellpoints with k = 64 x 10-s m/s

drawdown at the well estimated from steady-state purpose of analysis was modelled as a single
piezometer data, gives results similar to the equivalent well. The radius of the equivalent cir-
recorded drawdowns at a distance in excess of cular well was taken as 7.6 m, so as to give the
approximately one well spacing from the ejector same perimeter as the actual wellpoint system.
line. The ability of the parabolic isochrones After 15 min of pumping (Fig. 14(a)), the radial
model to predict drawdowns nearer the line of parabolic isochrones slightly underestimate the
wells is discussed below. actual drawdowns, but after 61 min (Fig. 14(b))
It is shown above that the parabolic isochrones the agreement is very close. At later times (Fig.
model and the analytical solution for radial flow 14(c), after 166 h), the drawdowns are signifi-
begin to diverge at a time factor T, = 0.1. In a soil cantly overestimated by the radial parabolic iso-
of given permeability and compressibility, the thrones. This is confirmed by Fig. 15, which
time t,., after the start of pumping at which T, = compares the timeedrawdown responses recorded
0.1 will depend on the well radius r,. For an by the piezometers at various distances from the
individual well, rw x 0.1 m, and t,., may be only wellpoint ring with those calculated using the
a few seconds. For rings of wellpoints modelled as radial parabolic isochrones approximation.
large equivalent single wells, t,., may be several During the first few minutes of pumping there is
days or weeks. Table 3 gives values of t,., at reasonable agreement, but at later times the para-
which the parabolic isochrones and the analytical bolic isochrones model significantly overestimates
solutions for radial flow begin to diverge, for soils the drawdown. The best fit to the data was
having lo-’ m/s < k < 10e5 m/s and 10 MPa < obtained using a drawdown at the wellpoints of
E,’ < 50 MPa, with 0.15 m < rw < 95.5 m. 3.9 m, which would imply the development of a
The dewatering system in case study WP3 was seepage face 0.6 m above the top of the wellpoint
a 12 m square ring of wellpoints, which for the screen.
92 POWRIE AND PREENE

The variation in flow rate with time calculated


by use of the radial parabolic isochrone model is
Parabolic isochrone model compared with the exact analytical solution in
Fig. 7, which shows that for T, > 0.1 the flow rate
is greater for the exact solution than for the radial
Recorded drawdown
parabolic isochrone model: this indicates that the
parabolic isochrones underestimate the hydraulic
gradient (which is equal to l/y, multiplied by the
slope of the isochrone) at the entry into the well.
Also, the integrated flow rate is greater for the
exact solution than for the radial parabolic iso-
throne model: this implies that the latter under-
estimates the volume of soil subjected to
pore-pressure reduction. These observations are
confirmed by Fig. 19, which compares radial
parabolic isochrones at T, = 10, 100 and 1000
with the numerical solutions to equation (a) pre-
sented by Rao (1973). The ‘exact’ isochrones are
steeper near the well, but have a greater distance
of influence; the parabolic isochrone solution will
therefore tend to overestimate drawdowns near
the line of wells, and underestimate drawdowns
further away. At T, = 10, the discrepancy between
Linear, iiw = 7 m
the two methods is not excessive, so T, = 10
8O
I 5, 10
I 15
, 20 25
,
might represent the limit of applicability of the
Time: days parabolic isochrone approximation, rather than
W T, = 0.1, at which the two solutions begin to
Fig. 13. Case study JZJ12: comparison of calculated and
diverge. In this case, the values of elapsed time
recorded drawdown: (a) 18 m from the pumped line of
(given in Table 3) at which the radial parabolic
ejector wells; (b) close to an individual well

Table 3. Period of pumping after which radial flow parabolic isocbrone and
exact solutions begin to diverge (T, = Wl)*

rw: m k: m/s

10-s 10-G 10-7

Relatively stifl soil (E’ = 50 x 10s kPa)

Single well 0.15 0.05 s 0.5 s 5s


Well group 3.18 20 s 200s 34 mitt
(5 m x 5 m)
Well group 6.37 80 s 13.5 min 2.3 h
(10 m x 10 m)
Well group 31.8 34 s 5.6 h 2.3 days
(50 m x 50 m)
Well group 95.5 5.1 h 2.1 days 3 weeks
(150 m x 150 m)

Relatively soji soil (E’ = 10 x lo3 kPa)

Single well 0.15 0.2 s 2.3 s 23 s


Well group 3.18 100s 16-9 min 2.8 h
(5 m x 5 m)
Well group 6.37 6.8 min 1.1 h 11.3 h
(10 m x 10 m)
Well group 31.8 2.8 h 28.2 h 11.7 days
(50 m x 50 m)
Well group 95-5 253 h 10.6 days 15.1 weeks
(150 m x 150 m)

* For well groups rw is based on equivalent perimeter.


CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 93

o-
0

r 0 Recorded drawdown
1 - k = 1.9 x IO ‘m/s
I-
---- k = 4.1 x 10 6 m/s
E t
E
t* Recorded drawdown
-k=l~9x105mis
B ‘\\
2 3- \ ---- k = 4.1 x IO ‘m/s
\
6 \

0
Parabolic isochrones
4-
1A PositIon of wellpolnts
51 ’ 1 I I 1 I I
-5 0 5 IO 15 20 25

1 I I I I I 1
4’
0 5 IO 15 20 25 30 35
5 ’ ’ 1 I I I
-5 0 Time: h
10 20 30 40
(b) (b)
Fig. 15. Case study WP3: comparison of calculated and
recorded drawdown: (a) 2 m from the ring of wellpoints;
(b) 8 m from the ring of wellpoints

“r 0 Recorded flow rate

r_]k=4.1 x 10 ‘m/s
6
Z~j-k=1-9xlO 5m/s
2 t

PosItIon of wellpoints

51 ’ ’ I I I I 1
-5 0 IO 20 30 40 50
Distance: m
w
Fig. 14. Case study WP3: isochrones of drawdown: (a) Time: h
after 15 min of pumping; (b) after 61 min of pumping: Fig. 16. Case study WP3: comparison of calculated and
(c) after 166 h of pumping recorded flow rate

isochrones solution ceases to be valid could be overestimate the drawdowns significantly. This is
multiplied by 100. consistent with the expected divergence between
For case study WP3, T, = 0.1 after -3 min, so the exact and parabolic isochronic solutions
after this time the exact solution and the radial shown in Fig. 19. For piezometers within about
parabolic isochrones approximation would be 50 m of the ring wellpoints (r/r, < 8) and for
expected to diverge. In reality, however, the dis- T, > 100 (t > -5 h) the parabolic isochrone solu-
crepancy between the approximate solution and tion gives larger drawdowns than the numerical
the data during the early stages of pumping is solution. Drawdowns may also have been affected
small (Figs 14 and 15), with the radial parabolic by the release of pore water from the upper part
isochrones slightly underestimating the recorded of the aquifer when the pore pressure falls to the
drawdowns. After a few hours of pumping, the air entry value U, Fig. 16, which indicates record-
radial parabolic isochrones solution begins to ed flow rates significantly higher than those cal-
94 POWRIE AND PREENE

stiffness E,’ = 50 MPa (0.5 N), 100 MPa (1 N),


Recorded drawdowns
200 MPa (2 N) and 800 MPa (8 N). The draw-
downs according to the true analytical solution
for a single well with constant pumped flow rate 4

i = (q/47rkD)[ -0.5772 - In (r2/4cHV t)] (7)


