You are on page 1of 20

Downloaded from http://qjegh.lyellcollection.

org/ at Duke University on October 13, 2012

Quarterly Journal of Engineering Geology and Hydrogeology

The analysis and planning of step drawdown tests


Lewis Clark

Quarterly Journal of Engineering Geology and Hydrogeology 1977, v.10;


p125-143.
doi: 10.1144/GSL.QJEG.1977.010.02.03

Email alerting click here to receive free e-mail alerts when new articles
service cite this article
Permission click here to seek permission to re-use all or part of this
request article

Subscribe click here to subscribe to Quarterly Journal of


Engineering Geology and Hydrogeology or the Lyell
Collection

Notes

© The Geological Society of London 2012


Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

Q. J1 Engng Geol. 1977. Vol. 10 pp. 125-143, 8 figs., 2 tables. Printed in Great Britain

THE ANALYSISAND PLANNINGOF


STEPDRAWDOWNTESTS
LewisClark
Water Research Centre, Henley.
Formerly with Hunting Technical Services Ltd.

SUMMARY
A step drawdown test is a pumping test in which the drawdown in a well is observed while the
discharge rate from the well is increased in steps. The methods of analysis and the criteria for
planning step drawdown tests are discussed and evaluated in order to produce an optimum test
programme and a comprehensive test analysis.
Step drawdown tests can be used to determine the aquifer transmissivity and the well loss
component of the drawdown in the pumped well. The aquifer storage coefficient can also be
estimated.

Introduction
The drawdown in a pumped well is made up of two components: aquifer loss
and well loss. The aquifer loss is that part of the drawdown caused by resistance to laminar
flow within the aquifer. The well loss results from resistance to turbulent flow in the zone
adjacent to the well, and through the screen. An additional component of well loss, which is
important in deep wells, is the frictional head loss during flow up the well.
Jacob (1946) suggested that the well loss is approximately proportional to the square of
the discharge rate and that the relationship between the two components of drawdown is
approximately:
Sw = BQ + CQ 2 (1)
where BQ = aquifer loss
CQ 2 = well loss
He devised the step drawdown test to determine the well loss in the above relationship, and
the effective radius of a pumped well.
The step drawdown test is now rarely used to determine the effective radius of a well
but it is widely used to determine well losses. Indeed, hydrogeological text books and most
publications on the subject give determination of well losses as the only use for a step draw-
down test. Eden & Hazel (1973) were the first clearly to suggest that a step drawdown test
should also be used to determine aquifer characteristics. The author and his colleagues have
been involved in several well testing programmes and the step drawdown test has been a
major tool for both aquifer and well evaluation.
In a recently completed project in Saudi Arabia, fifteen deep wells were drilled and
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

126 LEWIS CLARK

tested using both step drawdown tests and constant discharge tests; step drawdown tests
were also carried out on a further thirteen deep wells. This paper reviews the various methods
of analysis of step drawdown tests and discusses the principles evolved during the well testing
programme in Saudi Arabia, for both the analysis and planning of step drawdown tests.

Theory of step drawdown tests


The step drawdown test, as described by Jacob, is a test in which the drawdown of a well
is observed while the discharge rate from the well is increased in steps. The discharge rate is
kept constant throughout each step, and the change of discharge rate between the steps is
made as quickly as possible. The plot of the drawdown against time is a stepped curve;
hence the name of the test (Fig. 1).
The basic assumption in all step drawdown test analyses is that the increase in discharge
rate at the beginning of each step is equivalent to a new pump in the well with a discharge
rate equal to that increase in the discharge rate. The total drawdown in a step drawdown
test is equal to the sum of the drawdowns caused by the theoretical pumps responsible for
each discharge step in the test. Except in the first step of the test, this principle of super-
position of effects of the incremental changes in discharge rate ignores a basic assumption
underlying all well tests, which is that the piezometric surface should be at rest and in a
horizontal plane prior to each test. However, the principle of superposition is also used in
digital models of well field histories and the predictive capacity of these models suggests that
the principle is valid.

- Static Water Level

STEP 1

STEP 2

STEP 3 l

STEP 4

It I It 2 It 3 It 4 It s
TIME t

STEP DISCHARGE RATE TIME


1 Ol t2 - t l
2 Q2 t 3 - t2
3 Q3 t4- t3
4 Q4 t 5 - t4

FIG. 1. Diagrammaticplot of step drawdown test.


Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF STEP DRAWDOWN TESTS 127

The test illustrated in Fig. 1 is equivalent to pump 1 discharging at rate Q1 from time
tl to ts, pump 2 switching in from time t2 to t5 with a discharge rate of Q2-Q1, followed by
pumps 3 and 4 coming in at times t3 and t4 and continuing to t5 with discharge rates Q3-Qz
and Q~-Q3 respectively. It follows that the drawdowns through step 1 are the true draw-
downs for the discharge rate Q1, but the drawdowns in subsequent steps are not the true
drawdowns for the prevailing discharge rate because they are the sum of the effects of two or
more tests.
The mathematical expression for the drawdown to be expected throughout a stepdown
test in a confined aquifer has been modified from Jacob's original work by Sternberg (1968)
and later by Eden & Hazel (1973). When u, the lower limit of integration in the Theis
equation, is less than 0.01 then the Theis equation for a pumped well can be simplified to:
2.3Q lo 2.25Tt
Sw-- ~ g r~wS (2)
This equation can be written as follows:
sw = (a § b log t)Q (3)
2.25T
where 2.3 log rwS
a = 4IIT
2.3
b--
4IIT
Combining equations 1 and 3, the expression for a pumped well with well losses will be
Swt = (a + b log t)Q + CQ 2 (4)
where Swt = drawdown at time t
The drawdown in the well at the end of the step drawdown test illustrated in Fig. 1 is the
sum of the effects of all the steps, that is:
Swt5 = (a + b log (ts-tl))Q1 § (a + b log (ts-t2))
(03-01) § (a + b log (%-t~))(O~-O~) +
(a + b log (ts-t4))(O4-Q3) § CQ]
In the general case of a step commencing at time tx with a discharge rate increment of
Qx-Q,,-1 = AQx, the head loss at any time t due to the discharge increment AQx, will be:
sw~ = AQx(a + b log (t-tx)) § CQx2
The total drawdown due to all the steps will be"
, x=n

Swt = ~ AQx(a + b log (t-tx)) -q- CQ~


X=l

x~n X=ll

swt = ~ a AQx + ~ bAQx log (t-tx) + CQn2


x=l x=l

x=n

Swt-- aQn + b ~ AQx log (t-tx) + CQna (5)


X~I
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

128 LEWIS CLARK

The theory summarized above presupposes that the flow regime is in an unsteady state
throughout the test. This is not necessarily so. The water level in a well commonly reaches a
state of equilibrium within the period of an individual step. This is particularly the case
where leaky aquifers are being tested. In the step drawdown tests where steady state is
reached, the basic Jacob formula, equation 1, still applies but the factor B is no longer time
dependent and:
sw = BQ + CQ 2
2.3 ri
where B= ~ log - - in aquifers at equilibrium
rw
1 rw
or B = ~-H-~Ko -~ in leaky aquifers

The lack of time dependency in B means that the drawdown in each step, once steady state is
attained, is the true drawdown for the prevailing discharge rate.
The theory discussed above applies only to tests in confined aquifers. In unconfined
aquifers the conditions must satisfy the assumptions underlying the testing of confined
aquifers, plus the assumptions of Dupuit:
(i) the flow velocity is proportional to the tangent of the hydraulic gradient, and
(ii) the flow is horizontal and uniform through the thickness of the aquifer.
The latter implies that there must be no drawdown, but in practice the drawdown must be
kept low compared with the thickness of the saturated aquifer. The steady state drawdown
in a well fulfilling these conditions can be expressed:
2 _ 2.3Q ri
Sw Sw log - - (6)
2D 2IIT rw
where D ~- saturated aquifer thickness
This is directly comparable with the Thiem equation for the drawdown in a well at steady
state in a confined aquifer:
2.3Q ri
Sw = 2 ~ log
rw

Similarly the theory governing nonequilibrium confined aquifers can be applied to non-
equilibrium unconfined conditions provided no delayed yield takes place, by using
2
Sw
Sw--
2D
throughout instead of Sw.
The basic Jacob equation has been accepted as valid so far in this discussion. Rorabaugh
(1953) suggested that a more correct form of equation 1 would be:
Sw = BQ + CQ n (7)
where n ranges from 2.4 to 2.8 and averages 2.5.
Lennox (1966), supports Rorabaugh's expression but suggests that n may be as high
as 3.5.
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF S T E P D R A W D O W N TESTS 129

Bierschenk & Wilson (1961), however, used Jacob's equation successfully in work in
Iran. Eden and Hazel strongly support Jacob's equation and say that any apparent diversion
from the proportion, well loss varies with the square of the discharge rate, 'should indeed
be viewed with the utmost suspicion'. The author has found that the Jacob equation has
applied to most of the tests with which he has been involved. Of twenty-four step drawdown
tests carried out in the Saudi Arabian project, twenty-three complied with the Jacob
relationship. Variations from the relationship can usually be shown to be due to doubtful
data, or to changes in the well caused by development. This subject is further discussed in
the next section where a method for evaluating the Rorabaugh equation is given.

Step drawdown test analysis


The aim of step drawdown test analyses was originally to evaluate the well loss factor C, and
in Jacob's ease to determine the effective well radius rw. The aim of an analysis should now
be to evaluate the unknowns B and C in Jacob's equation 1, or a, b and C in Eden & Hazel's
equation 5. The evaluation of these unknown factors can then be used to determine the well
loss component in the well drawdown and the aquifer transmissivity, and to estimate the
aquifer storage coefficient.

Well losses
Methods of analysis: a review
The data from the step drawdown test of a well tapping a major confined sandstone
aquifer are given in Table 1 and are used to illustrate the various methods of analysis dis-
cussed. The results of the analyses, together with the results of a constant discharge test of
the same well, are given in Table 2 to allow comparison of the different methods.
Jacob developed the following general equation for the well loss factor C:
C = (ASiw/AQi) -- (Asiw-Z/AQi-1)
( A Q i - 1 _~_ AQi)

