You are on page 1of 60

Measurement of Hydraulic Conductivity

A. Laboratory methods

Constant Head Permeameter (from Fetter, 2001)

L = Length of the sample in the cylindrical sample chamber


A = Cross-sectional area of the sample
h = Head difference between the inflow and outflow
Q = Volumetric flowrate

The specific discharge q (flow per unit area perpendicular to the direction of flow) is given by
Darcy’s law
Q h
q= =−K sat
A L
Why do we use Ksat? (because the hydraulic conductivity depends on the volumetric moisture
content. The subscript sat refers to the saturated hydraulic as compared to the hydraulic
conductivity of a sample which is not completely saturated.

With reference to the diagram of the constant permeameter, we can see that

h= L+( P2
γ ) −( L+ h )
P2
But γ is zero because the height of the column of water above the porous plate at the
outflow end is zero
1
Therefore,

Q=−K sat A − ( )
h
L

QL
K sat =
Ah
Alternatively, if let V be the volume of water discharged from the permeameter in the time
interval Δt, then
V
Q=
Δt
Substituting for Q in equation 2 gives
VL
K sat =
A . Δt . h
Remarks
1. This approach is suitable for non-cohesive materials (silt and sands) with medium to
high K (i.e., K>10-5 m/s) (de Marsil, 1986; p.75)
L
h≤
2. Ideally, the head difference h should be less than L/2 ( 2 ). Otherwise we would
have non-laminar flow.
3. The hydraulic gradient should be nearly equal to that normally encountered in the
field
4. Flow must be steady (i.e. inflow = outflow)
5. The sample must be deaired as the presence of air bubbles reduces the cross-
sectional area of flow resulting in low values of saturated hydraulic conductivity.

2
Falling Head Permeameter

Falling Head Permeameter (From Freeze and Cherry, 1979)

The volume of water, Vt, draining from the permeameter shown above in the time interval Δt is
2
dt
V t =π ( h 0 −h )
4
where h0 is the initial water level above the outflow end, h is the water level at time t 0 + Δt, and
dt is the inside diameter of the falling head tube.

The rate at which water drains out of the falling head tube into the sample chamber Q in is
Vt dh
Q in = =−A t
Δt dt
Where At is the cross-sectional area of the falling head tube.
2
d
A t =π t
4
The volume of water draining from the sample chamber is given by Darcy’s law
h(t )
Q out =KA c
L
where h is the difference between the inflow and outflow water levels. Conservation of mass
requires
Qin =Q out
substituting for Qin and Qout in the above expression gives
dh (t ) h (t )
−A t =KA c
dt L
3
The above ordinary differential equation can be solved by separating the dependent and
independent variables (h and t, respectively) and integrating as follows:
At L
dh h t
− ∫
A c h0 h
=K ∫0 dt
A L
− t ln(h)|hh =Kt|t0
Ac 0

Therefore,
At L h
K sat = ln( 0 )
Ac t h
Substituting for At and Ac gives
d 2t L h0
K sat = ln ( )
d 2c t h

Remarks
1. This approach is suitable for materials with Ksat<10-5 m/s
2. The long pipe (falling head tube) is used to create a large hydraulic gradient

Using Pressure Heads and Elevation Heads


We can also derive an expression for K sat using pressure heads and elevation heads instead of
total head

The outflow rate Q2 is given by

Q 2 =Q out =KA ( h1 −h2


L )
P2 P2
h2 = + Z 2= +L
Where γ γ
P2
Since the pressure head γ is equal to zero,
h2 =L

similarly,
P1
h1 = + Z 1=h(t )+0
γ
Therefore,
4
h1 =h(t )

Substituting h1 and h2 in the expression for Q2 gives

Q 2 =KA ( h(tL)−L )
The inflow rate into the sample chamber Q1 is

dh(t )
Q1 =Qin=−at
dt
Since the inflow rate is equal to the outflow rate,

−at
dh(t )
dt
=KA
h(t )−L
L ( )
This equation can be solved by (using the method of separation of variables) separating
variables and integrating as follows:

H1 dh(t ) KA t 1
−∫H = ∫ dt
0 ( h(t )−L ) at L t 0
This integral can be evaluated by using integration by substitution. For example,
Let
u=h(t )−L
Then du=dh(t )
Our limits of integration have to be adjusted as follows: when h = H 0, u = H0 – L. Similarly,when h
= H1, u = H1 – L. Using these limits, we can write the above integral as
H 1− L du KA t1
−∫H − L
u a t L ∫t0
= dt
0

H −L KA
−ln(u )|H1 −L = (t −t )
0 at L 1 0
Since H0-L is equal to h0 and H1-L is equal to h, then
H −L h h0
−ln ( u )|H10 −L ==−ln ( u )|h0 =ln ( )
h

Therefore
5
at L h0
K sat = ln ( )
At h
2
dt L h0
K sat = ln ( )
2
dc t h

General remarks on laboratory methods


1. Ksat values obtained using these approaches sometimes do not bear any resemblance
to field-measured values. Also, the differences between replicate analyses may
be large
2. Ksat also varies with the size of the sample (scale effect)
3. It is desirable to use undisturbed samples.

1AQUIFER TESTING: METHODS, MODELS AND APPLICATIONS (Determination of hydraulic


properties: Field methods)

Hydraulic properties of aquifers are parameters that are used to describe the ability of a porous
medium to store water (Sy, S, and Ss) and to transmit water (K, T). These parameters can be
used to determine the quantity of water that can be taken into or released from storage in an
aquifer. They properties must be known in order to assess the quantity of water that is
available for pumping from a given aquifer. The methods used for determining these
properties are traditionally covered under well hydraulics.

Hydraulics is the study of the flow of liquids. In particular, well hydraulics is the study of the
flow of groundwater to wells. The response of an aquifer (time-drawdown, distance-drawdown
relationship) to pumping is influenced by many factors. Some are related to the physical nature
of the aquifer, others are related to conditions at the pumping well as well as groundwater flow
conditions. Below is a list of some of the factors influencing the response of an aquifer to
pumping.

Factors Influencing the Time-Drawdown (t-s) Response of an Aquifer to Pumping:


a) Factors related to the physical nature of the aquifer
 Aquifer type (confined, unconfined, leaky, etc)
 Geometrical properties of the aquifer: homogeneity & isotropy, dimensions of the
aquifer, boundaries (both physical and hydrologic), orientation of the confining layers,
etc.
b) Factors related to conditions at the pumping well
 well penetration - this is related to the length of the screened portion of the well
6
relative to the saturated thickness of the aquifer.
 well cross-section
 well diameter
 storage capacity of the well
 well losses
c) Factors related to groundwater flow conditions
 flow modes (steady-, unsteady-, quasi-steady-state flow),
 dimensionality of flow (1-, 2-, and 3-D flow)
 groundwater velocity
 pumping rate, etc.
d) Flow Regime

Design and Analysis of Pumping Tests


A pumping (aquifer) test involves withdrawal of water from an aquifer or injection of water into
an aquifer (either at a constant rate or variable rate) for a period of time and recording the
drawdown (increase in head) in the pumping well and/or observation wells at different time
intervals during and/or after pumping/injection. The data is used to determine, among others,
aquifer hydraulic properties.

