6-1

6. Groundwater flow equation

Mass conservation: Min – Mout = ∆M/∆t
Darcy’s law ∆M-∆h relationship
Now we can put them all together.
An aquifer having a cross
sectional area A. Top and
bottom are confined by A
aquitards, and one end is
connected to a lake.
How is the water level in wells related to the lake-water level?
Darcy’s law: qx(x0) qx(x1)
∂h ∂h
qx = −K qy = −K
∂x ∂y x = x0 x1
In general, h and q are functions of (x, y, z, t).
In this case, h and q depends on (x, t) only.
Min − Mout = Aρw[qx ( x0 ) − qx ( x1 )]
  ∂q  ∆x = x1 - x0
≅ Aρw qx ( x0 ) − qx ( x0 ) + ∆x x 
  ∂x 
∂q ∂ ∂h 
= − A∆xρw x = − A∆xρw  − K 
∂x ∂x  ∂x 
∂  ∂h 
= A∆xρw K 
∂x  ∂x 
 6-2

Mass change in a REV having volume V = A∆x.

∆M = VρwSs∆ψ from Eq.[4-2]
At a fixed location, A
∆h = ∆(z + ψ) = ∆z + ∆ψ = 0 + ∆ψ
∴ ∆h = ∆ψ
∆x
∴ ∆M = VρwSs∆h = A∆x ρwSs∆h
∆M
Putting them all together, Min − Mout = [6-1]
∆t
∂  ∂h  ∆h
A∆xρw  K  = A∆xρwSs
∂x  ∂x  ∆t
∂  ∂h  ∂h
Making ∆t very small,  K =
 s S [6-2a]
∂x  ∂x  ∂t
If K is constant (i.e. material is uniform), we have:
∂ 2h ∂h
K 2 = Ss [6-2b]
∂x ∂t
This is the one-dimensional groundwater flow equation. Note
that [6-1] and [6-2] represent exactly the same thing. Partial
differential equations are nothing more than a language to
describe the simple conservation principle.
We can solve [6-2] in a number of ways. The next example
shows a simple numerical technique to solve [6-2b].
 6-3
Propagation of a hydraulic pulse

1
z=0

∆x
Water level in two lakes has been kept at z = 0 for a long
time. The water level in the left lake suddenly rises at t = 0.
We divide the aquifer into a number of cells and analyze the
mass balance of each cell. cell-1 cell-2 cell-3

For example, for cell-2: M12 M23

Min − Mout = M12 − M23

∆x
h2 − h1 h −h
= −ρw AK + ρw AK 3 2
∆x ∆x
h − 2h2 + h3
= ρw AK 1
∆x
∆M h2 @t +∆t −h2 h2 @t +∆t −h2
= VρwSs = ρw A∆xSs
∆t ∆t ∆t
h1 − 2h2 + h3 h2 @t +∆t −h2
∴ ρw AK = ρw A∆xSs
∆x ∆t
K∆t
∴ h2@t +∆t = h2 + (h1 − 2h2 + h3 ) [6-3]
Ss (∆x) 2
 6-4
Eq. [6-3] indicates that the future state of cell-2 is dependent
on the present state of itself and its neighbors. We can write
[6-3] for all cells and solve them simultaneously using a
spread sheet (lab exercise).

x0 x1 x2 x3 x4 x5 x6 x7 x8 x 9 x 10
t=0 1 0 0 0 0 0 0 0 0 0 0
1 0
1 0
∆t
1 0

Eq. [6-3] is called forward finite difference equation, which is

just another way of writing the mass balance equation [6-1]
and the partial differential equation [6-2].
All of these give us the increment of storage. To calculate the
storage amount itself, we need to specify:
Initial condition: h = 0 for all x > x0 at t = 0