This equation can be solved graphically. The increments of drawdown in a test are deter-
mined by plotting the test data on a semi-log scale with time on the log axis (Fig. 2). The
line through the data of each step is then extended forward to the end of the succeeding
step. The incremental drawdowns and discharge rates from adjacent steps are substituted
into the general equation to evaluate C. In the test shown in Fig. 2, five values for C are
derived and their average is taken as the value for the test. This analytical method is used by
Walton (1962).
The Jacob method of analysis depends on the measurement of increments of drawdown
and discharge rate, and both these measurements are open to error. A faulty reading on one
step will affect the well loss calculation for that step and the one following, and commonly
leads to a wide spread in the calculated values for C. This effect is seen in Fig. 2 where an
anomalously high incremental specific capacity in Step 2 leads to a negative value for the
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

130 LEWIS CLARK

well loss coefficient in that step. Similar spreads of data points are seen in analyses published
by Jacob (1946) and Walton (1962).
Bierschenk & Wilson use a simpler approach based on equation 1. Dividing throughout
by Q, this equation becomes:
Sw
-- B q- CQ (8)
Q
The arithmetic plot of specific drawdown (sw/Q) against a discharge rate will, therefore, be a
straight line with a slope of C and an intercept of B. Bierschenk & Wilson use the same
graphical procedure as on Fig. 2 to determine the increments of drawdown. The total
drawdown for each discharge rate is then the sum of the incremental drawdown of that step
plus all previous steps. The total drawdown and discharge rate of each step are used to
prepare a specific drawdown--discharge rate plot for the test (Fig. 3). C, and hence well
losses (CQ ~) are obtained from this plot.

TABLE 1 Test well 1: Step Drawdown test data


Step I 2 3 4 5 6 l~ecovery
Discharge rate in m 3 / d a y i306 t693 2423 326t 4094 50i9 0

T i m e i n minutes from Drawdown in m e t r e s l~esidual


beginning os step Drawdown
in metres

1 - 5.458 8.170 t0 88t 15.318 20,036 0.599


Z - 5.529 8. 240 11 797 15.494 20.248 1.233
3 - 5. 564 8.346 11 90Z 15.598 20.389 4.050
4 - 5.599 8.451 12 008 15.740 20.529 4.683
5 1.303 5.634 8.486 12 078 15.846 20.600 4.578
6 2.289 5.669 8.557 12 t 4 9 15.88t 20.660 4.402
7 3.117 5.669 8.557 12 t 4 9 15.952 7-0.74t 4.26t
8 3.345 5.705 8.592 12 184 16.022 20.81t 4.226
9 3.486 5.740 8.672 12 219 16.022 20.882 4.050
10 3.521 5.740 8.672 t 2 325 16.093 20.917 4.014
iZ 3.592 5.810 8. 663 12 360 16.198 20.952 3.909
14 3.627 5.810 8.698 12 395 16.268 21.022 3.768
16 3.733 5.824 8 733 1 2 430 16.304 21.128 3.662
18 3.768 5.845 8 839 12.430 16.374 2t.i63 3.627
20 3.836 5.810 8 874 12.501 16.409 2t.t98 3.4t6
25 3.873 5.824 8 874 12.508 16.586 21.304 3.Z75
30 4.0t4 5.824 8 979 12.606 16.621 21.375 3.064
35 3.803 5.881 8 979 12.712 16.691 2t.480 -
40 4. 043 5.59t 8 994 12.747 16.726 2t.55i 2.711
45 4.261 5.59t 9.050 12.783 16.776 2t.6t9 -
50 4. 261 6. 092 9.050 12.8i8 16.797 2t.656 -
55 4.190 6.092 9.120 12.853 16.902 - -
60 4..120 6.176 9.12o 12.853 i6.938 21.663 2.359
70 4;t20 6.162 9.155 12.888 16.973 21.691 2.218
80 4.226 6.t76 9.191 12.9Z3 17.079 ZI.762 2.078
90 4;226 6.169 9.191 12.994 17.079 21.832 t.937
100 4.226 6.169 9.226 12.994 17.114 21.903 1.866
t20 4.402 6.176 9.261 13.099 17.219 2Z.008 1.726
150 4.402 6.374 9.367 13.205 17.325 22.t84 1.479
180 4.683 6.5t4 9.578 13. 240 17.395 22.325 i.303
240 1.02t
480 0.458
720 0.528
t570 0.035
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF S T E P D R A W D O W N TESTS 131

TABLE 2: Analytical results on test well 1


STEP TEST
Analytical Method Equation governing drawdown Theoretical well Theoretical drawdown Transmissivity Storage
in well loss (m) at (m) after 580 rain. at m3/m/day Coeffici ent
Q=4600 m 3 / d a y Q=4600 m 3 / d a y

Jacob ~ , 4 - - -

Bier schenk" &


s =0. 00345Q+1.03XI0-7Q 2 Z.Z 18.1 36Z -
Wilson %v

Hazel s =(0.00Z03+4.81XI0 -4
w lOgl0t)Q+ 1 98ZXI0-7QZ 3.9 18. I 380 5.7Xi0 -4

Eden and Hazel s = (o.o0zz3+s.sxbo -4_


w lOgl0t)Q+5.44X50-TQZ 3.1 19.3 36Z 5.3X50 -4

Theis (First Step) Data doubtful . - 346 5.3X50 -3


Q5 + Q Z + Q 3 + Q 4 + Q 5 + Q 6
Theis (Recovery) Using Q = 6 " - 241 -

Approximate Equi-
Sw=. 00344Q+i. 88X50 -7QZ 4.0 19.8 355 -
librium Method
CONSTANT DISCHARGE TEST Observed d r a w d o w n (m)
after 580 rain at Q
=4600 m3/day