are also shown (Cooper & Jacob, 1946) where h
is the drawdown at a radial distance r from the
well and D is the effective aquifer thickness.
Although this formula is for a well pumped at a
constant flow rate, it can be used without signifi-
Ev’ = 200 MPa
J
cant error for a well in which the drawdown is
(a) constant when T, > lo4 (Jacob & Lohman, 1952).
For case study, DW3, T, = lo4 after - 1.25 h, so
Parabolic tsochrone model
the use of equation (7) is not unreasonable.
For case study DW3 (for a given value of E,‘)
the radial parabolic isochrones model gives sig-
nificantly smaller drawdowns than the Cooper-
200 MPa Jacob solution. This is consistent with the
divergence between the radial parabolic iso-
thrones approximation and the true analytical
800 MPa
solution that occurs at larger values of T,, which
results in the underestimation of drawdowns at
distance from the well, as discussed above and
shown in Fig. 19. (For case study DW3 the
800 MPa piezometer at 12 m has r/r, = 96 and that at 29
m has r/rw = 232.) At a distance of 12 m from the
0 20 40 60 80 well (Fig. 17(a)), both the Cooper-Jacob analysis
Time: h with E,’ = 50 MPa and the parabolic isochrones
(b)
method with E,’ = 200 MPa give a reasonable fit
Fig. 17. Case study DW3: comparison of calculated and to the observed data. At a distance of 29 m from
recorded drawdown: (a) 12 m from the well; (h) 29 m the well (Fig. 17(b)), a reasonable fit between the
from the well CooperrJacob solution and the measured data is
obtained with E,’ = 200 MPa; the parabolic iso-
thrones analysis requires E,’ = 800 MPa
(E,‘/N = 8 MPa), and then gives less close agree-
culated using either analytical method, is ment. The implied four-fold increase in soil
consistent with the release of pore water due to stiffness between the two monitoring points is
partial desaturation, in addition to the consoli- plausible in neither case. The drawdowns (and
dation mechanism assumed in the analyses. hence the changes in effective stress) are of the
The radial parabolic isochrones model was also
used in the analysis of case study DW3. This was
a pumping test on a single well, in conditions that
matched the assumptions of both the approx-
imate and exact analyses in that the well pen-
etrated the full depth of the aquifer and the pore
pressures did not fall below the air entry value u, 0
00 000 0
The drawdown inside the well was - 18 m, but a
Jacob & Lohman (1952)
back-analysis of the piezometer drawdowns using /(k~=7xlo 5 m/s. h, = 2 2 m)
the methods of Cooper & Jacob (1946) indicates a -----i_-________________

drawdown in the soil at the face of the well of


only 2-3 m, implying combined losses and Parabolic lsochrone model
seepage face effects of - 15 m. Fig. 17 compares O-05 (k = 7 x 10 ’ m/s, !I, = 2-2 m)
the drawdowns measured in piezometers 12 m
(Fig. 17(a)) and 29 m (Fig 17(b)) from the well
(r, = 0.125 m) with those calculated by use of the Tome- h

radial parabolic isochrones model, using a draw- Fig. 18. Case study DW3: comparison of calculated and
down at the well of 2.2 m and values of soil recorded flow rate
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 95

same order of magnitude, so that the effect of a T, = 10 T, = 100 T, = 1000


high soil stiffness at small strains would be
expected to be similar at each location. Further-
more, the value of E,’ = 800 MPa required to
match the parabolic isochrones solution to the
measured drawdowns at a distance of 29 m from
the well is probably outside the range that could ---- Numerfcal solution (F&O, IWS)
reasonably be inferred from the site investigation - Parabohc ,sochrone