The time-drawdown response of an aquifer to pumping depends on many factors. Some of


these factors are listed above. The other factors that determine the time-drawdown response
of an aquifer to pumping are aquifer properties, such as, K, T, S s, S, Sy, etc. These parameters
appear in groundwater flow equations and must be known in order to solve such equations.

The determination of aquifer parameters from aquifer test data (i.e., time-drawdown
measurements) involves the identification of a mathematical model (analytical solution) that
correctly matches the data. Various mathematical models have been developed for interpreting
aquifer test data. They are based on conceptual models of different aquifer systems. An aquifer
test site conceptual model is a simplified (mathematical, numerical, pictorial, etc)
representation of the aquifer system under investigation. Among other things, a conceptual
model of the test site shows: the aquifer type; hydrostratigraphic units; confining layers (if any);
aquifer geometry; etc (see below). The development of an accurate conceptual model of the
aquifer system under investigation, depends on how well the aquifer has been characterized.
The selection a correct conceptual model and mathematical model is a crucial step in the
analysis of aquifer test data.

Aquifer Characterization
Aquifer characterization consists of (Anderson and Woessner, 1992):
(1) An aquifer test site conceptual model
 a geological map of the area of interest
 x-sections showing hydrostratigraphic units (i.e. geologic formations with similar
hydrogeologic properties), lateral and vertical dimensions of hydrostratigraphic
7
units, well features (radius and degree of penetration), boundaries
 time-drawdown graphs of aquifer test data and pumping rate
 a topographic map showing surface water bodies and divides
 contour maps showing the elevation of the base of the aquifers and confining
beds
 isopach map showing the thickness of aquifers and confining beds

Groundwater Hydrology
The diagrams below show the cones of depression that form around a pumping well during
groundwater withdrawal as well as the radius of influence (farthest distance from the pumping
well at which the drawdown is zero).

8
This diagram shows the relationship between topography and the water table which, as you can
see, is a subdued replica of the surface topography

Time Drawdown Response Aquifers to Pumping


The time-drawdown response of an aquifer to pumping is used to identify the type aquifer the
observed response comes from because some aquifers respond differently to pumping.
9
However, there are some aquifers which exhibit similar response to pumping as can be seen in
the diagram below.

This figure is from Renard et al. (2009). The article is titled: Understanding diagnostic plots for
well-test interpretation. Please download the article.

Basic Assumptions
Many mathematical (theoretical) models have been developed to analyze and interpret
pumping test (aquifer test) data. The models are based on many simplifying assumptions. Some
10
of the assumptions on which the derivation of such mathematical models are based are listed
below. As we go through the list of these simplifying assumptions, it is very important to
consider what would happen to the time-drawdown response (of an aquifer to pumping) if any
one of these assumptions is not consistent with the conditions at the test site.

 All geologic formations are horizontal and of infinite lateral extent.


 the potentiometric surface (water table ) of the is horizontal prior to the start of
pumping.
 The potentiometric surface (water table) of the aquifer is not changing with time prior
to the start of pumping.
 All changes in the position of the potentiometric surface (water table) are due to the
pumping well alone.
 The aquifer is homogeneous and isotropic.
 All flow is radial toward a well.
 Groundwater flow is horizontal.
 Darcy’s law is valid.
 Groundwater has a constant density and viscosity.
 The pumping well and the observation wells are fully penetrating; i.e., they are screened
over the entire thickness of the aquifer.
 The pumping well has an infinitesimal diameter and is 100% efficient.

Important uses of aquifer tests and Results of aquifer tests

 determination of hydraulic properties of aquifer systems


 design of water supply wells (determination of sustainable pumping rates, etc)
 prediction of rates and directions of groundwater flow
 design of groundwater remediation systems
 groundwater resources evaluation, development, and exploitation

MATHEMATICAL MODELS USED FOR INTERPRETING AQUIFER TESTS

Model 1:Steady Radial Flow to a Well in an Infinitely Extensive confined aquifer

Figure 1 shows a fully penetrating well pumping from a confined aquifer. In addition to the
pumping well, there are two other wells that are used to monitor water levels at different
distances from the pumping well. These are called observation wells.

11
Figure 1. Elements of groundwater flow towards a fully penetrating well in a confined
aquifer (after Dawson and Istok, 1991)

Terminology
 Static water level (SWL) - level at which water stands in a pumping well prior to
pumping

 Pumping (dynamic) water level (PWL) - level at which water stands in a pumping during
pumping

 Cone of depression – potentiometric surface in the area affected by a pumping well in a


homogeneous isotropic aquifer a cone. The apex coincides with the highest drawdown.

 Radius of influence (R) – is the horizontal distance from the centre of the pumping well
to the limit of the cone of depression.

Assumptions
 Same as Basic Assumptions with one important addition, that is the assumption that
“equilibrium conditions have been achieved in the aquifer” This means that there is no
further significant drawdown with time.

Derivation of the Thiem Equation


Consider steady flow into and out of imaginary cylindrical sections centred at the pumping well
(Figure 2). Under steady state (equilibrium) conditions, the inflow into the cylindrical section

12
must be equal to the outflow, which is the pumping rate Qw.

(1)
where b is the thickness of the aquifer, A is the surface area of the cylindrical element. Equation
1 can be rearranged as follows:

(2)
Integrating equation 2 gives

(3)
where is the constant of integration, and (T = Kb) is the transmissivity. Let h(r=R) = h 0.
Substituting for h in equation 3 gives
Qw
h0 = ln R+c
2 πT

(4)
Subtracting eq. 3 from 4 gives

(5)

where is the drawdown caused by abstraction from the well. Equations 3 and 4 apply
strictly within the radius of influence of the well.
When measurements of the piezometric head at two observation wells located r 1 and r2
distances from the pumping well are known, equation 3 becomes

(6)

for the well at , and

(7)

for the well at .

13
The constant of integration is eliminated from both equations to give

(8)
Equation 8 is commonly referred to as the Thiem equation (Thiem, 1906) in the literature. The

drawdown at the well can be obtained from equation 4 by substituting the radius of the well
for . That is

(9)

14
Figure 2. Concentric cylindrical sections centred at the pumping well
Note:
The Thiem equation may be used only when steady state conditions have been achieved, that
is, when there is no longer any significant change in drawdown with time.

Model 2:Steady Flow to a well in an infinitely extensive unconfined aquifer


The solution approach for an unconfined aquifer follows in a similar manner that of the
confined case. The assumptions for this problem are essentially the same as those outlined for

15
the confined flow case. The additional assumption that is made here is that there are no
vertical gradients (i.e. flow in the aquifer is horizontal). Unlike in a confined, the thickness of an
unconfined aquifer is not constant. It is equal to the saturated thickness, h, which changes with
pumping. This is due to the fact that an unconfined aquifer is dewatered during pumping.