Theis (using
p i e z o m e t e r data)
- . - 357 1. ZXI 0 -4

Theis (using pumped


well data cor~:ected 18.56 300 6.0X10 -3
for 3.0 m well losses)

@
~ .___~ ~Z~Swl

~-~ - - - ~- Step 1
~=~=~.~ z~S w 2

E IC
.,.,..
9 i 9 1 I ~ ~ ~" S t e p , 3

Step A SW ( m ) Z~Q (m3/day) Z~Sw/z~Q C= , ~ ~ ~-~,~


T sw,'J
z
oa xl0 -3 x 10-7 I ~ Step 4
1 4"60 1306 3.5 ~ AI
2 1-75 387 4.5 5.9 * ~ Sw5
3 2.65 730 3-6 -8"1
4 3"10 838 3-7 0"6 te D 5
5 3-70 833 4-4 4-2 ~-
2( 6 4"15 925 4-5 0"6 sw 6

C=0.64x10 -7
Step 6

I i I I I
10
I I I I I ! I I
lOO
I I I I I III
TIME t (min)

FIG. 2. Jacob analysis.


Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

132 LEWIS CLARK

Hazel (1973) has devised a rigorous graphical analysis of step drawdown tests in which
the true drawdown-discharge rate curve for each step is reconstructed. The test data are
plotted on a semi-log scale (Fig. 4). The resultant drawdown curve for step 1 is the true
curve for discharge rate Q~. The true drawdown curve for step 2 is built up using the
incremental drawdown between step 1 and step 2. The incremental drawdown at a point in
step 2 is measured together with the time from the start of step 2. The increment of draw-
down is then added to the drawdown in step 1 at a time equal to the time measured from the
start of step 2.
The transfer of one point is illustrated in Fig. 4; the incremental drawdown measured
120 minutes from the start of step 2 is added to the drawdown 120 minutes from the start of
step 1. Each data point in step 2 is transferred in this way to reconstruct the whole drawdown
curve at the discharge rate Q2. The curves for the further steps are then reconstructed in the
same way. Hazel uses his method of analysis to evaluate equation 4 completely. The slope of
the curves on the drawdown-time plot (Fig. 4) allow b to be calculated. The increment of
drawdown sw over one log cycle is measured, then b -~ A sw/Q. A value of b is obtained
for each step and the values are averaged for the whole test. A plot of specific drawdown-
discharge rate similar to Fig. 3 will have a slope of C and an intercept of (a -k b log t). If the
value of drawdown at unit time is taken from Fig. 4 and used to construct this plot then the
slope will be a because b log t will be zero.
Hazel also used Fig. 4 to obtain more points for the construction of the specific draw-
down plot. Consider two steps with discharge rates Qn and Qn-1, the drawdowns after time t
will be swn and Swn-~; therefore using equation 4:
Swn z (a q- b log t)Qn ~- CQn2
2
Swn-x ---- (a -]- b log t)Qn-1 -k CQn-1
Swn- swn-z = (a q- b log t)(Qn - Qn-1) -k C(Qn2 - Qn2-1)
therefore Swn -- Swn-1 = (a -k b log t) q- C(Qn q- Qn-1) (8)
Q n - - Qn-1

xlO -3

4.r

0
r I . . . . . . . . . . o .---''~
I
o
z
0
o
sw = BQ + CQ 2
- = -00345Q + 1-03 xlO-TQ 2
~ 3.5

u. ~'~"B T = 1'22/B
= 354 m2/day
D.
r

3'0 I I ~ 4
1 2
DISCHARGE RATE Q m3/day x 10 3

FI~. 3. Biershenk and Wilson analysis.


Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF S T E P D R A W D O W N TESTS 133

This is the same form of equation as equation 8 and a plot of


Swn - - Swn-1
Qn -- Qn-z
against Qn -~- Qn-x will be a straight line with a slope C and an intercept (a -~ b log t).
These data can therefore be plotted on the same graph as the specific drawdown and
discharge rate (Fig. 5). The method uses increments of drawdown and discharge and
suffers from the same defect as Jacob's method (see above), with the resultant scatter of
data points seen on Fig. 5.
Eden & Hazel have developed a method of analysing step drawdown tests based on
equation 5. This equation can be simplified:
X=12

Swt = aQn -k- b ~ AQx log (t-tx) q- CQn2


X=l

-~ aO. + bH § c o ~
x=n

where H -~ AQx log (t-tx)


X=I

The incremental drawdown from


120 minutes in step 2 is added to
the drawdown at 120 minutes in step 1
g - -

z
10

15
Reconstructed
.
d
curves
r a w d o ~

~ Step 3

~- -Step 4

'~..Step 5
4 i i
I
0,,-i J, z~sw is drawdown per log cycle

~ 20
Step ASw(m) O (m3day) Asw/Q s w,1 min (m) sw, lmin/O ".%
x10-4 x10-3 b= Sw/Q
= 4"8 x l 0 ~'4 Step 6
0.70 " 1306 5"36 3-00 2-30
2"3 Q
0-93 1693 5"49 4.15 2'45 T=
11 ~Sw
1.15 2423 4"75 6.25 2" 5 8
=, 380, m2/day
'1-65 3261 506 8.60 2 64
1-65 4094 4-03 12-15 2.97
2.08 50i9 4.14 15.60 3-11
30 I i I l J i I t I I I | i ! n ! | I I I I', 1 I I I I -

1 lo lOO 4000
TIME t (rain)

FIG. 4. Hazel analysis.


Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

134 LEWIS CLARK

The value of H is calculated for each observation using measured values of discharge rate
and time. Each value of H is then plotted against the drawdown producing a family of
parallel straight lines, each line corresponding to one value of discharge rate. The slope of
the lines is b and the intercept of each line on the drawdown axis is the drawdown at unit
time (Fig. 6).
These drawdown values at unit time are used with their respective discharge rates to
construct a specific drawdown-discharge graph, as in the Hazel analysis. The slope of this
graph will be C and the intercept on the specific drawdown axis will be a. The computations
involved in this method are tedious when done by hand but a computer programme has
been developed to handle the calculations (Holloway 1972).
The analysis of tests in which a steady state is reached in each step is much simpler. The
drawdown is not time dependent, therefore the steady state drawdown in each step is the
true drawdown for that discharge rate. The drawdown and discharge rate for each step can,
therefore, be used directly to calculate the specific drawdown for substitution in Fig. 3.
The author's colleagues commonly use an 'approximate equilibrium analysis', based on
the fact that the rate of drawdown in each step decreases with time. After a time the rate of
drawdown is generally so slow that the well can be said to be approaching equilibrium and
the test can be treated as a steady state test. Data from previous step drawdown tests have
been used to determine the length of steps needed before the errors introduced into well loss
calculations are acceptably low (Fig. 7). After about three hours the error in most cases is
below 15 per cent.
This is considered acceptable for an approximate analysis because the magnitude of
error is not much greater than that inherent in step drawdown tests. This method of analysis
should not be used when the well is clearly not approaching equilibrium, but it is useful
because, apart from the calculation of well losses, it allows an estimation of the aquifer
transmissivity using equilibrium theory.

xlo"
KEY *ltz "4t6
~1o 4
I min
"1' o Sw,, O vs 0 plot .s,6

"4~6 Swn - Sw vs O + Qn plot .3r5


On - 0
c -
using data fromsteps 4 and 6 /"~. "2t6
"5. .~e 2
V z
,5"~
.C = slope - 1 . 8 2 xlO - 7
~ 3 I/.3 a = intercept =2-03 x 10- 3
o

/ 2~3 " "2~4 .3~ b ( F i g . 4 | = ,4-81 x 1 0 - 4

o Sw = ( 2 0 3 x 1 0 - 3 4 81 x 10-4 I ~ ) Q + 1"82x 10"7Q 2

2 I I I I . I I I I I I
o 1 2 3 4 5 6 7 8 9 10

DISCHARGE RATE Q or Q § On ( m 3 / d a y x l O 3)

FIG. 5 Hazel analysis : Specific drawdown-discharge plot.


Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF S T E P D R A W D O W N TESTS 135

" ~."E . cc 0

Io c o ~

e 7

c~
c~

~x N
- =

t~

xW x ~1
\ ~-\ \ -

a e g'o~

$~" oooooo,,,,,,
e99999 ~ n ,\ ,\ ,\ ,i\ if
(uJ) ~s NMOOh~V~IQ
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

136 LEWISCLARK
Most step drawdown tests can be analysed using the methods described above. Those
cases which appear to fit Rorabaugh's equation (equation 7), should be examined carefully
for possible errors before the analyses based on Jacob's equation are abandoned. Tests
comprising only three steps are a common source of errors. A slight error in drawdown or
discharge measurements in any step will produce an erroneous specific drawdown value for
that step. One such point, plotted on the specific drawdown-discharge rate graph, can make
the well loss apparently proportional to a higher power than the square of the discharge
rate (Fig. 8). An extra step will usually confirm that the well loss is proportional to the square
of the discharge rate. A minimum of four steps therefore should be used in all step drawdown
tests to remove this source of error.

=Z
,,:z I,u

r,.)
.-=
4

,,=,~
LL
-- r
-2 x5
,._o
z 0 v8
x6
"' 0 -1

v3 7
102 10 3
LENGTH o f STEP (min)

1 Hazel, 1973. 5 "1


2 Jacob, 1946. 6 MacDonald's, 1975.
3 Lennox, 1966. 7 3
4 Walton, 1962. 8 Eden & Hazel, 1971,

Fro. 7. Accuracy of approximate methods of step drawdown test analysis.

~ d on f i v e steps 2
sw = BQ + C Q
....o
oneous curve based
on only three steps
sw = BQ + C Q n

DISCHARGE RATE Q

Fro. 8. Diagrammatic specific drawdown-discharge rate plots illustrating errors induced


in curves based on insufficient data points.
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF S T E P D R A W D O W N TESTS 137

