10 Step-drawdown analyses

Step-drawdown analyses are done for two reasons. In the case of exploration
wells, aquifer tests are commonly preceded by step-drawdown tests to deter-
mine the proper discharge rate for the subsequent aquifer test. In the case of
exploitation wells, the results of step-drawdown analyses are used to deter-
mine the optimum production capacity and to analyse the well performance
over time, for the purpose of maintenance and rehabilitation.
The analysis of step-drawdown analyses uses diagnostic plots in which
ratios of the steady-state drawdowns at the end of each step and the discharge
rate are plotted against the discharge rate on linear paper. According to the-
ory, such a plot should exhibit either a straight or a curved line. When the
data points fall on a straight line under a slope, Jacob’s method should be
applied; in all other cases, Rorabaugh’s method is to be used.
Below, field data will be used to show how step-drawdown data can be
analysed with SATEM. Both analysis methods may be applied to confined,
unconfined, and leaky aquifers.

I 10.1 Using Jacob’s method

i files and select Jacob’s method in the method selection form.

The first screen shows you the time-drawdown data plotted in semi-log for-
mat during Step 1. This graph shows you that the drawdowns had not yet sta-
bilised at the end of this step, i.e. no steady state developed during Step l .
This implies that the observed drawdown value at the end of Step 2 needs to
be corrected. In SATEM, the Hantush-Bierschenk method is used for this pur-
pose (see Chapter 5, Section 3). On the basis of visual inspection you now indi-
cate which part of the late data exhibit a straight line. You can do this by
entering numerical values for the lower and upper time limits, after the ques-
tion marks. SATEM then shows you the best-fitting straight line through the
selected points; this is done by linear regression. Next, use the arrow keys to
indicate which drawdown will be used for the extrapolation: choose a draw-
down value at the end of Step l which is on or close to the straight-line seg-
ment. Time limits of 10 and 180 min have been selected for the straight-line
segment and the drawdown value for 120 min has been taken for extrapola-
tion to Step 2.

115
 This first type of screen is repeated for all subsequent steps. No steady
state developed during these steps either. All data in these time-drawdown
plots were selected to determine the straight-line segments, except for Step 4
where the lower limit was taken as 2 min. The drawdown value for extrapo-
lation was not changed and remained the one for 120 min. Note that if you
change this drawdown value in a particular step, SATEM will adopt the same
time value in all other steps as well.
The second type of screen shows you the diagnostic plot; the expression
sJQ is known as the specific drawdown. The s values correspond to the time
value you selected for extrapolation in the first type of screen; for Steps 2 to 6
they have been corrected by extending the slope of the straight-line segments
through the drawdown values at 120 min for previous steps and subtracting
them from the observed values.
Figure 10.1 shows that the six data points lie almost exactly on a sloping
straight line. This implies that the Jacob method (P = 2) can be used for this
step-drawdown test. The straight line was based on the full range of discharge
rates; it was determined by SATEM using linear regression. The values of B
and C can be found directly from this straight line: its slope is equal to C and
its intercept with the y-axis (Q = O) is equal to B (see Chapter 5, Section 4).
The third type of screen displays the values of B and C. The general draw-
down equation for the pumped well has been calculated as

sw = 3 . 1 1~0 - ~ ~ + 2
~ . o0 ~
- ~ Q ~ (10.1)

This screen also displays a comparison between observed and calculated draw-
s/Q = B + C 9
smecific drawdown
i n n/<l/s)

0.2
10 20 30 40 50 60

discharge i n l/s

Figure 10.1 Diagnostic plot of step-drawdown data (‘Clark’) for a n application using the Jacob
method

116
Table 10.1 Comparison between corrected and calculated drawdown values, resulting from the
‘Clark’ step-drawdown analysis using the Jacob method
Discharge rate Corrected drawdown Calculated drawdown
(m3/d) (m) from Equation 10.1
(m)
1306 4.40 4.43
1693 5.87 5.87
2423 8.78 8.75
3261 12.35 12.32
4094 16.19 16.14
5019 20.61 20.70

downs. Keep in mind that SATEM corrected the observed drawdowns for non-
steady state at the end of each step. Table 10.1shows these results. The standard
deviation of the differences between corrected and calculated drawdowns has
been calculated as 0.06 m. The difference between the means is always equal to
zero because the line in Figure 10.1 has been determined by linear regression.
The fourth and last type of screen displays a graph showing the total draw-
down in the pumped well plotted against the discharge rate. Figure 10.2
shows this relationship as plotted points; the dashed line represents the draw-
down values without the contribution of the non-linear well-loss component
(sw= BQ). This figure shows that the non-linear well-loss is negligible up to
discharge rates of some 1800 m3/d. This implies that for higher discharge
rates the well efficiency will start to decrease.

