Hydrotogical Sciences-Journal-tles Sciences Hydrologiques, 45(4) August 2000

589

Well storage effect during pumping tests in an

aquifer of low permeability

V. S. SINGH
National Geophysical Research institute, Hyderabad 500 007, India
Abstract Well storage effect is generally considered while interpreting pumping test
data from large diameter wells. However, in an aquifer of low permeability, the well
storage is found to be significant during pumping tests conducted on bore wells. The
interpretation of such data gives ambiguous results unless well storage effect is taken
into account. A field example is presented to illustrate the difficulty in interpretation
of the pumping test data. In order to take into account the well storage effect, a finite
difference approach of interpreting pumping test data is suggested.

Effets de capacit de puits lors des essais de dbits dans des aquifres
peu permables
Rsum Les effets de capacit de puits sont en gnral pris en compte lors des essais
de dbits sur des puits de grand diamtre. La capacit de puits semble cependant
toujours tre un facteur significatif pour l'interprtation des essais de dbits dans des
puits fors en milieu peu permable. L'interprtation de telles donnes donne en effet
des rsultats ambigus moins de considrer l'effet de capacit du puits. Dans cet
article, nous prsentons un exemple concret afin d'illustrer la difficult d'interprtation des donnes d'essais de dbits de puits fors dans de tels milieux. Afin de
prendre en compte l'effet de capacit de puits, nous proposons une approche utilisant
la mthode des diffrences finies pour l'interprtation des donnes.

INTRODUCTION
A reliable assessment and management of groundwater potential in any area needs
accurate and representative estimation of aquifer parameters. The pumping test is one
of the widely practiced methods to determine aquifer parameters. The timedrawdown data obtained during the pumping test is matched with model values and,
after obtaining the best match, the model parameters are considered to be
representative of field parameters. The model becomes a true representative of field
situations only if actual field conditions are taken into account. Many of the field
conditions are simplified as they cannot be exactly represented through models. It is
seldom that exact field conditions are known and, by and large, one arrives at
equivalent models representing the complex field situations. One of the boundary
conditions in most of the models is that the pumping well has an infinitesimally
small diameter. This implies that the effect of well storage is negligible and the entire
pumped water comes from the aquifer. However, under many circumstances, the well
storage is found to be significant and the estimation of aquifer parameters becomes
ambiguous if the well storage effect is not taken into account.

Open for discussion until 1 February

2001

V. S. Singh

590

WELL STORAGE
Papadopulos & Cooper (1967) considered the effect of well storage while developing
an analytical method to estimate aquifer parameters. The model proposed by them has
been used mostly for the cases where well storage effect is significant due to "large
diameter" of the wells. It has been shown that the effect of well storage is significant
till time, t which is expressed as:
t > 250

(1)

where T is the transmissivity of the aquifer and rc is the radius of that part of the well
where the water level declines. It is evident from the above expression that well
storage is significant for a longer duration of pumping test when either the radius of
the well is large or the transmissivity of the aquifer is low. In the case of larger well
radius (i.e. large diameter well), several models have subsequently been presented for
different field conditions (Singh & Gupta, 1986, 1991; Rushton & Redshaw, 1979).
However, in the case of bore wells, it is often presumed that the well storage effect is
negligible. Considering a bore well of 0.08 m radius (which is common in practice,
particularly in hard rocks) and using equation (1), the duration for which the well
storage is significant for various values of aquifer transmissivity is shown in Fig. 1. In
the case of an aquifer of low permeability, the well storage effect becomes significant
for the larger part of the pumping duration and the estimation of aquifer parameters
becomes ambiguous unless the effect is taken into account. This phenomenon is
observed not only in the pumping well, but also in the observation well.

NO WELL STORAGE EFFECT

t = 2 5 0 r c 2 / T for

10"

(0'

10'

c = 0-08

1C?

TIME IN MIN

Fie. 1 Duration of well storage effect for various values of T.

CASE STUDY
In hard rock, aquifers are of poor permeability. To carry out pumping tests, a bore well
of 0.08 m radius with an observation well is often used. A pumping test is considered

Well storage effect during pumping tests in an aquifer of low permeability

591

here to illustrate the well storage effect. The pumping and observation wells tapped
weathered and fractured granite. From the drilling data, the aquifer was found to be
located between 24.6 and 28.8 m below the ground surface. The granite in this depth
range is semi-weathered and fractured. It is underlain by hard massive granite. The
total depth drilled was 29 m. The aquifer behaviour indicates that it is confined. An
observation well was constructed at 10 m distance from the bore well. The pumping
test was carried out for about 1300 min with a constant discharge of 51.23 m J day"1.
Drawdowns were observed in both the pumping well and observation well during the
pumping phase. Recuperation in both the wells was observed for the duration of
300 min.
The time-drawdown plot of observations made in the pumped well is shown in
Fig. 2. The initial time-drawdown data (up to the first 20 min) observed in the
pumping well clearly indicate the effect of well storage. Figure 3 indicates the aquifer
contribution to the total pumpage from the well. It is evident that, during the early
stage of pumping, the well storage dominated and most of the pumped water came
from this. However, the aquifer discharge gradually increased with time and became
equal to the total pumpage from the well. Therefore, the aquifer contribution did not
remain constant throughout duration of the test. Due to variation in discharge rate, the
time-drawdown observed in the observation well changes its slope and conventional

TIME

IN MIN

Fig. 2 Time-drawdown plot of pumping well.