The outflow is given by

(10)
equation 10 can be integrated as follows
Q dr
∫ hdh= 2 πK ∫ r
(11)
to gives

(12)
where is the constant of integration. If h(r=R) is H, then substituting for h in equation 12
gives.

(13)
Subtracting eq. (12) from (13) gives

(14)
When measurements are made of the water table levels at two observation wells, eq. (12)
becomes

(15)

for the well at , and

16
(16)

for the well at .


The constant of integration is eliminated from both equations to give

(17)
Eq. (17) can be written as

(18)
When the saturated thickness is large compared to the drawdown due to the pumping well,

then the following approximation can be made: and . With these


approximations, eq. (18) becomes

(19)

where is an equivalent transmissivity for the unconfined aquifer, and eq. (19) is now
identical to the confined steady flow to a well. With these assumptions, the observed
drawdown at two observation wells can be expressed similar to the confined case, that is

(20)

17
Model 3: Unsteady Radial Flow in a Confined Aquifer - Theis Method
CONCEPTUAL MODEL

Figure 3. Cross-section of a pumped confined aquifer

Definition symbols used:

Let h = head at a distance r from the pumping well


hw = head in the pumping well
h1 = head at a distance r1 from the pumping well
h2 = head at a distance r2 from the pumping well
H = head before pumping begins
K = hydraulic conductivity
Q = pumping rate
D = aquifer thickness
S = aquifer storativity, dimensionless
s = drawdown
rw = radius of the pumping well

Assumptions
Same as for steady radial flow except for the equilibrium assumption. So, the cone of
depression continues to expand with pumping.

18
MATHEMATICAL MODEL

Governing equation

2 2
∂ h ∂ h S ∂h
+ =
∂ x2 ∂ y2 T ∂ t (21)

Because of radial symmetry, the above equation can be transformed to polar coordinates as
follows
x=r cos θ
y=r sin θ
2 2 2
r =x + y
r= √ x 2 + y 2
Using the chain rule of differentiation, we get
∂h ∂ h ∂r
=
∂ x ∂r ∂ x (22)
∂ r ∂( √ x + y ) x
2 2
= =
∂x ∂x r (23)
Therefore
∂h ∂ h x
=
∂ x ∂r r (24)

( ) ( )
2
∂ h ∂ ∂h ∂ ∂h x
= =
∂ x 2 ∂ x ∂ x ∂ x ∂r r (25)
Using the product rule of differentiation, equation 25 can be written as follows:

( ) ( )
2
∂ h ∂ ∂ h x ∂ ∂ h ∂ r ∂h ∂( x /r)
= = +
∂ x 2 ∂ x ∂ x r ∂ r ∂r ∂ x ∂ r ∂ x (26)
But
x x
= 2 2
r √x + y

Therefore

19
∂2 h x2 ∂2 h ∂h ∂( x / √ x + y )
2 2
= +
∂ x 2 r 2 ∂ r 2 ∂r ∂x (27)

( ( ) )
1 2x
√ x + y −x 2 2 2 2 2
2 2 2
∂ h x ∂ h ∂h
= 2 2+
√ x +y
∂ x r ∂ r ∂r
2 2 2
x +y (28)

( )
x2
√x +y − 2 2
2 2 2
∂ h x ∂ h ∂h
= 2 2+ √ x 2+ y 2
∂ x r ∂ r ∂r
2 2 2
x +y (29)

= +
(
∂2 h x2 ∂2 h ∂h ( x 2 + y 2 )−x 2
∂ x 2 r 2 ∂ r 2 ∂r ( x 2 + y 2 )3 /2 ) (30)

Similarly,

= +
(
∂ 2 h y 2 ∂ 2 h ∂h ( x 2 + y 2 )− y 2
∂ y 2 r 2 ∂r 2 ∂ r ( x2 + y 2 )3/2 ) (31)

Summing equations 30 and 31 gives

2
∂ h 1 ∂h S ∂ h
+ =
∂ r 2 r ∂ r T ∂t (32)

where :  s(r,t) = H - h
Equation 32 was solved by Theis (1935) subject to the following conditions:

Initial Conditions
The drawdown in the aquifer before pumping begins is zero
20
s (r , 0)=0 , r≥0

(33)
Boundary Conditions
lim s( r , t )=0 for t ≥0
r→∞ (34)

∂ s (r ,t ) Q
lim r =
r→ 0 ∂r 2 πT (35)

The Theis (1935) solution for the drawdown, s(r,t), is


Q
s(r , t )= W (u )
4 πT (36)
2
r S
u=
where: 4 Tt (37)
r = radial distance from the pumping well
S = storativity
T = Transmissivity
t= time since pumping started

∞ −u
e
W (u)=∫ du
u u (38)

Equation (33) is known as the Theis equation and W(u) is referred to as the Theis well function.
The Theis well function can be expressed as a series
∞ −u 2 3 4
∫ eu du=−0.577216−ln(u)+u−
u
+
u

u
(2)2! (3)3! (4)4!
+...
u (39)

The storativity S and transmissivity T are determined from 36 and 37, provided Q, r, t, and s are
known in addition to theoretically defined values of W(u) and u. The procedure for the
determination of S and T from field measurements by the Theis’ match point method is
described below.

21
Match Point Method

Case 1: one observation well

1. Plot theoretically defined values of W(u) versus 1/u on log-log paper (this is known as
the Theis type curve)
2. Plot drawdown s(r,t) versus t using the same logarithmic scales used to prepare the type
curve (do not draw a line through the data points)
3. Place the Theis type curve over the data type curve. Keeping the axes parallel, adjust the
data type curve until it overlies the Theis type curve or until a best fit is obtained (Figure
4).
4. From the best fit section of the curves, choose a convenient match point (e.g. W(u) = 1
or 1/u = 1). Note the match point values on both curves, that is, W*(u), 1/u*, s*(r,t), and
t*.
5. Substitute values of Q, s*(r,t), W*(u) into equation 6 and solve for T
6. Determine S using equation 37

Figure 4. Illustration of the Theis match point method

22
23
MODEL 3: SPECIAL CASES
Cooper-Jacob straight-line time-drawdown method
The Cooper-Jacob straight-line time-drawdown method is based on the observation that at
large t (t = time after pumping started), the contribution of the terms beyond ln(u) (in equation
39) to the Theis well function is insignificant. In that case the higher power terms in the Theis
well function can be ignored. We can therefore write

W (u )≈−0 . 577216−ln(u ) (40)


Recall that
r2 S
u=
4 Tt
When t is large, u is very small. Dawson and Istok (1991) suggest that higher power terms in the
Theis well function can be ignored when u < 0.03. However, it must be pointed out that
different authors use different values of u. When u < 0.03, the drawdown in the aquifer can be
expressed as:
Q
s (r , t )= (−0 .5772−ln u )
4 πT (41)
This equation can be written as

s(r , t )=
Q
4 πT (
−ln(1. 78 )−ln(
r2 s
4 Tt
) ) (42)

s(r , t )=
0. 183 Q
T
log
(
2 . 25Tt
r2 s ) (43)

Procedure
1. Plot s(r,t) versus log(t) on semi-log paper – with s on the arithmetic axis and log(t) on the
log-axis.
2. Fit a straight line to the data (ignore points at early time as they may not fall on the
straight line portion of the data)
3. For the fitted line compute the change in drawdown, Δs, for one log cycle (Figure 5). The
change in drawdown for one log cycle is given by
0 . 183Q
Δs= log(10)
T
4. Compute T using the equation above

24
Figure 5. Plot of s versus t

5. Determine the value of t at which s(r,t) = 0 and use it and equation 43 to compute S.

s ( r , t =t 0 )=0=
0 .183 Q
T
log
r2 s (
2. 25 Tt 0
) (44)
This means that

log
( 2 .25 Tt 0
r2 S )
=0
(45)

Cooper-Jacob straight-line distance-drawdown method


This method requires drawdown data from 3 or more observation wells. Also the condition u <
0.03 must be met.