data: this would suggest strongly that the radial


parabolic isochrones approximation is inapprop-
riate in this case. Fig. 19. Discrepancy between parabolic radial isochrones
Figure 18 compares the recorded flow rate (as a and numerical solutions of Rao (1973)
function of time) for case study DW3 with flow
rates calculated using the radial parabolic iso-
thrones approximation and the Jacob-Lohmam
analytical solution (equation (7)), with E,’ = 50 outside represents a seepage face where water is
MPa. The parabolic isochrones approximation entering the well, well losses due to the probable
underestimates the flow rate by an order of mag- local reduction of soil permeability following well
nitude. The Jacob-Lohman solution gives a installation, and surface tension effects which
closer estimate, but still underestimates the might tend to resist the exit of water from the soil
recorded tlow rate by a factor of -2. The calcu- pores (Hall, 1955). Any analytical solution based
lated flow rate decreases with increasing soil on the outward progression of drawdown away
stiffness for both analyses, which suggests that from a well will be of limited use for individual
neither method is capable of modelling all aspects wells, since at present the drawdown required to
of the observed behaviour. The discrepancy overcome well loss effects cannot be estimated to
between the JacobLohman solution and the field better than within an order of magnitude. This is
data might be due in part to the estimates of the in contrast to the data from case studies involving
soil permeability and the drawdown at the well. lines of closely spaced wells (WPl, WP3 and
These would have a similar effect on the radial WP7), where the well losses were sufficiently
parabolic isochrones solution, but the main small (O-l m) not to affect significantly the accu-
problem with the approximate method is that it racy of the drawdown calculations.
underestimates the hydraulic gradient in the
vicinity of the well. It is not surprising that if the
parabolic isochrone is approximately correct for CONCLUSIONS
plane flow, it will not be appropriate in regions The principal mechanism of pore-water pres-
where there is significant convergence of a radial sure reduction in fine soils is often consolidation.
flow pattern in the circumferential direction. The time taken for a construction dewatering
A further limitation in the application of any system to achieve a given drawdown may be an
analytical model to an individual well in a fine important factor in programming the progress of
soil is illustrated by case study EJ12. Fig. 13(a) a construction project, and should be considered
shows that the plane flow parabolic isochrones at the design stage. This Paper gives approximate
model, using drawdown at the well estimated solutions for vertical consolidation due to the
from steady-state piezometer data, gave close esti- plane and radial horizontal flow to dewatering
mates of drawdown at distances of one well systems comprising lines and rings of closely
spacing or more from the line of wells. The draw- spaced wells. These solutions are based on the
down in the vicinity of an individual well is now assumption that the isochrones representing the
investigated by use of the radial parabolic iso- variation of drawdown with distance from the
thrones model. Fig. 13(b) shows the drawdown line or ring of wells at any given time are para-
(as a function of time) recorded in a pneumatic bolic.
piezometer 2 m from an individual ejector well. The approximate methods are compared with
The ejector developed a vacuum of -45 kPa corresponding exact analytical solutions where
inside the well. Assuming that this is equivalent to these exist, and with field data from a number of
an additional lowering of the water level in the sites. In plane flow to lines of wells, the parabolic
well of 4.5 m, the total drawdown inside the well isochrone approximation matches both the exact
was 28 m. However, data from surrounding solution and the field data.
piezometers indicate a drawdown immediately For inward radial flow to rings of wells, there is
adjacent to the well of only 7-8 m after 20 days. initially close agreement between the flow rates
The difference of -20 m between the drawdown calculated by use of the exact and approximate
in the well and that in the soil immediately methods, but at a time factor T, ( =cHv t/r,‘) of
96 POWRIE AND PREENE

