CIVL2006 Soil Mechanics

Topic 3: Seepage and Flownets

Dr. Fiona Kwok

Room 521, Haking Wong Building
The University of Hong Kong
Tel: (852) 2859 2655
Email: fiona.kwok@hku.hk
 3.1 Fluid Flow in Soils

Relevant Readings:
Knappett & Craig (2012): Chapter 2.1-2.2
Atkinson (2007): Chapter 6.3, 14.1-14.2
Fluid Flow in Soils
 As mentioned in Topic 1, saturated soil is a two-phase material, comprising of soil particles
and pore water. Pore water has a significant influence on the overall mechanical behaviour
of the soil mass, and all geotechnical theories must take into account of fluid flow through
its pores.
 The ability of engineers to understand and predict the flow of fluids (usually water) in soils
is essential for many applications in civil engineering.
 Flow of water through soils is called seepage. Seepage takes place when there is
difference in water levels on the two sides of the structure such as a dam or a sheet pile.

Failure of Failure of
earth retaining
dam: wall:
Pore Water Pressure
 Due to the existence of the inter-
connected voids, soils are permeable. The
permeable soils will allow water flow from
points of high energy to points of low
energy (Ancient wisdom: water flows
downhill).
 In general, the water in the voids of an
element of saturated soil will be under
pressure, either due to the physical
location of the soil or as a result of CRT scan of porous network of a typical clay
external forces. sample

 This pressure is the pore water pressure u. It is measured

relative to atmospheric pressure.
hw
 When there is no flow, the pore pressure at depth hw below
the water surface is: u

(Atkinson, 2007)
Hydrostatic States (No Flow)
 The level in the ground at which the pore pressure is zero (equal to atmospheric) is
defined as the water table or phreatic surface. The pore water pressure is really a
"gauge" pressure in that the atmospheric pressure has been subtracted out.
This is the standard
symbol for water
table

hw

At hydrostatic state, standpipe always level with At no flow condition, the standpipe always have
the water table the same water level with phreatic surface,
irrespective of the its depth inserted.

 When there is no flow, the water surface will be at exactly the same level in any standpipe
placed in the ground below the water table. This is called a hydrostatic pressure state.
 The pore pressure at depth hw below the water table is: u
 Below the water table the soil can be considered to be saturated.
(Atkinson, 2007)
Steady Flow States (Seepage)
 In fluid mechanics, Bernoulli’s equation states that for steady non-viscous
incompressible flows, the total (piezometric) head (h) at a point can be
expressed as the summation of three independent components: pressure head,
elevation head and velocity head.

Total head
2 h
Elevation P
Pressure Velocity head z
head head

ARBITRARY DATUM

 To define the elevation head, a reference datum has to be defined. The datum
can be put anywhere, but it is best to put it low down to avoid negative head
values.
 In most geotechnical problems, the seepage velocity can be considered very
small when water flows through the porous soil. Therefore, it is often assumed
that the velocity head is negligible.
 For fully saturated soils, the pressure head derives solely from water. Hence, P
is simply the pore water pressure.
Hydraulic Gradient h

 When water flows through soils, some energy

s
is lost in overcoming the resistance provided by A
the soils. This loss of energy, expressed as B
total head loss (h), is simply the difference in
water levels. ARBITRARY DATUM

 Hydraulic gradient is the total head loss per unit length. When water flows from
Point A to Point B, the total head at A has to be greater than that at B. The
average hydraulic gradient between A and B, is the total head lost between A
and B divided by the length AB along the flow path.
∆
∆

The negative sign is present in the equation to indicate that the flow is always in the direction
of decreasing head. If i is positive, the flow rate will be in the negative s-direction. If i is
negative, the flow rate will be in the positive s-direction.

Standpipes (Piezometers) are devices used to measure the pore water

pressure in the ground. When everything is in equilibrium, the
pressures of water just inside and just outside the pipe are equal.
u
Flow Rules
1. Fluids flow in soils only when there is a gradient in piezometric head (h). Lack
of a gradient in piezometric head implies that fluid is not flowing.
2. Whenever there is fluid flow in soils, there is energy dissipation (i.e. head loss).
3. In soils, fluids always flow down the gradient in piezometric head (i.e. hydraulic
gradient). That is, fluids flow from high energy regions to low energy regions.

1
2

REFERENCE DATUM
h1 > h2
h1 = h2
Fluid flow down the hydraulic gradient from Point 1 to 2
No flow between Points 1 and 2
(seepage not necessarily downhill)
(Swan, 2011)
Flow Regimes
 The overall flow velocity has a close relationship with the hydraulic gradient, and
can be divided into 3 regimes in general.
1. Laminar flow zone (Zone I)
2. Transition zone (Zone II)
3. Turbulent flow zone (Zone III)
 In most soils, the flow of water through the void spaces can be considered
laminar (Zone I): ∝

 In fractured rock, stone, gravel

and very coarse sands,
turbulent flow conditions may
exist.