Procedure
1. Plot s(r,t) versus log(r) on semi-log paper (r being the distance from the pumping
well to the observation well).
2. Draw a straight line through the data points and compute the change in drawdown,
Δs, for one log cycle (Figure 6).
3. compute T from the following expression:
Q
4 πT {
s 2 −s 1 = (−0 .5772−ln u2 )−(−0 . 5772−lnu1 ) }
(46)
0 . 366Q r
s 1 −s 2 =−Δs= log( 2 )
T r1 (47)
25
0. 366 Q
T= log10
Δs
4. Determine the value of r at s(r,t)=0 and use and equation 43 to compute S

s(r , t )=0=
0 . 183 Q
T
log
(
2 .25 Tt
r 20 S ) (48)

2 .25 Tt
S=
r 20 (49)
where r0 is value of r at s(r,t)=0 and t is the time at which measurements of drawdown were
taken in the wells whose data are plotted in Figure 3.

Figure 6. Plot of s versus r

RECOVERY TESTS
Recovery tests consist of measurements of residual drawdown in an aquifer and time since
pumping stopped either in the pumping well or observation well located at a distance r from
the pumping . In theory, residual drawdown data can be analyzed by the same methods used
for time-drawdown measurements taken during the pumping period. Residual drawdown s ’ is
given by
s '=s p (t )−s i (t ' )
(50)
The minus sign in equation indicates that the imaginary well is injecting water into the aquifer
at the same rate as the real withdrawal of abstraction and at the same location as the real
recharge well. So, the net effects of both the real pumping well and imaginary injection at the
26
observation is the sum drawdown caused by the pumping well and the imaginary well.
'
s =−
0 . 183Q
T
log
( 2
r S )
2. 25 Tt 0. 183 Q

T
2 .25 Tt '
log( 2
r S
)
(51)
0. 183Q t
s '= log( )
T t' (52)

Where t = time since pumping started, t’ =time since pumping stopped, and s’ = residual
drawdown.

Figure 7. Plot of drawdown and residual drawdown versus t

Analyses of time-drawdown measurements taken during the recovery period can be used to
check results obtained from measurements taken during the pumping period. In some cases
recovery tests give better results than other methods because the recovery of water levels
occurs at a constant rate, whereas in an actual pumping test it may be difficult to keep the
pumping rate constant.

ANALYSIS OF RECOVERY TEST DATA

 uses the method of images (image well theory)

According to the method of images, the recovery of water levels after the pumping period can

27
be simulated by an imaginary recharge well. The image well injects water into the aquifer at the
same location as the pumping well and at the same pumping rate. According to the principle of
superposition, the residual drawdown s’ is given by
s '=s p (t )−s i (t ' ) (51)
where: sp(t) = predicted drawdown for the
pumping well
si(t’) = predicted drawdown for the image
well
t = time since pumping started
t’ = time since pumping stopped

Procedure: Analysis of Residual drawdown data by the Jacob Method


As with drawdown data, the Jacob straight-line, time-drawdown method can be used with the
portion of the residual drawdown data for u’<0.03, where u’=(r 2S/4Tt’). In that case the residual
drawdown s’ is given by (Dawson and Istok, 1991)

s '=h0 −h( t ' )=


2 .303
4 πT
log 2
r S (
2 .25 Tt
−log
2 .25 T t '
r2 S ) (52)

' 2 . 303Q t
s == log '
4 πT t (53)
t
'
A plot of s versus t on semi-log paper gives a straight line, and the residual drawdown Δ s
' '

t
'
per log cycle of t is

' 2. 303 Q
Δ s ==
4 πT (54)

Example

A plot of residual drawdown versus t/t’ is shown in Figure 8. The data is from Driscoll (1986, p.
254). It is important to note that single well recovery data can only be used to determine
transmissivity.

28
Figure 8. Residual drawdown versus t/t’

29
STEP DRAWDOWN TESTS

One of the assumptions made in the development of mathematical models used in the analysis
of pumping tests is that flow in the aquifer is laminar during pumping. This assumption implies
that the drawdown is directly proportional to the pumping rate, Q (Driscoll, 1986; p. 555-556).
Therefore, we can write
s=BQ
where B = constant of proportionality
s = drawdown
Indeed the well hydraulics equations we have encountered so far are of this form. However,
analyses of real pumping tests have shown that this assumption does not always hold. In
reality, the drawdown observed in a pumping well during pumping, s w, is greater than that
predicted by the equations we have seen so far. The actual drawdown in the pumping well,
sw, generally has two components:

 a contribution to sw due to groundwater flow through the aquifer (formation or aquifer


loss)
 a contribution to sw due to head losses sustained as a result of
(a) resistance to flow as groundwater passes through the zone around the pumping
well
(b) resistance to flow as groundwater enters the well through the well screen
(c) an increase of flow velocity in the immediate vicinity of the pumping well
(turbulent flow)

Head losses (decrease in head) due to the last three factors are usually referred to as well
losses. Figure 9 shows the two types of drawdown observed in a pumping well during pumping.

30
Figure 9. Components of drawdown

While the drawdown due aquifer loss is unavoidable, the extra drawdown due to well losses
may be reduced to some extent. Well losses are caused by a number of factors, including
(Driscoll, 1986):

 damage to the formation around the borehole during drilling


 choice of well screens with insufficient open area
 poor distribution of screen openings
 insufficient length of well screen
 poorly designed filter packs
 presence of drilling fluid in the formation around the pumping well

Well losses have a significant impact on well performance. The smaller the well losses,
the more efficient is the well. The efficiency (well efficiency) of a pumping well and its
maximum sustainable pumping rate may be determined by step drawdown tests. In addition,
step drawdown tests may be used to evaluate well losses.
Procedure for conducting a step drawdown test
In a step drawdown test

31
 a well is pumped at 3 (or more than 3) successively higher rates (usually 5-8 steps) while
recording sw and t (time since pumping started)
 for each step, the well is pumped until there is no significant change in drawdown with
time
 at the end of each step, the pumping rate is increased (the pumping water level may be
allowed to recover to the static water level, if time permits)

The data collected consists of the following:

pumping time Qi Si Si/Qi


step
1 N/A Drawdown at
the end of
the ith step
2
i is the step counter and is equal to 1, 2, 3, etc