0.1-10, significant divergence begins to occur. support, to WJ Engineering Resources Ltd for the
This may not in practice be a serious short- facilities used in the collection of the field data,
coming: the elapsed time at which T, = 0.1 will and to L. J. Robinson for his comments on the
depend on the consolidation coefficient of the soil mathematical derivations presented in Appen-
cHv and the radius of the equivalent well r,,,, and dices 3 and 4.
may vary from a few seconds for an individual
well to several weeks for a large excavation.
Theoretical considerations aside, it is important APPENDIX 1. ANALYTICAL SOLUTIONS
Planeflow to an infinitely long fully penetrating slot
that the boundary conditions in the field should Drawdown 6 at distance x from the face of the slot
be reasonably close to those assumed in analysis. can be expressed as
Significant local desaturation of the aquifer may
result in discrepancies between theory and per- I;lh, = 1 - erf [x/2d(c,v t)]
formance. Close sources of recharge, which have (Carslaw & Jaegar, 1948) where 6 is the drawdown at
not been mentioned explicitly in this Paper, the face of the slot, t is the elasped time since start of
would almost certainly have a deleterious effect. pumping, cHv is the coefficient of vertical consolidation
The soil parameters used in the analysis (i.e. the due to horizontal flow and erf (0 is the error function,
horizontal permeability k, and the vertical values of which are given by Carslaw & Jaegar.
stiffness E,‘) will have a significant effect on the
results. In the cases back-analysed in this Paper,
Inward radialjow to a fully penetrating circular well
the selection of soil permeability according to the
Drawdown h at radius r from the centre of the well
methodology given in Appendix 5, together with
can be expressed as
the estimation of soil stiffness from oedometer
tests carried out over an appropriate stress range 1: m
or even empirically from the SPT blowcount,
5-2 1
w “=I
gave calculated drawdowns that were generally
close to those measured. In cases of uncertainty,
however, it is recommended that a sensitivity
study be carried out over the probable range of
(9)
k, and E,‘.
The case study data indicate that, for lines of (Richardt, 1957) where 5 is the drawdown at the face of
closely spaced wells, losses and seepage face the well and
effects are small (O-l m) and do not significantly
u@,) = J&G&,) - Y,@,)J,M (104
affect calculations. However, where wells are
widely spaced and radial flow components to u,@, N) = J&G NY,@,) ~ Y,@, NJ&,) (lob)
individual wells are significant, losses at the soil-
U&, r/r,) = J&.r/r,)Y&.) - Y&G r/r,)J&,) (1@4
well interface may be large, perhaps of the order
of 10 m or more. At present there is no reliable J,, J, are Bessel functions of the first kind and zero and
way of estimating these losses. This tends to mili- first order respectively, Y, , Y 1 are Bessel functions of
tate against the use of methods based on effective the second kind and zero and first order respectively, a,,
drawdowns at individual wells for the analysis of a:2, ... *:n> are the roots of the equation
radial flow. J,(m, NY&,) - Yl@. MJ,(a,) = 0 (11)
The principal advantage of the parabolic iso-
throne models is the comparative ease with which N = RI/r,, R’ is the radius to the outer aquifer bound-
ary which is assumed to be impermeable, T, is the radial
the distribution of drawdown at any given time
time factor (see equation (5)) and other terms are as pre-
can be calculated. However, there are limitations, viously defined.
both theoretical (T, < 0.1-10 for inward radial Flow rate 4 from the well can be expressed as
flow) and practical (no close sources of recharge
or significant desaturation of the aquifer). Provid- 4 = 27&D& G( T,) (12)
ed that their limitations are recognized and the (Jacob & Lohman, 1952) where G(T,) is a well function
approximate solutions are not misapplied, the tabulated by Jacob & Lohman, k is the soil per-
results should usually be sufliciently accurate for meability, D is the aquifer thickness and the other terms
design and programming purposes for dewatering are as previously defined.
systems comprising lines and rings of wells in fine
soils.
Outward radialPow to a ring slot
Drawdown h at radius r from the centre of the ring
can be expressed as
ACKNOWLEDGEMENTS
The Authors are grateful to the Science and J&B./r,)
(13)
Engineering Research Council for financial (&/r,)J,(/?.) exp (- ‘&*)
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 97

(Carslaw & Jaegar, 1948) where h, is the drawdown at where


the inner face of the ring slot, /I,, pz.. fi,, are the
positive roots of J,(b) = 0 and other terms are as pre- 12cnv t (R - r.J3 (R - rJ2
-_=++ (22)
viously defined. 2 3r,’ 2
r, rW

To make this solution more useful, an explicit expres-


APPENDIX 2. SUMMARY OF DRAWDOWN sion for the radius of influence R must be obtained. This
FORMULAE FOR PARABOLIC ISOCHRONE can be achieved by considering equation (22) as a cubic
SOLUTIONS in (R - r,)/r,. Let a = (R - r,)/r,, and substitute into
Plane/low to a slot equation (22). Then
For t > 0, equations (2) and (3) apply.
36c,v t
a3 + 3a2 - 2 = 0 (23)
Planeflow to two parallel slots a distance b apart r,
For 0 i t < b2/48c,,, equations (2) and (3) apply. which if expressed in the form
For t > bz/48c,,
a3 + Aa’ + Ba + C = 0 (24)
where A = 3, B = 0, C = - 36c,, t/r,,,‘, can be solved
by substituting the identities
(14) b = a + A/3 (2W

Znward radialjow to a well


(25’3
For t > 0, equation (4) applies.
For 0 < t < rW2/9c,v
AB 2A’
RJr, = 2 cos (Q/3) 2q=T-11-C
(15)
where cos 0 = 18T, - 1. and solving /I” - 3pp - 2q = 0, to give, for 0 ,< t ,<
For t > r,‘/9cHv
rwz/9c,,, equation (15), where cos 0 = 18T, - 1 and
R/r, = exp (o/3) + exp (~ 8/3) equation (5) applies. For t > r,,,‘/9c,,
(16)
R/r, = 2 cash (O/3) (26)
where
where cash 0 = 18T, - 1, which can be expressed in
Q = In ((18T, - 1) + J[(18T, - 1)’ 11) (17)
terms of exponentials as equation (16) (where equation
and equation (5) applies. (17) applies). From these equations, R can be substi-
tuted into equation (21) to calculate drawdown at
radius r from the well axis.
Outward radialjow to a ring slot of radius rW
For 0 < t < rWz/18c,v and (rW - L) < r < r,