In those cases where the data definitely fit the Rorabaugh equation, the specific draw-
down for each step is determined as in the Bierschenk & Wilson analysis. This can then be
used to solve the equation:
Sw
-- B = CQ n-1
O
which is analogous to equation 8. The term (sw/Q - B) is plotted against the discharge rate
on log-log paper. B is unknown; therefore a series of values are assumed for it, and the true
value is the one to give a straight line plot. The slope of this line is then n-1 and the inter-
cept on the (sw/Q - B) axis at unit discharge rate will equal C.
An interesting anomaly in well behaviour was found in certain deep wells in Saudi
Arabia. The drawdown-discharge graph did not pass through the origin but had a negative
intercept on the discharge axis. In these wells the Jacob equation had an additional com-
ponent and was of the form:
Sw ~ - - A + BQ + CQ ~
The main cause of this anomaly was the change in temperature conditions in the well in the
first few minutes of the test. The top of the screen in the well was about 1200 metres from the
surface and the piezometric surface was some 150 metres from the surface. Before pumping
the temperature of the water column was governed by the geothermal gradient and ranged
from about 20 to 52~ Within minutes of starting pumping the entire column was at the
aquifer temperature of 52~ This replacement of the column by less dense warm water can
decrease the drawdown of the first step by several metres. This anomaly does not affect the
well loss calculations because it is constant in all steps. It does, however, seriously affect any
transmissivity calculations derived from the specific drawdown-discharge rate plot, because
it has the effect of reducing the aquifer loss component on this graph.

The use o f well loss calculations


The analysis of a step drawdown test gives the discharge rate-drawdown relationship of
the well over the range of discharge tested. The specific drawdown at any discharge rate can
be obtained from Fig. 3 and hence the specific capacity (reciprocal of specific drawdown) can
be calculated. Alternatively, the values of drawdown and discharge rate used to construct
Fig. 3 can be plotted against each other to show directly the variation of specific capacity.
For wells reaching equilibrium in each step, this is a true specific capacity; otherwise
specific capacity at any one discharge rate varies with time. It is therefore necessary to quote
time and specific capacity together.
The well loss calculation is vital in the analysis of tests on pumping wells with no
piezometers. The knowledge of the well loss component of drawdown allows correction of
drawdown data to yield the true drawdown, thereby making possible the analysis of constant
discharge tests in such wells. The calculation of transmissivity in solitary wells by ignoring
the well losses leads to an underestimation of the transmissivity.
Well loss has commonly been used to determine a factor called 'well efficiency', defined
as the ratio of the aquifer loss to the total drawdown:
BQ
Well efficiency = (BQ + CQ 2) • 100 (in per cent)
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

138 LEWIS CLARK

The idea of well efficiency could suggest that high loss, ineff• wells are badly designed.
This is not necessarily so. Well losses comprise turbulent losses about the well screen and
frictional losses in the casing string. Design can seek to minimize the well losses but the
length of the casing string is governed by the depth of the aquifer. Well efficiencies quoted
merely as percentages are therefore of no value. Efficiency calculations are, however, legiti-
mate for comparisons between wells of similar depths in the same well field.

Aquifer parameters
Calculation of aquifer transmissivity
The data from the first step in any test, after correction for well losses, can be used in a
constant discharge analysis; Theis, Thiem or Leaky Aquifer, whichever is applicable.
The specific drawdown-discharge rate plot (Fig. 3) can be used to estimate transmissivity
in those tests reaching equilibrium in each step. The 'approximate equilibrium analysis'
uses the same principle. When only the pumped well is considered, the equilibrium formula
of Thiem can be simplified as follows:

2.3Q log ri
T - 2IIsw r'-~
where rw = effective well radius
ri -~ radius of influence of the well
The discharge and drawdown can be measured but rw and ri are unknown; the accuracy of
any value of transmissivity therefore depends on the accuracy of the estimate of ri/rw.
Experience has shown that ri/rw is about 1000 but, even if this proportion varies considerably,
the log ri/rw term wilI vary very little (Muskat 1937). Logan (1964) suggested that a value of
3.3 should be taken for log ri/rw and hence the Thiem equation will reduce to:
1.22Q
T--
Sw

This simple equation may be recognized as one of a group now widely used in hydrogeology
of the form"
Q
T =- constant
Sw

The constant varies from area to area, or more precisely between groups of wells of differing
designs and depths. The constant increases with the increase in the welt loss component of
the well drawdown. The Logan equation assumes no well loss: for alluvial wells in the Lower
Indus Plain, Pakistan, the constant was 1.32 (MacDonald with Hunting 1962); for alluvial
wells in the Yom Basin, Thailand, it was about 2.1 (Redecon 1973). The well losses decrease
with the discharge rate, therefore the specific drawdown value at the zero discharge intercept
in Fig. 3 can be substituted in the Logan equation to get an approximate value of trans-
missivity. The accuracy of this determination will depend on the accuracy of the log ri/rw
estimate but it cannot be grossly in error and does provide a valuable crosscheck for other
methods.
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF STEP DRAWDOWN TESTS 139

Application of the equilibrium method to systems in which true equilibrium has not
quite been reached, does not give large errors.
Hazel's graphical analysis enables one to calculate the transmissivity from each step.
When the drawdown is plotted against t, as in Fig. 4, then the slope of the line for each step
is b. The values of b from all steps are averaged and then substituted in the equation:
2.3
b -- 4IIT (see equation 3)

Eden & Hazel's analysis gives a family of curves with a slope of b, which can be used in
the same equation.
The analysis of recovery data is incorporated in the Eden & Hazel analysis but it can
also be analysed as in a constant discharge test to calculate transmissivity. The residual
drawdown is plotted against t/t" on semi-log paper. The increment of residual drawdown,
Sw" over one log cycle can be substituted in the equation:
2.3Q
T--
4IIs w
The main question in this calculation is the value of discharge rate to be used. We have used
the weighted mean of all the discharge rates in the test.
The transmissivity determinations lack precision under the best of conditions because
of errors inherent in the data used. It is therefore important to obtain values by as many
methods as possible to avoid the occasional wild result. It is recommended that, whenever
possible, at least one analysis based on non equilibrium conditions, and one on equilibrium
conditions, are used in each test.