Analysis of Step Drawdown Tests


Step drawdown test data may be analysed by a number of methods (Kruseman and de Ridder,
1991; Dawson and Istok, 1991). Only the Cooper-Jacob straight-line, time-drawdown method
will be covered. This method can be applied to confined, unconfined, and leaky aquifers
(Kruseman and de Ridder, 1991). A necessary condition for the use of the Cooper-Jacob method
is that u < 0.03 (where u = r 2S/4Tt). When this condition is satisfied, the drawdown, s, is given
by

s(r , t )=
0. 183 Q
T
log
(
2 . 25Tt
r2 s ) (55)

In particular, for r = rw, equation 55 becomes

sw=
0 .183 Q
T
log
(
2. 25 Tt
r 2w s ) (56)
where: Q = pumping rate
T = transmissivity
sw = drawdown in the pumping well
rw = radius of the pumping well
S = aquifer storativity
t = time since pumping started

32
Equation 2 can be written as
s w =BQ

(57)

where B = aquifer/formation loss coefficient (a time-


dependent parameter)

BQ is “the theoretical loss of the hydraulic head due to groundwater flow through the porous
medium” Kresic (1997) - the formation loss. As written, equation 2 assumes that the drawdown
observed in the pumping well is due to formation loss only. This is usually referred to as the
theoretical drawdown. As pointed out, the drawdown observed in the pumping well is always
greater than the theoretical drawdown. Also, s is not always proportional to the pumping rate,
Q. The actual drawdown may be expressed as
2
s w =BQ+ CQ (58)
where C = well loss coefficient

rearranging equation gives


sw
=B+CQ
Q (59)
The specific capacity Q/sw is defined as
Q 1
=
s w CQ+B (60)
It decreases with increasing drawdown and increasing Q. The decrease of specific capacity with
increasing Q is attributed to the presence of turbulent losses (i.e. an increase of flow velocity in
the immediate vicinity of the pumping well). Figure 10 shows plots of pumping rate and
drawdown versus time. It will be used to illustrate the Jacob method of determining B and C
from step drawdown test data.

33
Figure 10. Drawdown versus time of a step test

Procedure for Determining B and C from Step Drawdown Tests Using the Jacob Method

1. Plot si/Qi versus Qi


2. Fit a straight line to the data
3. Determine B and C (C=slope, B=intercept).

Figure 11 illustrates the method. With B and C, it is possible to calculate the actual or expected
drawdown in the pumping well for any pumping rate. According to Kresic (1997, p. 153), C
values ranging from 2500 to 3000 s2/m5 are acceptable.

34
Figure 11. Plot of specific drawdown versus pumping rate

Well Efficiency

Well efficiency,ε, is defined as the ratio of the theoretical drawdown to the actual drawdown,
and is expressed as a percent. Thus,
BQ
∈= x 100 %
BQ+CQ 2 (61)

100 %
∈=
1+CQ/ B (62)
values less than 70% are not acceptable for new wells (Kresic, 1997). Well efficiency can also be
calculated using the distance drawdown method as shown below or the procedure described in
in Kresic (1997, p. 154-157, see handouts). The distance- drawdown method requires at least 3
observation wells.

Other Methods of Calculating Well Efficiency


The Cooper-Jacob straight-line distance-drawdown method can be used to calculate well
efficiency if drawdown measurements from at least 3 observation wells are available (Figure
12).

35
Figure12. Well efficiency from distance-drawdown data

The distance-drawdown graph is extrapolated until it intersects the vertical line r = r e. The
actual drawdown is equal to the sum of the aquifer loss (theoretical drawdown) and the well
loss (Figure 12). The well efficiency is given by

Theoretical drawdown
∈= x100 %
Actual Drawdown

36
MODEL 4: TRANSIENT, CONFINED, FRACTURED-POROUS SYSTEM

ANALYSIS OF PUMPING TESTS IN FRACTURED-POROUS AQUIFERS

Several models have been developed for the analysis of pumping tests data from fractured-
porous (dual porosity) aquifers. Two of these are widely accepted in the literature. These are

the Warren and Root (1963) pseudo-steady-state


∂s
∂ t
≈0 ( )
, block-to-fracture flow model and

the Moench (1984) transient


∂s
∂ t(≠0 )
block-to-fracture flow model. Before we describe these
two approaches, it is useful to describe the time-drawdown behavior of fractured porous
aquifers as well as that of unconfined aquifers.

Time-Drawdown Response of a Dual Porosity Aquifer and an Unconfined Aquifer to


Pumping
The time-drawdown responses of dual porosity and unconfined aquifers to pumping are very
similar. At early time, the time versus drawdown graph follows the Theis (1935) non-equilibrium
(transient or non-steady state) response; at intermediate times the drawdown is less than that
predicted by the Theis model; while at late time, the response once again conforms to the Theis
model. However, the explanations given for the behavior of these two different aquifers to
pumping at intermediate times (i.e., the response between early time and late time) is different.
In the case of dual porosity aquifers, the time-drawdown response is attributed to the onset of
matrix to fracture flow, while in the case of unconfined aquifers it is attributed to gravity
drainage.

The pseudo-steady-state flow model of Warren and Root (1963) is used in the Department of
Water Affairs (Botswana) aquifer test analysis software: TESTCURVE.

Major Features of Dual (double) porosity aquifers


 They consist of matrix blocks which are separated by fractures (both the fractures and
the matrix blocks constitute the aquifer)
 Transmissive voids - pores (in the matrix blocks) and fractures
 Fractures are more conductive but have lower storage capacity than matrix blocks

37
Occurrence of Dual Porosity Aquifers
Double porosity aquifers underlie extensive areas of Botswana. They include
 fractured sandstone
 layered bedrock (ssts, quartzites, etc). Conductive layers are equiv. to fractures, while
less conductive layers equiv. to matrix blocks
 fractured crystalline bedrock
 karst (although the transmissive voids formed as a result of different processes)

Conceptual Model of a Dual Porosity Aquifer


Figure 13 shows a conceptual model of a dual porosity aquifer. In the conceptualisation of a
fractured-porous aquifer
 there is no distinction between the matrix blocks and the numerous low-permeability
fractures with small apertures
 matrix blocks are idealised as a systematic array of identical, rectangular parallelepipeds
(Streltsova, 1988)
 fractures are assumed to be uniformly spaced, continuous with constant apertures and
are oriented parallel to one of the principal axes of hydraulic conductivity

Figure 13. Conceptual model of a dual porosity aquifer as


(alternating layers of matrix blocks and fractures)

Definition of symbols used:

38
Sf = Storage coefficient of the fracture (early storage coeff.)
Sm = Storage coefficient of the matrix blocks
S = Storage coefficient of the aquifer (late storage coefficient)
Kz = Hydraulic conductivity of the matrix blocks
Bz = Vertical distance between the groundwater head and the point
in the matrix blocks at which the average drawdown occurs (see
Neuman, 1972, p.1034; Fetter, 1994, p.277, Figure 8.2)
Note: The thickness of a dual porosity is taken as its saturated thickness