(Swan, 2011)
Darcy’s Law (1)
 In 1856, Darcy established an empirical relationship for
the flow of water through porous media. In one-
dimensional form:

where k is the coefficient of permeability (also known as hydraulic conductivity)

of the soil which has dimensions of velocity (L/T)
A is the cross-sectional area of the flow
q is the volumetric flow rate
 Darcy’s Law states that the mean velocity of groundwater flow is generally
dependent on 2 factors:
a) the larger the hydraulic gradient (i), the faster fluid will flow.
b) the coefficient of permeability (k) of the porous medium
 The fundamental premise for Darcy’s law to work are:
a) Darcy’s Law is typically valid for values of the hydraulic gradient less than
about 5, and for laminar flow in the voids.
b) The soil is fully saturated.
c) The flow is in steady state. There are no temporary variations.
(Das, 2001)
Darcy’s Law (2)
 Note that v is the mean discharge velocity based on the v
gross cross-sectional area. Macroscopically the flow
path (in 1D) can be considered as a smooth line.
 However, on the microscopic scale, the water seeping
through a soil follows a very tortuois path between the
solid particles. This true seepage velocity (vtrue)
through the void spaces is greater than v.
 Assuming the void ratio for the cross-section is the
same as the volumetric void ratio, the cross-sectional
area of the void is smaller than the total area A by a
factor e/(1+e). Hence,
1
(Note porosity n )

 This means that if you use dye or a tracer to examine groundwater flow you will
measure vtrue, which is not the same as the velocity given by Darcy’s Law.

(Atkinson, 2007)
Permeability (1)
 The coefficient of permeability (k) is a measure of the ease with which water can
flow through the voids between solid soil particles.
 The value of k depends on a number of factors, including the particle size
distribution, particle shape, soil structure. void size, void ratio and the viscosity
of the pore fluid.
 Void size (which is related to particle size) has by far the most significant effect.
In coarse grained soils, the effective grain size D10 has good correlation with
permeability. Hazen (1911) suggested that, for uniform sands (Cu < 5) having D10
of 0.1-3 mm, in its loosest state, k and D10 are related by:
k (m/s) = 0.01D102 (mm2)
 Typical permeability values for the common soil types, are summarized in the
following table. When k is less that 10-6 cm/s, the soil is practically impervious.

Permeability (m/s):

(Craig, 2004)
Permeability (2)
 A similar table of values of k according to USCS group symbol:
Group Symbol Typical coefficient of
permeability (cm/s) The permeabilities of typical
GW 2.5 × 10-2 sands/ gravels can be a million
times greater than those of typical
GP 5 × 10-2 clays!
GM > 5 × 10-7 As the grain size of the soil
GC > 5 × 10-8 decreases, the hydraulic
conductivity decreases
SW > 5 × 10-4
significantly. This is due to the
SP > 5 × 10-4 higher SSA of fine-grained soils.
SM > 2.5 × 10-5
SM-SC > 10-6
SC > 2.5 × 10-7
ML > 5 × 10-6
ML-CL > 2.5 × 10-7
CL > 5 × 10-8
OL ---
MH > 2.5 × 10-7
CH > 5 × 10-8
OH ---
Laboratory measurement of permeability (1)
1. Constant Head Test (for coarse-grained soils)
 Water from a constant head tank flows through the sample in a cylinder. During the test,
the water supply at the inlet is adjusted in such a way that the difference of head
between the inlet and outlet (distance L) remains constant. After a constant flow rate is
established, water is collected in a measuring cylinder (volume: Q) for a known duration
(t). ∆
&
∆
 In practice it is best to vary the rate of flow (q) in stages and plot v against i; in this way
Darcy’s law can be verified with k evaluated.
 The constant head test is unsuitable for fine-grained soils where the flow rates are so
small that evaporation from the measuring cylinder could lead to significant error.

(Powrie, 2007)
Laboratory measurement of permeability (2)
2. Falling Head Test (for fine-grained soils)
 A standpipe (with area a) is connected to a cylindrical sample of fine-grained soil (with
area A). The water drains into a reservoir of constant level.
 The standpipe is filled with water and a measurement is made of the time (t1) for the
water level (relative to the water level in the reservoir) to fall from h0 to h1.
 At any arbitrary intermediate time t (t < t1), the water level is at a general height h, and its
rate of change is given by -dh/dt. And at time t, the head loss between the top and
bottom of the sample is h.
Specimen: ∆