i = k {C - [(2Lr, - r,*) + 2(r, - L)r - r’]} (18) APPENDIX 4. DERIVATION OF PARABOLIC


ISOCHRONE ANALYSIS FOR OUTWARD RADIAL
FLOW TO A RING SLOT
L/r, = 1 + J3 sin (fJ/3) - cos (e/3) (19)
Consider a radial section through a thin layer of soil
where cos 0 = 1 - 18T,. of thickness &, for outward flow to a ring slot of dia-
Fort > r,*/18c,v and 0 < r < rW meter rW (Fig. 20(a)). If the pore pressure in the ring is
rapidly lowered at t = 0, and it is assumed that the dis-
tribution of excess pore pressure U can be represented
h=h,[l -exp(&8T,)+Gexp(i-87)]
by a parabola, then there are two cases to consider:
before pore pressure reductions start at the centre of the
(20) ring (t -c t,) (Fig. 20(b)), and after pore pressure
reductions start at the centre of the ring (t > t,) (Fig.
where equation (5) applies.
20(c)).

APPENDIX 3. DERIVATION OF DISTANCE OF


If U is represented by a parabola (Fig. 20(b))
INFLUENCE FOR INWARD RADIAL FLOW TO A WELL
The background to this solution is described, and U = Dr* + Er + F (27)
some symbols are defined, in the main text. The draw-
down i after time t at radius r from the axis of a well of The boundary conditions are
radius rW was derived by Powrie & Roberts (1990) and
r=r,:ti=O (284
can be expressed as

h r = r, - L: U = y,h, (28’4
h=” (r’ - 2Rr + R’) r,<r<R (21)
(R - r,S r=r,-L:at2/dr=o (284
98 POWRIE AND PREENE

Drawdown ,k On substitution from equation (18)


t?, at
outer face rr
of ring Y
.4”=211Yw~ (hw/LZ){L2 - [(2Lr, - r,‘)
E”’ I r.-L
:
I + 2 (rw - L)r - r2]}r dr (33)

k
(4
which, after much algebra,

“Y ii 6z
becomes

AV= Ww (4Lr, - P)
6E”’

Darcy’s law can be applied at the inner face of the ring


(r = r,) to calculate the volumetric flow rate 4 from the
layer. The slope of the isochrone at any point is
&i/ar = y, i, where i is the hydraulic gradient

q=-_Ak,j=-2nr
WSzk ‘E
” Y, ar ,=,*
Radial distance r

(b) = 2nr, Szk, ah (35)


ar ,=I.
so

4nr, bzk, hw
4= (36)
L
The magnitude of 4 can be equated to the rate of
change of volume

(37)

Fig. 20. Analysis of radial flow to pumped ring slot: (a)


and from equation (34)
pumped ring slot of radius rr ; (b) distribution of excess
pore pressure at time t < t,, (c) distribution of excess av -
- =* (Zr, _ L) g
ire iressure at time f > t, at “1 (38)

If the magnitudes of equations (36) and (38) are equated


Hence 4n&k, /iwr,
(39)
D = -Y,t?,JL? (2W L
SO
E = - 2y, h,(L - r,)/L? (29b)
12k, E,‘r, ’
dt = ‘{2r,L - L?} dL (40)
F = -Yy,~Jrw2 - 2Lr,)/LZ (294 Yw s0 s0
giving, for t < t, and (rW - L) < r < rw Hence

1
Y k c L
U = +-- [(2Lr, - rw2) + 2(r, - L)r - r’] (30) 12r,c,vt = r,L? -7
[ 0

which, expressed in terms of drawdown for t -c t, and 12c,,t c L3


2 2 for t < t, (42)
(r, ~ L) < r < r, , is equivalent to equation (18). r, rw 3rw3
The vertical compression p at radius r from the centre
of the ring (r,,, > r > 0) is t,, the time at which pore pressure reductions start at
the centre of the ring, can be found by evaluating equa-
p = GzAcrv’JEv’ (31a) tion (42) when L = r,

t, = rW2/18c,v (43)
AC,’ = -AU = y,i (31b)
To obtain an explicit expression for L, equation (42) can
The overall reduction in volume of the layer at time t is be expressed as a cubic in L/r,