Flow in a Dual Porosity Aquifer

 flow occurs through both matrix blocks and fractures


 flow through the fractures (or highly conductive layers) to the pumping well is
(assumed) radial/horizontal and in a transient state
 matrix-to-fracture flow is vertical (because the bedding-parallel fractures are the most
important conduits (Botha et al., 1998))

Response of a Dual Porosity Aquifer to Pumping


Figure 14 shows the time-drawdown response of a dual porosity aquifer to pumping. The time-
drawdown-curve of a dual porosity aquifer typically has three components

 the early-time component is similar to that predicted by the Theis model


 the second or intermediate period/component is characterised by a significant
dampening of the pressure response signal and deviation of the time-drawdown curve
from the Theis model
 the third (late-time) segment of the curve again conforms to the Theis model

39
40
Figure 14. Time-drawdown response of a dual porosity aquifer

Explanation of the Pressure Response of a Fractured-Porous Aquifer

 The early-time component is influenced by fracture flow only and conforms to the Theis
solution (early Theis)
 The intermediate period component marks the onset of matrix-to-fracture flow. The
horizontal nature of this component indicates that groundwater flow from the fractures
to the well Q is approximately balanced by flow from the matrix to the fractures. This is
the reason why the drawdown (in the fractures) is almost constant. Transition from
fracture flow to flow from fractures and matrix blocks.
 As time progresses, drawdown increases again as flow from the matrix to the fractures
occurs at ever-increasing distances away from the well (Streltsova, 1988; p. 376). This
segment can be fitted to the Theis type curve to obtain the late storage coeffient (i.e.,
the storage coefficient of a dual porosity aquifer)

Additional Hydrogeologic Factors Considered in the DWA Approach

A. Skin Effect
Drawdown in and around a pumping well may be different from that predicted by theories of
groundwater flow to wells based on idealized models of aquifer systems (e.g. the Theis model).
The diagram below illustrates the effect of the damage zone and non-laminar flow on
drawdown in a pumping well.
41
Damage Zone
Damage (skin) zone refers to a zone of low conductivity developed on the borehole wall and/or
around the borehole as a result of “formation damage” during drilling. It may consist of
 finely ground drilling debris
 filter cake (a low permeability film consisting of dry drilling fluid)
 colloidal particles
 cuttings entrained during drilling

The extent of the damage zone depends on well construction and well development
procedures.

Head loss in the damage zone is generally assumed to be proportional to the pumping rate, Q.
Hence, Δsw1 may be expressed as

Δsw 1 =BQ (63)


According to Theis (1935), the drawdown in a pumping well under confined conditions is
Q
s(r w ,t )= W (u)
4 πT (64)
42
If there is resistance to flow in the vicinity of the pumping well, equation (64) has to be
modified to account for this effect (the so-called skin effect). In the Department of Water Affairs
manual for TESTCURVE, the skin effect is accounted for as follows:
Q
s(r w ,t )= (W (u )+2 ξ )
4 πT (65)
r S
w2
u=
Where: 4 Tt
2 ξ=cons tan t skin factor
Using the Cooper-Jacob approximation, we can write

s (r , t )=
2. 303 Q
4 πT { 2 .25 Tt
log 2
r S
+

2. 303 } (66)
Let rs denote the radial distance from the axis of the pumping well to the edge of the damage
zone. If the drawdown within rs has stabilized, we can write
∂h
Q=2 π bKr
∂r (67)
H = h(r,t)
The expression for the hydraulic head from the Thiem equation is

It is worth noting that K = Ks within the damage zone. A solution of (67) subject to the following
boundary conditions:
h(r =r w )=hw
h(r =r s )=hs

h s=hw +
Q
2 π bK
ln
rs
rw ( ) (68)
Note: this equation assumes that there is no K variation within the damage zone. This means
that K = Ks which is not true.

In actual fact, if a positive skin exists, there is more resistance to flow in the vicinity of the
pumping well than in the formation beyond this radial distance from the pumping well. As a
result, the head in the well is going to be
Q
hw − 2ξ
4 π bK (69)
Q
− 2ξ
Where hw is the head in the well without taking into account the effect of skin and 4 π bK is
the incremental drawdown in the pumping well caused by resistance to flow in the vicinity of

43
the pumping well due to the presence of skin. Accordingly, equation 68 has to be modified as
follows:

h s=hw −
Q
4 π bK
2ξ+
Q
2 π bK s
ln
rs
rw ( ) (70)

Combining equations 68 and 70 and solving for 2ξ gives

hw +
Q
2 π bK
ln
rs
rw
=
( ) ( hw−
Q
4 π bK )
2ξ +
Q
2 π bK s
ln
( )
rs
rw
(71)

{ }( )
K rs
ξ= −1 ln
Ks rw
(72)

As equation (72) shows, if there is no damage zone, K =Ks and ξ=0

Additional head losses may occur in the well as a result of an increase in groundwater velocity
in the vicinity of the pumping well. These are generally assumed to be proportional to Q raised
to some power n (n ≈ 2). This additional non-linear head loss (Δsw2) may be expressed as
2
Δsw 2 =CQ (73)
TESTCURVE accounts for both skin effects and non-linear head losses by incorporation of the
so-called pseudo-skin factor, ξp which is expressed as
ξ p =ξ +2 π TCQ
2 ξ p=2ξ +4 π TCQ
Taking into account the influence of skin and non-linear head losses, the drawdown may be
expressed as

s (r , t )=
2. 303 Q
4 πT {
2 .25 Tt
log 2
r S
+

+
4 π TCQ
2. 303 2 . 303 } (74)

B. Wellbore Storage
Most of the conceptual models used for interpreting pumping test data are based on the
assumption that the pumping well has an infinitesimal radius. However, real wells have finite
radii and some wells have very large diameters. When a large diameter well is used in a
44
pumping test, some of the initial well discharge will come from storage within the well casing
(Dawson and Istok, 1991). As a result, early-time drawdown data will be less than that
predicted by models that ignore the effect of wellbore storage.

When a large diameter well is used in a pumping test, the pumping rate Q has two components
at the beginning of pumping: one from the well casing and another from the aquifer. The first
contribution is given by
∂ sw ( t )
Qw =πr 2w
∂t (75)
where rw , sw, and t are the radius of the pumping well, drawdown in the well, and time since
pumping started, respectively. Equation (75) can be written as follows

Qw
∂ sw= ∂t
πr
w2
Which means
Qt
sw=
πr 2w (76)
A plot of sw versus t on log-log paper, gives a straight line. This component of drawdown is
referred to as the wellbore storage effect. The contribution from the surrounding aquifer, Q a, is
given by
∂ sw (r w , t )
Qa =−2 πr w T
∂r (77)
T =Kb
where T is aquifer transmissivity. Therefore the pumping rate Q is
∂ sw ( rw , t ) ∂s w ( t )
Q=−2 πr w T + πr 2w
∂r ∂r (78)

As explained, from equation (76) we have

log sw =log t+log


( ) Q
πr 2w
(79)
On a plot of log sw versus log t, the wellbore storage effect is recognized in the early-time
drawdown data as a straight line with a unit slope.