The two end cells (0 and 10) only have one neighbor, and we
cannot apply [6-3]; i.e. there is no h-1 or h11. One way to deal
with this problem is to specify:
Boundary conditions: h = 1 at x = x0, h = 0 at x = x10
In this case, we force the boundary cells to take a specified
value, which reflects the physical constraint on the system.
This is called specified head or 1st type boundary condition.
 6-5
Next, suppose that the aquifer is “plugged” at x = x10. The
mass balance equation for cell-10 is:
∆M10
= M9,10 − 0
∆t
Therefore, we do not have to include h11 in our calculation.
This is another way of imposing a physical constraint on the
system, and called specified flux or 2nd type boundary
condition. We can also write this as:
∂h
=0 at x = x10
∂x

Solution with 1st type boundary conditions (BC’s)

This graphs show the solution of [6-2b] with the 1st type BC’s;
h = 1 m at x = 0 and h = 0 at x = 100 m
1

0.8

0.6
h (m)

0.4 t = 120 min

0.2 t = 30 min
0
0 20 40 60 80 100
x (m)

As t → ∞, the graph becomes a straight line.

(1) The mass storage does not change any more.
(2) Specific discharge is constant with respect to both t and x.
 6-6
Solution with 1st and 2nd type BC’s
Suppose we have the following BC’s to go with [6-2b].
h = 1 m at x = 0 and ∂h/∂x = 0 at x = 100 m
1

0.8

0.6
h (m)

0.4

0.2

0
0 20 40 60 80 100
x (m)

At steady state, the solution looks different from the last one.
Differential equation ↔ Governing processes
Boundary conditions ↔ Physical constraints

Steady-state solution
∆S/∆t = 0 at steady state, which means ∂h/∂t = 0 and h is a
function of x only.
d 2h d 2h
Eq.[6-2b] is now written as: K 2 =0 or 2
=0
dx dx
It is easy to show that the steady-state solution is given by:
h = C1x + C2
where C1 and C2 are constants that are dependent on BC’s.
 6-7
Effects of hydraulic conductivity
Let’s go back to the example with 1st type BC’s. The graph
shows the h(x) profile at t = 120 min. In this case, K= 10-5 m/s.
If K = 10-6 m/s, how would the profile look like at t = 120 min?
How about K = 10-4 m/s?

h=1 h=0
1
h (m)

0.5

0
0 20 40 60 80 100
x (m)

Suppose the middle part of aquifer is filled with a material

having a smaller K, say K2 = K1/2. At steady state, how will
the h(x) profile look like?

h=1 K1 K2 K1 h=0
1
h (m)

0.5

0
0 20 40 60 80 100
x (m)
 6-8
Bulk hydraulic conductivity h = hin h = hout
Water is flowing through a box
at Q (m3/s). We are not given
Q Q
the information of the material.
We use Darcy’s law to assign
the bulk hydraulic conductivity Area A
(Kb) of the box.
L
h −h A∆h
Q = Aq = − AKb out in = − Kb
L L
QL
∴ Kb = −
A∆h
Kb is dependent on the property of sediments and how they
are arranged in the box. Let’s examine two important cases.

(1) Parallel layers

∆h ∆h
Q = − A1K1 − A2 K 2 A1 K1
L L Q
∆h A2 K2
− A3 K 3
L A3 K3
K1 A1 K 2 A2 K 3 A3
∴ Kb = + + L
A A A

Kb is given by a weighted arithmetic average of each material.

The layer thickness serves as a weighting factor. Thicker
layers have a heavier influence on Kb than thinner layers.
 6-9
(2) Serial layers
h
In this case, we need to have the
same value of q in all layers.
∆h1 ∆h ∆h
q = − K1 = − K2 2 = − K3 3
L1 L2 L3
K1 K2 K3
Head drop in each layer must add up
to the total head drop ∆h.
∴ ∆h1 + ∆h2 + ∆h3 = ∆h L1 L2 L3
−1
 L L L 
∴ K b =  1 + 2 + 3 
 K1L K 2 L K 3 L 
Kb is given by a weighted harmonic average of each material.