AV=
s ‘r
r,-L
2rrrp dr = 2ny, -
62 ‘-
Ev’ s ,,-I.
hr dr (32)
d
7,
3
3c
r,
*+7=0
36c,v
rW
t
(44)
CONSTRUCTION DEWATERING SYSTEMS IN FINE SOILS 99
which, if expressed in the form of equation (24) where If the magnitudes of equations (53) and (54) are equated
a=L/r,,A=-3,B=OandC=36c,vt/rW2,canbe
solved by substituting the identities in equation (25) and ny, iw i-w 92 am
4?r6zk,&(l - m) = 2E ,
solving fi3 - 3pp - 2q = 0. This gives equation (19), for V at
t < r,‘/18c,, , where cos 0 = 1 - 18T, and equation (5)
applies. These expressions can be applied to equation so
(18) to calculate drawdown, and to equation (36) to cai-
8k,E,’ ’ m 1
culate flow rate. - dt= - (56)
2 o (1 -m)dm
yWrW s ,, s
(b) t B t, Hence
If ii is represented by a parabola (Fig. 20(c)), then

U = G? + Hr + I (45) ?(L-fJ= -ln(l-m) (57)


The boundary conditions are

1
r=rW:ti=O (464 m = 1 - exp 8c,v(t -
- r,2 Q for t 2 t, (58)
r = 0: U = (1 - m)y,& Wb)
and substitution oft, from equation (43) gives
r=O:aii/ar=o (46~)
Hence m-l-exp[i-y] fort>r,*/l8c,, (59)
G = - (1 - m)y,h,/rw2 (474
This can be substituted into equation (49) to give an
H=O VW expression for drawdown (equation (20), for t > t, and
I = (1 - m)r, 5 0 < r < r,) where equation (5) applies, and substituted
(47c)
into equation (53) to give an expression for flow
giving
4 = 4n6zk, fi, exp [$ - 8 T,] (60)
U = (1 - m)y, h,(l - r*/r,‘) for t > t,
OGrGr, (48)
APPENDIX 5. SELECTION OF SOIL PERMEABILITY
or, in terms of drawdown
To obtain the optimum permeability estimate for a
fine-grained aquifer, the available data should be used
6 = h,[m + (1 - m)r’/rW2] for t> t,
as follows.
OGrGr, (49) The most reliable permeability estimates are from
pumping tests: permeability should be based on
The vertical compression p at radius r from the centre analysis of drawdown in piezometers. The method of
of the ring (r, > r > 0) is given by equation (31). The analysis should be appropriate to the aquifer boundary
overall reduction in volume of the layer at time t is conditions.
If no pumping test data are available, and the soil is

A1/=
s‘=2mp
0

On substitution
62
dr = 2ny, F

from equation
Iw-

s II
(49)
hr dr (50)
of isotropic permeability and contains less than -20%
of silt and clay size particles,

to represent
particle size analyses
should be used. Provided that sufficient samples have
been obtained local variations in the
aquifer, Hazen’s rule should be used to estimate per-

1
meability from grading curves. The optimum per-
tn+(l-m)$ rdr (51) meability estimate shall be

(a) for tube samples: the mean permeability from all the
grading curves
ny, h, r,*Sz (b) for bulk samples: the minimum permeability from
AV= (1 +m) (52) all the grading curves.
Ev’
If the soil contains more than - 20% particles of silt
Darcy’s law can be applied at the inner face of the ring and clay size, or has significantly anisotropic per-
(r = r,) to give equation (35) and meability (or no grading curves are available), in situ
tests in boreholes and piezometers should be used. Pro-
4 = 4n&k, h, (1 - m) (53) vided that sufficient tests have been carried out
throughout the aquifer, the optimum permeability esti-
The magnitude of the flow rate 4 can be equated to mate shall be the maximum result.
the rate of change of volume, and from equation (52) Permeability determined from laboratory per-
meameter or consolidation tests should not be used
av
_=
ny www
i r 2S~ am unless confirmed by other permeability methods and by
(54)
at 2Ev’ at visual examination. Intact samples should be examined
100 POWRIE AND PREENE