Wellbore storage effect diminishes as pumping progresses and eventually ceases. In Testcurve,
it is suggested that the time tc at which wellbore storage effect ceases (Agarwal, 1970) is

45
2
0 .2 r w
t c=
T (80)
However, Agarwal’s is not the only criterion (see Streltsova, 1988).

46
C. Boundary Problems in Groundwater Flow
Most aquifer test methods assume, among others, that the aquifer is of infinite lateral extent.
In reality, most aquifers are bounded (i.e. they have finite dimensions) by either impermeable
or recharge boundaries. While recharge boundaries (e.g. canals, rivers, lakes, etc) can be easily
observed in the field, very often impermeable boundaries are not exposed.

The presence of physical boundaries, e.g., impermeable boundaries, surface streams, lakes, etc,
may cause the observed time-drawdown response to deviate considerably from the predicted
time-drawdown response based on the assumption of an infinitely extensive aquifer. It also has
a significant impact on the long term sustainability of groundwater supply. In view of this, it is
important to take the presence of boundaries (both real and potential) into account when
determining the long term sustainable yields of water supply wells. Figure 15 shows two types
of physical boundaries commonly in hydrogeology.

Figure 15. (a) Recharging boundary due to a stream. Impermeable boundaries due
to: (b) lateral termination, (c) dyke, and (d) fault. From Singhal and Gupta (1999)

D. Solution of Boundary Problems in Groundwater Flow Problems: Method of Images


The presence of impermeable and recharge boundaries can be easily seen on semi-log plots of
time-drawdown data. For instance, a recharge boundary causes a flattening of the time-
drawdown plot, while a barrier boundary causes the slope of the time-drawdown graph to
increase. The method of images is commonly used to determine the location of boundaries or
47
predict the potential impact of boundaries on the sustainability of groundwater supply
schemes.

Figure 16. Effect recharge and impermeable boundaries on time-drawdown graphs.

48
Intersecting Boundaries

Figure 17. From Kresic (1997)

49
Simulation of Impermeable and Recharge Boundaries

Figure 18. Application of the method of images From Freeze and


Cherry (1979)

TESTCURVE takes into account the effects of physical boundaries and discontinuities on
drawdown. TESTCURVE can handle the following:
1. Single barrier boundary
2. Single line discontinuity
3. Two barrier boundaries intersecting at 1200
4. Two perpendicular barrier boundaries
5. Two barrier boundaries intersecting at 600
6. Two parallel barrier boundaries (compartment)
50
7. Closed square aquifer

A line discontinuity is defined as a change in aquifer hydraulic properties; say T and S, from T 1
and S1 to T2 and S2, respectively. The line over which these changes take place is referred to as a
linear discontinuity.

Procedure for Determining the Location and Trend of a Barrier Boundary


The location and trend of a barrier boundary can be determined time-drawdown
measurements from at least 3 observation wells provided the cone of depression has
intercepted the boundary.

For each observation well:

1. Plot time-drawdown data on semi-log graph


2. Extrapolate the plot beyond the point of inflection
3. Choose an arbitrary value of drawdown s1r and a corresponding value of time t1r
before the effect of the boundary on the time-drawdown plot.
4. Determine the time intercept t 1i (after the effect of the boundary on the time-
drawdown plot) at which the drawdown s 1i (= s1r) caused by the effect of the
boundary occurs
5. Use the Cooper-Jacob formula to calculate the distance of the image well from the
pumping well

( ) ( )
2 .3 Q 2. 25 Tt 1 r 2. 3 Q 2 . 25Tt 1i
s 1r = log =s1i= log
4 πT 2
r 1r S 4 πT 2
r1 i S

r 1r √ t 1i
r 1i =
t1 r
Do this for all three observation wells. Draw circles of radii t ji (j = 1,2, 3 and i = 1,2,3) centred at
the jth observation well. The intersections of the arcs give the location of the image well. The
perpendicular bisector of the line joining the image well and the pumping well marks the strike
of the boundary.

51
MODEL 5: TRANSIENT, CONFINED, LEAKY 9FLOW IN A LEAKY, CONFINED AQUIFER)

So far we have considered flow in confined aquifers in which there is no leakage from the overlying
and/or underlying aquifers into the main (pumped) aquifer. In practice, leakage of water into the main
aquifer through confining layers may occur. Aquifer systems in which this occurs are known as leaky
aquifers (Figure 19). As a result of leakage, the pumping rate Q in a leaky, confined aquifer system has
two components: one from the main aquifer, the other from outside the pumped aquifer (leakage).
Leakage must be accounted for in the derivation of well hydraulics equations for leaky aquifer systems.
Equation 32 can be modified as follows to account for leakage:
'
∂2 h 1 ∂h (h 0−h) K S ∂h
+
2 r ∂r
− =
∂r T b '
T ∂t (81)
Two cases will be considered.

Case 1: No water comes from elastic storage in the aquitard.


Basic assumptions
1. The (pumped) aquifer is bounded above by an aquitard
2. The aquitard is overlain by an unconfined aquifer (the so-called “source bed”) and bounded
below by an aquiclude.
3. The water table in the source bed is horizontal before pumping starts
4. All layers are horizontal and extend infinitely in the radial direction.
5. The initial potentiometric surface in the confined is horizontal.
6. Groundwater flow in the aquitard is vertical
7. The water table does not fall during pumping of the aquifer. This assumption is not valid unless
(a) there is continuous recharge to the unconfined aquifer or
' ' 2
S (b )
t< '
10 b K
or
b' ' K ' ' > 100 bK
where:
t = time since pumping started
S’ = storativity of the aquitard
b’ = thickness of the aquitard
b = thickness of confined aquifer
b’’ = saturated thickness of the
unconfined aquifer
K’ = vertical hydraulic conductivity of
aquitard
K’’ = hydraulic conductivity of the
unconfined aquifer
K = hydraulic conductivity of confined
aquifer

A number of methods have been developed for the analysis of pumping test data in leaky aquifers.
Virtually, all of them assume:

52
1. Infinitesimal well radius (no wellbore storage)
2. Horizontal flow in aquifers
3. Vertical flow in aquitards

The Hantush-Jacob Leaky Method


'
The vertical flow component per unit aquifer area, q , is given by
'
K ( h0 −h )
q' = '
b (82)

Let

B=
T b'
K' √ (83)

The solution to equation 81 is


Q
h0 −h= W (u , r /B )
4 πT (84)
where W(u,r/B) is the Hantush-Jacob well function. Values of W(u,r/B) are provided in most
hydrogeology textbooks. In deriving equation 84, Hantush and Jacob further assumed that the
storativity of the aquitard is zero, that is, the aquitard is incompressible.

Note:
According to Dawson and Istok (1991), equation 84 can be used to predict drawdown when

[ ( )]
2
2S 10 r
t>30 r 1−
T B (85)
and
r
<0 .1
B (86)

A plot of W(u,r/B) versus 1/u yields a family of type curves for different values of r/B. The following
description of the Hantush-Jacob Match-Point Method is from Dawson and Istok (1991).