26 I I I I I I I I I

-
___ 1 inear drawdown-discharge r e l a t ionship + -

-
+ -

O 600 1200 i800 2400 3000 3600 4200 4800 5400

discharge i n n3/d

Figure 10.2 Discharge-drawdown relationship of the pumped well (‘Clark’) using the Jacob
method

117
 If you have only drawdown values at the end of each step, i.e. one draw-
down value per step, SATEM will skip the first type of screen. With only one
drawdown value per step you cannot verify whether the drawdowns did actu-
ally stabilise at the end of each step, nor can you make any corrections. To
show you the differences between a series of drawdown values per step and
only one drawdown value per step, the drawdown values at 180 min of the
previous test have been entered in the ‘Clarksv’file. If you select this file and
select again Jacob’s method in the method selection form, SATEM shows you
the diagnostic plot as the first screen on your screen. You will see that the six
data points now deviate more from a straight line than in Figure 10.1. When
you again select the full discharge range, the general drawdown equation for
the pumped well is now calculated as

S, = 3.4X10-3Q+2.1X10-7Q2 (10.2)

Equation 10.2 is not very different from Equation 10.1, so the fact that in this
test the drawdowns were not in a steady state at the end of each step does not
lead to significantly different values for B and C. The standard deviation of
the differences between observed and calculated drawdowns has increased
from 0.06 to 0.11 m.
Although in this example the value of the parameters did not change much,
it is advisable to always use the time-drawdown data during each step. This
will enable you to check whether the drawdowns at the end of each step did
actually stabilise and, if not, to correct them. Such an approach will generally
yield more consistent results and more reliable analyses.

10.2 Using Rorabaugh‘s method

The way the Rorabaugh method is used in SATEM will be illustrated using
the step-drawdown in Kruseman and de Ridder (1990), which are stored in
the ‘Sheahan’ file. Select this file from the list of existing files and select
Rorabaugh’s method in the method selection form.
The first screen immediately shows the diagnostic plot, because only one
drawdown value per step is available for this test. This implies that you must
assume that the observed drawdowns were in a steady state at the end of each
step. This screen shows that the four data points are located on a concave line,
indicating a P value larger than 2. The values for B, C, and P cannot be found
directly from this plot. It can be shown that a plot of [sdQ - BI versus Q on
log-log paper yields a straight line under a slope for only one particular value
of B (Chapter 5, Section 5).
You need to give an initial B value. Once you have done that, the second
screen shows you the corresponding log-log plot of [sJQ - BI versus Q for that
B value, together with a straight line which is drawn through the first and last

118
 ç/Q = B + C Q A (P - 1)

Ln <ç/Q - B>
-i I

-4
3 4 5

Ln Q

Figure 10.3 Plot of s, /Q-B versus Q for a B value of 0.0835 (‘Sheahan’) for a n application using
the Rorabaugh method

data points. The intermediate points are usually not located on this line, so you
t
need to change the B value repeatedly until you have found the best match. It
is simple to change the B value: if the intermediate points are located above the
straight line, you decrease the B value, and vice versa. A value of 0.0835 for B
produced the best match; Figure 10.3 shows the resulting plot. The remaining
parameters C and P can now be found from this straight line: its slope is equal
to P-1, while its intercept with the Q = 1axis is equal to log C.
The next screen displays the values of B, C, and P. The general drawdown
equation for the pumped well has been calculated as

S, = 9.7X10-4Q+2.7X10-10Q2.78 (10.3)

This screen also displays a comparison between observed and calculated

drawdowns. Table 10.2 shows these results. The standard deviation of the dif-
ferences between observed and calculated drawdowns has been calculated as
0.03 m.
The final screen displays a graph showing the total drawdown in the
pumped well versus the discharge rate. Figure 10.4 shows this relationship as
plotted points; the dashed line represents the drawdown values without the
contribution of the non-linear well-loss component (s, = BQ). This figure
shows that up to discharge rates of some 1000 m3/d the non-linear well loss is
negligible. For higher discharge rates, the non-linear well loss increases rap-
idly; above 5000 m3/d, the non-linear well loss contributes more than 50% to
the total drawdown.

119
Table 10.2 Comparison between observed and calculated drawdown values, resulting from the
‘Sheahan’ step-drawdown analysis using the Rorabaugh method
Discharge rate Corrected drawdown Calculated drawdown
(m3/d) (m) from Equation 10.3
(m)
2180 2.62 2.62
3815 6.10 6.11
6540 17.22 17.18
9811 42.98 43.01

Let’s suppose that you need a capacity of 9000 m3/d for drinking-water
supply. If you install one exploitation well for this purpose, the drawdown
will be some 35 m according to Equation 10.3. If you were to install three
wells, each with a capacity of 3000 m3/d, each well would have a drawdown of
some 4 m. The total pumping costs of these three wells would be substantially
lower than that of a single well of 9000 m3/d, due to the sharp increase of the
non-linear well losses from such a well. Although the construction costs of
three smaller capacity wells are higher than those of one high-capacity well,
the increased pumping costs for the latter will be spread over the well’s eco-
nomic life time - say a period of 20 to 30 years. In other words, in cases as
depicted in Figure 10.4, if a well field is considered there could be a n economic
trade-off between increased construction costs versus decreased pumping
costs.
If you were to analyse the same data with the Jacob method you would