visually to confirm the broad range of permeability and Carslaw, H. S. & Jaegar, J. C. (1948). Conduction ofheat
degree of anisotropy, and, if necessary, to exclude in solids. Oxford University Press.
unrepresentative values obtained by other methods. Cooper, H. H. & Jacob, C. E. (1946). A generalized
graphical method for evaluating formation con-
stants and summarizing well field history. Trans.
NOTATION Am. Geophys. Un. 21, 526-534.
area of flow Hall, H. P. (1955). An investigation of steady flow
distance between two parallel slots towards a gravity well. Houille Blanche 10, No. 8,
coefficient of vertical consolidation for horizo- S-35.
ta1 drainage Hazen, A. (1892). Physical properties of sands and
effective aquifer thickness gravels with reference to their use infiltration, p. 539.
10% particle size Report to the Massachusetts State Board of Health.
void ratio Head, K. H. (1980). Manual oj soil laboratory testing.
the error function London: Pentech.
soil stiffness modulus in one-dimensional verti- Hvorslev, M. J. (1951). Time lag and soil permeability in
cal compression groundwater observations. Bulletin 36. Vicksburg:
well function (Jacob & Lohman, 1952) Waterways Experimental Station.
drawdown of piezometric level Jacob, C. E. & Lohman, S. W. (1952). Nonsteady flow
drawdown at a well or slot to a well of constant drawdown in an extensive
hydraulic gradient aquifer. Trans. Am. Geophys. Un. 33, No. 4, 559-569.
Bessel functions of the first kind and zero and Parry, R. H. G. (1971). A direct method of estimating
first order respectively settlements in sands from SPT values. Proceedings of
soil permeability symposium on interaction of structures and founda-
permeability in horizontal direction tions. Birmingham: Midlands Soil Mechanics and
distance of influence Foundation Engineering Society.
flow rate from the well Powrie, W. & Roberts, T. 0. L. (1990). Field trial of an
radial distance from the centre of a well ejector well dewatering system at Conwy, North
radius of a well (individual or equivalent) Wales. Q. J. Engng Geol. 23, 169-185.
distance of influence for radial flow Preene, M. (1992). The design of pore pressure control
elapsed time from start of pumping systems in fine soils. PhD dissertation, University of
elapsed time when T, = 0.1 London.
time factor cav t/r_* Preene, M. & Powrie, W. (1993). Steady-state per-
excess pore pressure formance of construction dewatering systems in fine
(negative) pore pressure at air entry soils. GCotechnique 43, No. 2, 191-205.
initial excess pore pressure Rao, D. B. (1973). Construction dewatering by vacuum
volume wells. Indian Geotech. J., 217-224.
linear distance from edge of a pumped slot Richardt, F. E. (1957). A review of the theories for sand
Bessel functions of the second kind and zero drains. J. Soil Mech. Fdn Engng Div. Am. Sot. Civ.
and first order respectively Engrs 83, SM3.
root of equation (11) Roberts, T. 0. L. (1988). Seepage in shallow unconfirmed
positive root of J,,(B) = 0 aquI$rs: permeability limits for gravity drainage.
unit weight of water PhD dissertation, University of London.
infinitesimal thickness of a horizontal soil layer Schofield, A. N. & Wroth, C. P. (1968). Critical state soil
vertical compression of soil layer mechanics. London: McGraw-Hill.
vertical effective stress Stroud, M. (1989). The standard penetration test-its
application and interpretation. Penetration testing,
pp 29-49. London: Thomas Telford.
REFERENCES Terzaghi, K. & Peck, R. B. (1967). Soil mechanics in
Bolton, M. D. (1991). A guide to soil mechanics. Cam- engineering practice, 2nd edn. New York: Wiley.
bridge: M. D. & K. Bolton. Theis, C. V. (1935). The relation between the lowering of
Brand, E. W. & Premchitt, J. (1982). Response charac- the piezometric surface and the rate and duration of
teristics of cylindrical piezometers. Giotechnique 32, discharge of a well using groundwater storage.
No. 3,203-216. Trans. Am. Geophys. Un. 16, 519-524.

You might also like