Calculations of aquifer storage coefficient


The Hazel and the Eden & Hazel analyses both evaluate the equation:
2.3 2.25T
a - - 4IIT log r2wS (see equation 3)

In a Theis curve-matching analysis of the corrected step 1 data the following equation can be
evaluated :
4Tt
S-- 2
rw

Provided the transmissivity is known, both these equations can theoretically be solved for S,
the aquifer storage coefficient. Unfortunately the effective radius of the well, rw, is not known
and must be estimated. Errors in this estimation will be squared in the calculation. The
calculation may therefore give a useful estimate of the storage coefficient but it should be
remembered that the results may be greatly in error.

Summary of analyses
The results of the different analyses of the step drawdown test of Test Well 1 are given in
Table 2 together with the result of a constant discharge test on the same well.
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

140 LEWIS CLARK

Four analytical methods have given equations governing the drawdown-discharge


relationship in the well. The theoretical drawdowns after 180 minutes of the constant dis-
charge test from these equations may be compared with the observed drawdown in the well.
There is a spread of values but all theoretical values are within 10 per cent of the true value,
and the average of the theoretical values is within 1 per cent of the true drawdown in the
well. This close agreement illustrates the value of step tests in predicting the drawdown
behaviour in a well but also shows the importance of using several methods of analysis in
any one test.
The well loss element of the drawdown shows a spread of values from 1.4 to 4.0 metres.
The lowest value is from the Jacob analysis and could well be the result of the errors inherent
in this method. The spread of values emphasizes the importance of using several methods of
analysis.
The transmissivity values derived by six different analyses are remarkably consistent
except for that by the Theis recovery method. The low value of transmissivity by this method
suggests that the average discharge rate used in the calculation should be weighted to the
later step discharge. The transmissivity values derived from the step drawdown test also
agree with that from the constant discharge test based on data from a piezometer. The
transmissivity from the pumped well is slightly lower than the rest but this could be due to
the well loss correction used being too low.
The storage coiefficient values from three analytical methods average 2.1 • 10-3, com-
pared with 1.2 x 10 -4 from the more dependable Theis analysis of data from the piezometer
in the constant discharge test. This error is most likely due to a wrong estimate of the effective
well radius, taken arbitrarily as 0.25 metres.
The aquifer constants derived from the step drawdown test and shown in Table 2
illustrate the fact that the storage coefficient can be estimated from such tests but cannot be
determined accurately. The transmissivity, however, can be measured to the same degree of
accuracy as by the classical analyses of constant discharge tests. The step drawdown test
therefore serves as an important check and aid to the analysis of the constant discharge test.

Planning a step drawdown test


Equipment
The analysis of a step test will be no more accurate than the data on which it is based;
therefore care should be taken when collecting the data to ensure the accuracy of any
observations. The data comprise measurements of discharge rates and water levels. In
shallow wells the water levels can be measured by electric tapes, but when water levels are
deep, an airline may have to be installed. Commonly the pressure gauge used on the airline
is graduated in 1.0 pound per square inch divisions, which are each equivalent to 2.3 feet.
More accurate test gauges can be obtained and should be used. Whenever possible, standby,
or duplicate facilities should be available; for example, two pressure gauges can be put on
each airline. High discharge rates are commonly measured by the orifice plate method. The
margin of error in this method under optimum conditions is ~ 2 per cent. Great care
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND PLANNING OF S T E P D R A W D O W N TESTS 141

should be taken to comply with the conditions governing the method (see Johnson 1966,
pp. 84-6); otherwise the margin of error may be much greater.

The number of steps


The specific drawdown-discharge rate graph (Fig. 3) is basic to most of the analyses des-
cribed. The accuracy of such a plot will depend on the number of data points on the graph;
that is, the number of steps in the test. It is therefore recommended that at least four steps
should be used to ensure accuracy. Three steps in a test are insufficient.
The specific drawdown-discharge rate graph will be most accurately defined if the data
points are spread out. Thus it is important that the discharge rates chosen in the step draw-
down test should cover most of the total discharge range of the well and the increments of
discharge rate should be fairly equal.

The length of steps


The length of the steps in a test can be varied to fit local conditions. The Eden & Hazel
method can handle steps less than an hour long but steps are usually chosen to be one hour
long for ease of timing, or 100 minutes long for ease of analysis (i00 minutes fills two Iog
cycles on a semi-log plot). It is commonly difficult to stabilize the discharge rate in the first
few minutes of a test; therefore a minimum step of 100 minutes is recommended to allow
adequate stable data to be obtained in each step. If time should allow, then three-hour steps
are recommended, because then there would be adequate data for a Theis analysis of the
first step, and the equilibrium methods would be more likely to apply. Steps longer than
3 hours cannot really be justified.
The length of steps in a step drawdown test need not be constant but the analysis of the
test is much simpler with steps of equal length. Also there is no need to start at a low dis-
charge rate and progressively increase the rate between steps. In theory, one could start at a
high discharge rate and decrease the rate between steps. In practice, however, the former is
simpler to organize.