10"'

10
TIME IN MIN

I01

I0 Z
-

Fig. 3 Aquifer contribution during pumping test.

V. S. Singh

592

Table 1 Estimated aquifer transmissivity (m" day' ).

Pumping phase:
Entire data
Initial part
Later part
Recovery phase:
Entire data
Initial part
Later part

Pumping well

Observation well

3.4 (Hantush)
1.95(Hantush)
13.3 (Theis)

36 (Hantush)
5.7 (Hantush)
93 (Hantush)

9.22 (Theis)
11.4 (Theis)
126 (Theis)

121.9 (Theis)
113.6 (Theis)
156.0 (Theis)

Theis: confined aquifer; Hantush: leaky aquifer.

interpretation techniques give misleading results. Table 1 shows the values of

transmissivities interpreted from different parts of the time-drawdown curve.
Considering all the data points from the observation well, best matches were
obtained with Hantush type curves suggesting the aquifer to be leaky (Fig. 4), which is
not the case. Similarly, a semi-log plot of time-drawdown from the observation well
shows a reduction in slope, indicating either a lateral increase in transmissivity, or a
nearby recharge boundary. However, such a distortion in time-drawdown curve was
found to be mainly due to variation in the aquifer discharge due to well storage effect,
as shown in Fig. 3.
Therefore, in order to take into account the well storage effect, while interpreting
the time-drawdown data, the radial flow finite difference method as suggested by
Rushton & Redshaw (1979) was adopted.
i

i
i

P- "i
-T

t::r_
1

V T

5
o

>

I !..

"T"

f\

_ ._._ -_

Sj^J

I
i

kL ,
ii

JL
^-

10

100

Time [min]
Fig. 4 Time-drawdown match with Hantush type curve.

jt r

_j_t.- T :

--

u
T

t-4 -\[

Well storage effect during pumping tests in an aquifer of low permeability

593

FINITE DIFFERENCE RADIAL FLOW MODEL

In order to obtain a drawdown solution to the radial groundwater flow equation, a
versatile, discrete time, discrete space numerical model was presented by Rushton &
Redshaw (1979). The method can be employed to take into account the variety of
boundary conditions which occur during the pumping tests. As the well is pumped, the
flow towards the well becomes radial and is considered to be symmetrical. The entire
aquifer zone from the well centre to the outer boundary is divided into number of grids
such that the grid spacing Aa = lnr (where r is the distance from the well centre).
Similarly, the time since pumping started is also discretized. Initial conditions such as
the drawdown at each node are written down at the beginning of the test (zero in this
case). The boundary conditions at the well (the discharge) and at the boundary
(prescribed potential or prescribed flux) are also described. The solution to the partial
differential equation describing radial groundwater flow is obtained by using the finite
difference technique with initial and boundary conditions. The finite difference
equation is expressed as:
7 T ( 5 + i - 2s + s+i ),+A, = S -f- (s,+A/ - snt )+q
A<2~

At

A,

r2

(2)

<+y

where sn is the drawdown (m) at nth node of radial distance r (m) and time t (day), kr is
the hydraulic conductivity (m day'1), S is the storage coefficient, q is the pumping rate
(m3 day' 1 ), and m is the saturated thickness of the aquifer (m).
The above expression, written at various nodal points, forms simultaneous
equations which may be solved for drawdown under given conditions.
The well storage is considered by assuming that the aquifer extends into the region
of the well. The properties of this region are considered differently so that it represents

10

100
TIME

IN MIN

1000
>

Fig, 5 Time-drawdown/recovery plot of (a) pumping and (b) observation well.

594

V. S. Singh

free water into the well. In the radial flow model, the horizontal hydraulic resistance
(Aa2/mkr) and time resistance (At/Srn2) at the node representing well area, are suitably
modified to represent free water in the well. Similarly, the well loss is accounted for by
modifying the horizontal hydraulic resistance representing the edge of the well.
Initially, the aquifer parameters from Table 1 are taken as the input values and the
time-drawdown/recovery values are calculated using the computer code described by
Rushton & Redshaw (1979). These are then matched with observed time drawdown
values. The aquifer parameters are then progressively varied till a best match is
achieved. A final match is shown in Fig. 5 and the aquifer parameters are
T = 130 m2 day"1 and 5 = 0.0012.

CONCLUSIONS
Well storage effect is found to be significant during pumping tests in hard rock
aquifers of low permeability. The conventional interpretation technique gives rise to
ambiguous results if well storage is ignored. In order to take into account the well
storage effect, a radial flow model is suggested which yields a reliable estimation of
aquifer parameters.

Acknowledgements The author is thankful to the Director of the National

Geophysical Research Institute, India, for his kind permission to publish this paper.
K. Radha, Joseph Gabriel and K. Sankar helped in the preparation of the manuscript.

REFERENCES
Papadopulos, I. S. & Cooper, H. H. Jr, ( 1967) Drawdown in a well of large diameter. Wat. Resour. Res. 3, 241-244.
Rushton, K. R. & Redshaw, S. C. (1979) Seepage and Groundwater Flow, 339. Wiley, Chichester, West Sussex, UK.
Singh, V. S. & Gupta, C. P. (1991) Interaction computer programme to interpret pumping test data from large diameter
wells. Wat. Resour. J. 169, 33-41.
Singh, V. S. & Gupta, C. P. (1986) Hydrogeological parameter estimation from pumping test on large diameter well.
J. Hydrol. 87, 223-232.

Received 20 October 1998; accepted 7 March 2000