Hantush-Jacob Match-Point Method

1. Plot W(u,r/B) versus 1/u on log-log paper. This plot is known as the type curve.
2. Plot s versus t to construct the data curve.
3. Overlay the data curve on the type curves. Keeping the axis of the two plots parallel, shift the
plots relative to each other until a good fit between the data curve and one of the type curves is
53
obtained. Note that this method does not give a unique solution for late-time drawdown data.
4. Select a match point from the best fit portion of the curves and record values of W(u,r/B)*,
1/u*, r/B*, s*, and t*
5. Use equation 4 to compute T
6. Use the expression for u to compute S, and use the expression for B to compute K’.

The Hantush-Jacob Match-Point Method is illustrated in Figure 9.1 using data in Table 9.2 (Dawson and
Istok,1991; p. 113).

Figure 19. Plot of W(u,r/B) versus 1/u

Hantush Inflection-Point Method

Hantush (1956, quoted in Dawson and Istok (1991)) developed another method for the analysis of
pumping test data in leaky aquifers. The method is known as the Hantush Inflection-Point Method.
Figure 20 shows the time-drawdown response of a leaky aquifer to pumping.

Figure 20. Time-drawdwon response of a leaky aquifer


54
(after Domenico & Schwartz, 1998)

The upward inflection of the time-drawdown graph (Figure 20) indicates the addition of water from the
overlying aquifer. Before the inflection all the water pumped comes from the pumped aquifer.

Procedure for the Inflection-Point Method:


1. Plot s versus t on semi-log paper with s on the arithmetic scale
2. Determine the maximum drawdown, (h 0-h)max. Some extrapolation may be necessary to
determine (h0-h)max.
3. Define (h0-h)i = (h0-h)max/2. This is defined as the drawdown at the inflection point.
4. From the graph of s versus t determine the time t i at which the inflection point occurs (this the
time at which the time-drawdown graph reaches (h 0-h)i). Determine the slope of the graph at t i.
Call this m i. It is better to express the slope as the drawdown per log cycle.

The Hantush Inflection-Point Method is illustrated in Figure 9.3 (Example 9.2 of Dawson and Istok
(1991)).

Figure 21. Illustration of the inflection-point method (After


Dawson and Istok (1991))

55
MODEL 6: TRANSIENT, UNCONFINED (WATER TABLE AQUIFER)

 Aquifer is dewatered during pumping


 Water is released from storage due to: (1) elastic compression of the aquifer matrix, (2)
expansion of the water, and (3) gravity drainage
 The storage coeffiecient is several orders of magnitude larger than that of confined aquifers
 Dewatering leads to a decrease in transmissivity

Figure 22 the time-drawdown response of an unconfined aquifer to pumping.

Figure 22. Time-drawdown response of an unconfined aquifer (after Domenico and Schwartz
(1998))

Three distinct segments can be distinguished in the s versus t graph: the early-, intermediate, and late-
time segments (also referred to as the first, second, and third segments, respectively).

First Segment (early-time drawdown data)


 Time-drawdown data follows the Theis model (i.e., the unconfined aquifer behaves as a
confined aquifer)
 Water is released from storage due to elastic compression of the aquifer matrix and expansion
of the water
 The storativity for this segment is comparable to that for confined aquifer
 Flow in the aquifer is horizontal
Second Segment (Intermediate time)
 Is characterized by the flattening of the time-drawdown graph
 It marks the onset of gravity drainage ( delayed yield, delayed drainage, delayed gravity
response)

56
Kh
 The shape of the graph depends on the
K
z ratio, the thickness of the aquifer, and the
distance to the pumped well (Domineco and Schwartz, 1998)
 The aquifer is being dewatered
 The is an increase in the storage coefficient

Third Segment (late-time drawdown data)


 The time-drawdown data follows the Theis model again
 The storage coefficient is equal to the specific yield

Analysis of time-drawdown data


A number of methods have been developed for interpreting aquifer test data from water table aquifers.
These include the curve-matching method of Neuman (1975). Neuman developed a set of type curves
shown in Figure 23.

Neuman’s (1975) procedure for analyzing time-drawdown data from water table aquifers.

Figure 23. Type curves for an unconfined aquifer (Domenico and Schwartz

57
Type A curves merge out of the Theis curve on the left and type B curves merge into the Theis curve on
the right. Type A curves are used to interpret early-time drawdown data, while type B curves are used to
analyze late-time drawdown data. The drawdown, s, and arguments u A,uB, and η are given by
Q
s= W (u A ,u B ,η )
4 πT (87)
r2 S
uA =
4 Tt (88)

r2Sy
uB =
4 Tt (89)
2 '
r K
η= 2
b K (90)

Procedure

1. Plot theoretically defined values of W(u A,uB,η) versus 1/uA and 1/uB on log-log paper (Theoretical
type curve)
2. Plot drawdown s(r,t) versus t using the same logarithmic scales used to prepare the type curve
(do not draw a line through the data points)
3. Place the data type curve over the Theoretical type curve. Keeping the axes parallel, adjust the
data type curve to fit the early-time drawdown data to the type A curves select a suitable match
point and record W(u Aη), 1/uA, t, and s. Do the same with the late time drawdown data and
record W(uB,η), 1/uB, t, and s or until a best fit is obtained (Figure 4).
6. Determine T and S using equations 87 and 88, respectively

58
Well Performance Tests
Determination of Sustainable Yield
According to Artiola et al. (2012), “a well’s sustainable yield is the (maximum) rate at
which a well can be pumped while ensuring that the water level I not drawn down to the
pump intake’’. Artiola et al. (2012) go on to say that “well yields are highly dependent on
the characteristics of the aquifer, the construction of the well, and the maintenance of
the well and the pumping equipment”.

HYDROGEOLOGIC CONDITIONS THAT AFFECT TIME-DRAWDOWN GRAPHS

One of the basic assumptions about the hydraulic conditions in the aquifer during
an aquifer (pumping) test is that “all changes in the position of the potentiometric
surface are due to the effect of the pumping well alone. However, apart from the effect
of the pumping well, there are other factors that can influence the hydraulic behaviour of
the aquifer during pumping. These include hydrogeologic boundaries (recharge and
impervious boundaries), vertical leakage, wellbore storage, etc. Let us look at the
effects of hydrogeologic boundaries on drawdown during pumping.

Effect of a Recharge boundary

Figure 1. Idealized cross-section of a well in an aquifer bounded on one side by a


stream.

When a well in an aquifer bounded on one side by a fully penetrating stream is


pumped, the cone of depression will expand until it intersects the stream. Thereafter two
things may happen: the river stage may or may not decline in response to pumping. Let
us consider the case in which the river stage is not affected by pumping.

The solution for the drawdown at the pumping well “is obtained by replacing the
semi-infinite aquifer to the left of the stream in Figure 1 and the constant head boundary
by the mathematically equivalent system composed of an infinite aquifer with a pumped
59
well (discharge rate Q)located at x = a, y = 0 and with a recharge well (discharge rate -
Q) located at x = -a, y = 0.

60

You might also like