Example
Suppose
A1 = A2 = A3 = A/3 L1 = L2 = L3 = L/3
K1 = K3, K2 = 0.01 K1
What is the Kb for parallel and serial cases?
 6-10

Anisotropy of hydraulic conductivity

In layered sediments, the value of Kb varies with direction.
This is called the anisotropy of hydraulic conductivity.
Note that heterogeneity at a small scale appears as anisotropy
at a larger scale.
Darcy’s law in anisotropic material is given by:
∂h ∂h ∂h
qx = −Kx qy = −K y qz = −Kz
∂x ∂y ∂z
Majority of shallow unconsolidated sediments have horizontal
strata, and their hydraulic conductivity is higher in horizontal
direction (Kx and Ky) than in vertical direction (Kz).
 6-11

Similarity between Darcy’s and Ohm’s Law

We saw that the Kb is the arithmetic average for the parallel
case, and the harmonic average for the serial case. Does this
remind us of something? V
V
R1
R1 R2

R2
I = V/R R = L/(AKE) KE: electrical conductivity

Electricity and groundwater obey the same form of equation.

Darcy’s law Ohm’s law Fourier’s law
dh dV dT
q = −K i = −KE qH = − K H
dx dx dx
i : current density qH : heat flux density

Laws like Darcy’s law are called constituitive equations or

phenomenological laws. In physical sciences, we combine
them with the conservation equation:
∆S
Qin − Qout =
∆t
This concept is the foundation of many scientific disciplines.
 6-12
Two-dimensional flow equation
y1
qx

y0
y x0 x1
x
qy
b
∆x = x1 - x0 ∆y = y1 - y0

Suppose a confined aquifer having a constant thickness (b).

We can analyze the mass balance of a box in this aquifer in a
way similar to the analysis of one-dimensional flow.
We saw that Min - Mout in x-direction is given by (see p.6-1):
∂  ∂h  ∂  ∂h 
[Min − Mout ]x = A∆xρw K
 x  = b∆y∆xρ w  Kx 
∂x  ∂x  ∂x  ∂x 
Similarly, Min - Mout in y-direction is given by:
∂  ∂h  ∂  ∂h 
[Min − Mout ]y = A∆yρw  y 
K = b∆x∆yρw  K y 
∂y  ∂y  ∂y  ∂y 
Storage change is (see p.6-2):
∆M = Vbox ρwSs∆h = b∆x∆y ρwSs∆h
Putting them all together into Min - Mout = ∆M/∆t,
∂  ∂h  ∂  ∂h  ∆h
b∆y∆xρw  Kx  + b∆x∆yρw  K y  = b∆x∆yρwSs
∂x  ∂x  ∂y  ∂y  ∆t
[Min – Mout]x [Min – Mout]y ∆M/∆t
 6-13
Dividing both sides by b∆x∆yρw
∂  ∂h  ∂  ∂h  ∂h
 Kx  +  K y  = Ss [6-4]
∂x  ∂x  ∂y  ∂y  ∂t

This is the 2-D groundwater flow equation. At steady state,

Eq. [6-4] becomes:
∂  ∂h  ∂  ∂h 
 Kx  +  K y  = 0 [6-5]
∂x  ∂x  ∂y  ∂y 

Graphical solution of 2-D flow equation

Suppose a two-dimensional, Cross section
confined aquifer of homogeneous,
isotropic sediments. North and south
boundaries of the aquifer are
Plan view
impermeable, and east and west
boundaries are connected to rivers.
Water levels in rivers are constant.
How does hydraulic head change
from the west to east?
50 m
h = 10 m
h = 15 m

x 0
0 100 m
 6-14

12
15

14

11

10
13
Observe:
(1) Flow lines are normal to contour lines.
(2) Contours meets the impermeable boundaries at 90º.
These are common features of the steady-state flow in
isotropic aquifers.

Water level in groundwater wells indicate hydraulic head in

the confined aquifer. Joining the well water levels, we can
define an imaginary surface of hydraulic head. The contours
in the flow diagram above show the shape of the surface.
 6-15
Equipotential and flow line
The potentiometric surface is an imaginary surface defined
by the water level in wells (hydraulic head, h) in a single
aquifer. Like any other surface, we can draw contours of
constant h, called equipotentials.
74
y
potentiometric surface
77 x
81
73

sand
78 h = 1075 m
impermeable clay 82
h = 1080 m
From Darcy’s law:
∂h ∂h
qx = −Kx qy = − K y
∂x ∂y
In vector form, we can write,
(qx, qy) = (-Kx∂h/∂x, -Ky∂h/∂y)
If the material is isotropic (Kx = Ky), we can just use K and
(qx, qy) = K(-∂h/∂x, -∂h/∂y)
i.e. the flow direction is parallel to the gradient vector.
 ∂h ∂h 
 − −  What does this vector represent?
 ∂x , ∂y 
 6-16

Groundwater flow direction is is normal to equipotentials

when Kx = Ky.