Drawdown-discharge relationshiD for Dumped well

drawdown in m
46
I I I I I I I I I

O 1000 2000 3000 4000 5000 6000 7000 8000 9000

discharge in m3/d
Figure 10.4 Discharge-drawdown relationship of the pumped well (‘Sheahan’) using the Rora-
baugh method

120
obtain values for B and C that are orders of magnitudes lower than the values
presented

s, = LOX 10-4Q+4.2X10-7Q2 (10.4)

The standard deviation of the differences between observed and calculated

drawdowns now increases from 0.03 to 1.24 m, also indicating that the four
data points do not lie on a straight line.
So, you may conclude that the Sheahan test data can be fitted better with
the Rorabaugh method than with the Jacob method; in other words, the P
value differs from 2. Note that the adoption of the Jacob method would result
in substantially higher non-linear well losses than according to the
Rorabaugh method.

10.3 Guidelines
For the analysis of step-drawdown data, you may use either the Jacob method
or the Rorabaugh method. It is up to you to judge which of the two methods
gives the most consistent results for your data.
The best approach is always to start with the Jacob method. From the diag-
nostic plot you can then decide whether your data basically exhibit a straight
line (yes or no). To augment visual inspection, which is of course a subjective
measure, you can use the value of the standard deviation of the differences
between observed and calculated drawdowns as a more objective measure. If
you are satisfied with both, there is no reason to use the Rorabaugh method
as well. If, however, you nonetheless decide to apply the Rorabaugh method,
it will then yield a P value around 2 and a standard deviation of the same
order of magnitude as that obtained by applying the Jacob method. The B
value will have approximately the same value, but the C value will be differ-
ent because P is different. For example, try to analyse the Clark data with the
Rorabaugh method. In these cases, it is advisable to base your analysis solely
on the results of the Jacob method.
If the diagnostic plot shows that the data points clearly exhibit a curved
line, you are advised to use the Rorabaugh method. This method will then
give you a better match and a significantly lower value for the standard
deviation of the differences between observed and calculated drawdowns.
With the Jacob method you may even encounter a negative B value if the
P value is significantly higher than 2. Such a negative value of B is, of course,
physically impossible. You will see this phenomenon if you use the Jacob
method to analyse the data stored in the ‘Featurel’ file; its data pertain
to a step-drawdown test performed in a highly conductive fracture. If,
however, you analyse the same data with the Rorabaugh method - a value
of 0.765 for B produces the best match - the analysis will yield the following

121
general drawdown equation for this pumped well

S, = 8.9X10-3Q+2.3X10-
14Q5.57
(10.5)

The standard deviation of the differences between observed and calculated

drawdowns is now 0.01 m, whereas with the Jacob method it was 0.64 m.
As was discussed at the beginning of this chapter, there will be no non-linear
well losses for relatively low pumping rates. Whether such a situation did
occur with your data can be verified by visually inspecting the discharge-
drawdown relationship as presented in the last screen of SATEM. If you use
the Jacob method to analyse the data stored in the ‘Feature2’ file, the analy-
sis will yield the following general drawdown equation for this pumped well

S, = 1.2~10-~Q+2.0XlO-~Q~ (10.6)

From the last screen you will see that up to 1800 m3/d the non-linear well
losses are negligible. In such cases, you may not use the values of B, C and P
from Equation 10.6 to estimate drawdowns for discharge rates higher than
those used in the test itself.
If you analyse the same data with the Rorabaugh method, the analysis will
yield the following general drawdown equation for this pumped well

S, = 1.2X10~3Q+1.0X10~2Q1~01 (10.7)

Note that in this case the C value does not represent the non-linear well loss
coefficient, but rather the linear well loss coefficient. If the value of P is taken
to be 1exactly, Equation 10.7 can be rearranged to

,S = 1.12X10-2Q (10.8)

For all practical purposes, Equations 10.6 and 10.8 yield the same linear loss
coefficients. Note that in this case the dashed line in the last screen does not
represent the linear well losses. It actually represents the first term of the
right-hand side of Equation 10.7. Because P is almost equal to 1, the second
term of the right-hand side of Equation 10.7 also represents the linear well
losses. If the P value found from the analysis is significantly lower than 2 it
can thus be concluded that the last screen does not provide correct informa-
tion.
If non-linear well losses are very small or absent, it is impossible to judge
which of the two methods will yield the most consistent results. If you are
interested in a general drawdown equation for the pumped well, which you
can also use for extrapolation, you must repeat the step-drawdown test, but
with significantly higher discharge rates in each step.
Finally, the well may not have been properly developed before a step-draw-

122
 down test was performed. The diagnostic plot will then reveal sJQ values
that decrease with increasing values for Q. This will lead to negative values
for C when you try to apply the Jacob method. An analysis with the
Rorabaugh method is then impossible. You will encounter this phenomenon
when you analyse the data stored in the ‘Feature3’file.

123