Standpipe:
Integrating between the limits:

ln Filter

 k can be determined by plotting ln(h0/h1) against t and finding the

gradient (equal to kA/al).
 The falling head test is unsuitable for coarse-grained soils where
the flow rates are so large that water level will drop from h1 to h2 Reference Datum

within a few seconds, thus not allowing enough time to take the
measurements properly. (Craig, 2004)
Field measurement of permeability (1)
Well-pumping Test
 In the field, the average hydraulic conductivity of soil deposits in the direction of flow can be
determined by performing pumping tests from wells.
 Water is pumped from one well and the resulting fall in groundwater level can be monitored
at a number of locations spaced radially from the pumping well using standpipes or
observation wells (boreholes).
 Observation wells differ from standpipes in that an observation well is open to the soil (via a
wellscreen and granular filter) over a significant proportion of its length, whereas a standpipe
is open to the soil over only a short length (150-300 mm) at the bottom.
 In general, a standpipe only measures the pore water pressure at a point, whereas an
observation well might measure the maximum value of total head over its open length.
 Pumping from a borehole can be used for deliberate groundwater lowering in order to
facilitate excavation.

Pumping well Observation well Standpipe (Powrie, 2007)

Field measurement of permeability (2)
 In well pumping tests, measurements are taken when steady seepage is established,
radially towards the well, resulting in the water table being drawn down to form a ‘cone of
depression’.
 The flow of groundwater is always horizontal so the head is always the same along any
vertical line.
 In the next slides, we will introduce the principles of well-
pumping tests performed in 2 different site conditions: 1)
confined aquifer and 2) unconfined aquifer.
Note that if a borehole is drilled into saturated clay with a
water table near the ground surface the hole will remain
dry for a considerable time. The reason for this is that clay
has low permeability in general and it will take very long
time for sufficient water to flow from the ground to fill the
borehole.
This means you can only determine pore pressures, and Water
groundwater conditions, from observation in boreholes in table

soils with relatively high permeability. In clays, water

levels may continue to fall for days or weeks after
commencement of pumping. In coarse-grained soils,
steady state conditions may be reached quite quickly.
For clays and soils with low permeability special
piezometers have to be employed. (Powrie, 2007)
Unconfined vs Confined Aquifer
 Unconfined Aquifer: The aquifer is not overlain by impermeable strata and its upper
boundary is the ground water level (GWL). Steady state flow conditions can be
reached quickly.
 Confined Aquifer: The aquifer is bounded at the top and bottom by relatively
impermeable strata. This prevents any water entering or leaving the aquifer in vertical
flow.
 This configuration gives a perched water distribution in the aquifer, with the initial
piezometric level (shown by the standpipes) always remaining above the upper
surface of the aquifer.
Well-pumping Test in Unconfined Aquifer (1)
Assumptions:
a) |dh/dr| <<1 (i.e. the hydraulic gradient is small), flow is therefore approximately horizontal
b) homogeneous soil
c) cylindrical symmetry about axis of pumping well
 The flow area at a general radius r is 2rh, where h = f(r) is the thickness of the aquifer.
 Apply Darcy’s Law: 2

Centreline (Powrie, 2007)

Well-pumping Test in Unconfined Aquifer (2)
 Rearranging the previous equation: 2

 Taking the integration between the limits (r: r to R0; h: h to H):

2
ln

ln

 Alternatively, the permeability k can be determined by plotting (H2 – h2) against ln r

ln ln ln ln

 The slope of the plot will have a negative magnitude of

Note that the plot on the right has natural logarithm on the horizontal axis.
Most semi-log graph papers are on log10 scale. If the plot is made on these graph papers, the
slope S of the line will become: 2.3

(Powrie, 2007)
Well-pumping Test in Confined Aquifer (1)
 The flow area at a general radius r is 2rt, where t is the thickness of the aquifer.
 Apply Darcy’s Law: 2
 At the steady state, q is the known pumped flow rate from the well.
 The drawdown (i.e. the depth of the new piezometric level below the initial level)
decreases with increasing distance from the well. Eventually, at a radius R0, the
drawdown is zero (i.e. the initial GWL is beyond R0 is unaffected by pumping). R0 is the
radius of influence of the well.

Centreline (Powrie, 2007)

Well-pumping Test in Confined Aquifer (2)
 Rearranging the previous equation: 2

 Taking the integration between the limits (r: r to R0; h: h to H):

2 2
ln

ln
2

 Alternatively, the permeability k can be determined by plotting (H – h) against ln r

2
ln ln ln ln
2 2

 The slope of the plot will have a negative magnitude of

2
Note that the plot on the right has natural
logarithm on the horizontal axis.
Most semi-log graph papers are on log10
scale. If the plot is made on these graph
papers, the slope S of the line will become:
2.3
2
(Powrie, 2007)
Example 3.1
A 25 cm diameter well is pumped at the rate of 0.2 m3/min. The aquifer is 30 m thick.
After some time, the well sides cave in and are replaced with a 20 cm diameter tube.
The drawdown is 2 m. The water table recovers its original thickness 750 m from the
well. What will be steady flow from the new well?

q = 0.2m3/min (25 cm dia. well)

q = ??? (20 cm dia. well)

s
r

s0 = 2 m
H = 30m h

R0 = 750 m
Example 3.1 Solutions
Write down all the available information on the diagram.
Identify the type of aquifer and think through how to attempt the question.