The well testing programme


A well testing programme can comprise only a step drawdown test but this is not advisable
because it gives little indication of long-term pumping effects. Hence a constant discharge test
should be run in conjunction with the step drawdown test to give this long-term information.
The optimum well testing programme is probably one in which a 4- or 5-step step drawdown
test is followed by complete recovery, which is then followed by a 24- or 72-hour constant
discharge test at the production discharge rate. This test would end with complete recovery
of the water level.

Conclusions
Step drawdown tests are used principally to determine well losses. They can, however, be used
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

142 LEWIS CLARK

also to determine aquifer transmissivity and to estimate the aquifer storage coefficient. The
step test can, therefore, be used to supplement and check the analysis of a constant discharge
test of a well.
Eight methods of step drawdown test analysis have been discussed, six based on non-
equilibrium criteria and two on equilibrium criteria. The most rigorous methods of analysis
are those devised by Hazel and by Eden & Hazel. It is recommended that at least two
methods of analysis are used in each test, preferably one based on equilibrium and one on
non-equilibrium criteria, if possible.
In most tests, the well loss is proportional to the square of the discharge rate but in
some tests the well loss appears to be proportional to a higher power of the discharge rate.
This deviation from a squared relationship is suspect and all cases, where it occurs, should
be checked for inaccurate data.
The quality of a step drawdown test analysis is dependent on the quality of the test data.
The greatest care should be taken to obtain accurate raw data. The step drawdown test
should comprise at least four steps with discharge rates spread evenly over the full range
of the discharge capacity of the well. The steps should be at least 100 minutes long but
preferably three hours.
A step drawdown test can be used to evaluate the discharge-drawdown relationship of a
well over a restricted period of time. Information on well behaviour over an extended period
must be derived from a long, constant discharge, test. The step drawdown test, therefore,
should be run in conjunction with a constant discharge test to give the maximum information
on well and aquifer conditions.

Acknowledgements: I would like to acknowledge my debt to all my associates involved in well testing over the
past few years, in particular: Mr. W. Bakiewicz, Mr. R. A. Farbridge, Mr. R. F. Stoner and Mr. D. M, Milne.
I thank Messrs. Bakiewicz and Farbridge for reading and criticizing the manuscript of this paper.

List of symbols
Q Discharge rate from pumped well L Leakage factor
Sw Drawdown in pumping well BQ Aquifer loss
s,~ Residual drawdown in well CQ 2 Well loss
t Time
A Increment, AQ is increment of discharge rate a = 2.3 log 2.25T
from pumped well 4IIT r2wS
rw Effectiveradius of pumping well 2.3
b--
ri Radius of influence of pumping well 4IIT
T Aquifer transmissivity
x=n
S Aquifer storage coefficient
Ko Modified Bessel function H = ~ Qx log (t-tx)
x=l

References
BEIRSCHENK,W. H. & WILSON,G. R. 1961. The exploration and development of groundwater resources in
Iran. In Symposium of Athens: Groundwater in Arid Zones. Publication No. 57. Internat. Assoc. Sci.
Hydrology. 2, 607-27.
Downloaded from http://qjegh.lyellcollection.org/ at Duke University on October 13, 2012

THE ANALYSIS AND P L A N N I N G OF STEP D R A W D O W N TESTS 143

EDEN,R. N. & HAZEL,C. P. 1973. Computer and graphical analysis of variable discharge pumping tests of
wells. Civil Engineering. Trans. Inst. Eng. Austr. 5-10.
HAZEL, C. P. 1973. Lecture notes: Groundwater Hydraulics. Australian Water Resources Council. 1973
Groundwater school Adelaide. Unpublished.
HOLLOWAY,H. G. 1972. Analysis ofpumptest data. Programme WU71. Systems Branch. Irrigation and Water
Supply Commission. Queensland.
JACOB,C. E. 1946. Drawdown test to determine effective radius of artesian well. Proc. Am. Soc. cir. Engrs.
72, 629-46.
JOHNSON,E. E. 1966. Groundwater and wells. Edward E. Johnson Inc. Saint Paul.
LENNOX, D. H. 1966. Analysis and application of step-drawdown tests. Proc. Am. Soc. cir. Enffrs. Jl
Hydraulics Division. HY6. 25-47.
LOGAN,J. 1964. Estimating transmissibility from routine production tests of water wells. Groundwater, 2,
No. 1.35-7.
SIR M. MACDONALD & PARTNERS. 1975. Riyadh Additional Water Resources Study: Final Report. Ministry
of Agriculture and Water, Riyadh, Saudi Arabia.
MUSKAT, M. 1937. Theflow of homoffeneous fluids through porous media. McGraw Hill Book Co., New York.
RESOURCEAND DEVELOPMENTCONSULTANTSLTD. (Redecon). 1973. Water Resources Investigation Phrae and
Sukhothai Provinces; Groundwater Report. Royal Irrigation Department, Bangkok, Thailand.
RORABAUGH,M. I. 1953. Graphical and theoretical analysis of step-drawdown tests of artesian wells. Proc.
Am. Soc. civ. Engrs. 79. Separate No. 362. 1-23.
STERNBERG,Y. M. 1968. Simplified solution of variable rate pumping tests. Proc. Am. Soc. cir. Enffrs. Jl
Hydraulics Division HY I, 177-80.
WALTON, W. C. 1962. Selected analytical methods of well and aquifer evaluation. Bull. 49. State of Illinois.
Department of Registration and Education, Urbana, Illinois.

You might also like