Hydraulic conductivity of anisotropic

materials varies with direction, and flow
lines may not be normal to equipotentials.
 6-17

Flownet construction Cross section

Let’s go back to the analysis of the

confined aquifer. This time it has Plan view
somewhat irregular shape.
Hydraulic head is 10 m in the west
river and 7 m in the east river.

Can we draw equipotentials?

Remember the flow is at steady state.

Qin Qout
Is q constant throughout the aquifer?

50 m
imper
m eable
h = 10 m

h=7m

x 0
0 impermeable 100 m
 6-18
Equipotentials and are shown below. Note that they meet the
impermeable boundary at 90º. Why?

Can we draw a few flow lines?

(1) Flow lines are normal to equipotentials.
(2) Spacing between flow lines?
l1

b1

b2

l2
9.5

8.5

8.0

7.5

7.0
9.0
10

7.5

7.25
The strip between two flow lines is called a “flow tube”. It
is customary to draw flow lines so that each flow tube has
the same Q. Suppose a horizontal layer in the aquifer having
a thickness (normal to the page) of 1 m. The flow rate
through a 1-m thick tube is given by:
dh bi
Qi = −(1)bi K = K∆h where ∆h = 0.25 m
dl li
If we chose the spacing so that bi = li, then Q in each tube is
equal to K∆h.
Flow nets constructed this way provide a useful tool for the
analysis of steady-state flow in homogeneous and isotropic
materials.
 6-19

8.5
9.5

8.0

7.5

7.0
9.0
10

Above is an example of a properly constructed flow net.

Suppose that K = 10-4 m/s. We can say that:
(1) There are four and half flow tubes.
(2) Each tube carries Q = 0.25 × 10-4 m3/s.
(3) 1-m thick layer has a total flow rate of 1.1 × 10-4m3/s.
(4) If the aquifer thickness is 5 m, the total flow rate will be
5.5 × 10-4 m3/s.

Effects of heterogeneity
Can we construct a flow net for this case?

50 m
K1 = 10-3 m/s
/s
m

h = 10 m
h = 15 m

3
10 -
×
3

y
0.
=
2
K

K1 = 10-3 m/s
x 0
0 100 m
 6-20
Flow lines and equipotentials are still normal to each other.
However, heterogeneity creates some features that were not
seen in the homogeneous case.
(1) Flow lines refract at the zone boundaries.
(2) Equipotentials are much denser in the low-K zone.
14

13
15

10
11
12

Groundwater takes the “paths of least resistance” by taking

the shortest path through the low-K material. The shortest
path through the low-K zone is achieved by flowing straight
across the zone.

How will the flow net look like in this case?

50 m
K1 = 10-3 m/s
h = 10 m
h = 15 m

y
K2 = 5 × 10-3 m/s

x 0
0 100 m
 6-21
In this case, the paths of least resistance are achieved by
channeling the flow through the high-K zone.

15

14

13

12

11

10
K1 < K2
K1
K2

h
Tangent law (optional)

+
h
Note that, at steady state, the flow q1
volume (Q = Aq) crossing the A1 L1

h
interface is equal on both sides. θ1
Let’s say q1 and q2 are the specific θ2 A2 q2
discharges along flow directions. L2
From mass balance principle,
A1q1 = A2q2 ∆h ∆h
∴ A1K1 = A2 K2
L1 L2
where ∆h is the head difference between the two contours.
Note that contours are continuous across the interface.
K1 L1 / A1 tanθ1
∴ = =
K2 L2 / A2 tanθ2

K2 is greater than K1, which means θ2 must be greater than θ1.

The flow lines refract.