Unconfined aquifer
25
Radius of the pumping well: rw   12.5 cm
2
Water depth at centre of pumping well: : h0  30  2  28 m

Apply well-pumping equation for unconfined aquifer:

q  R0  0.2  750 
k ln    ln  
 ( H 2  h 2 )  r   (30 2  282 )  0.125 
= 0.004774 m/s

20
For the relined new well, rw   10 cm
2
Apply well-pumping equation to find the new steady flow rate
q  R0  q  750 
k ln   0.004774  ln   q = 0.195 m3/min
 (H  h )  r 
2 2
 (30 2  282 )  0.1 
Example 3.2
The permeability of a semi-impervious soil was evaluated in a falling-head permeameter whose
head decreased from 100 to 38 cm in 5 min, The body diameter was 13 cm, the standpipe
diameter was 0.3 cm, and the sample length was 5 cm. What was the permeability of the soil?
Example 3.2 Solutions
Be clear of which permeability test the question is referring to. And think deeply from your
mind the mechanisms and calculation procedures of the associated test.

Falling-head test

Refer to the figure on the right:

  0.32
a  0.0707 cm2
4
 132
A  132.7323 cm2
4
Filter
l = 5 cm
h0 = 100 cm
h1 = 38 cm
t1 = 5 min = 300 s

Apply the equation for falling-head test: Reference Datum

al h0 0.0707 5  100  6
k ln  ln   8.6 10 cm/s
At1 h1 132.7323 300  38 
3.2 Flow Nets
Two-dimensional flow – Laplace’s Equation
 Laplace's equation indicates that any imbalance in flows into and out of an
element in the x direction must be compensated by a corresponding
opposite imbalance in the y direction.
 For a rectangular element with dimensions dx, dy and unit thickness, in the x
direction the velocity of flow into the element is

the negative sign being required because flow occurs down the hydraulic
gradient.
 The velocity of flow out of the element is

 Similar expressions can be written for the y direction.

 Balance of flow requires that

and this is Laplace's equation.

Flow Nets
 Graphical solutions to Laplace's equation for two-dimensional seepage can be
presented as flownets.
 Seepage is from high hydraulic head to locations of low hydraulic head.
 Decrease in hydraulic head along the flow path.
 Two orthogonal sets of curves form a flow net:
equipotentials connecting points of equal total head h
flow lines indicating the direction of seepage down a hydraulic gradient
 There can be no flow along an equipotential, because there is no hydraulic
gradient, so there can be no component of flow across a flow line.
 The flow lines define channels along which the volume flow rate is constant.
 Flow nets are usually constructed through trial-and-error sketching.
 The two-dimensional drawing of the system must be in scale.
Flow line - Paths along which water can flow through.
Equipotential line - Lines of equal (potential) energy level. As the water flows through
the pore space its energy is dissipated by friction, where equipotential lines act like
contours to show how energy is lost
 If standpipe piezometers were inserted into the ground with their tips
on a single equipotential then the water would rise to the same level in
each standpipe.
 The pore pressures would be different because of their different
elevations.
Boundary Conditions
 A surface on which the total head is fixed (for
example, from the level of a river, pool,
reservoir) is an equipotential.
 A surface across which there is no flow (for
example, an impermeable soil layer or an
impermeable wall) is a flow line.
 For the situation shown, with flow occurring
under a sheet pile wall, the axis of symmetry
must also be an equipotential.
Flow nets are constructed according to the
following rules
 Once you have defined the boundary conditions, start trial sketching
of flow lines and equipotentials, following the rules below.
 Flow lines enter and leave pervious surfaces perpendicular to those
surfaces.
 Flow lines are parallel to the flow direction.
 Equipotential lines are drawn perpendicular to flow lines such that
the resulting cells are approximately square and the intersections are
90 angles.
 Try to make the flow net consist of curvilinear "squares“.
Theoretically, it should be possible to draw a perfect circle within
each cell that touches all four boundaries, even though the cell is not
actually square.
 Typically, 3-5 flow channels are sufficient.
 Keep sketching and refining until you have a good set of "squares"
which satisfies the boundary conditions.
Suggested procedures for drawing flow nets:
1. First identify boundary conditions (Which boundaries are
impermeable? Which are constant head?)
2. Next think: where is water entering the system? Where can it
leave?
3. Always look for any symmetry in the boundary conditions.
4. Decide on the number of flow tubes you want to use.
5. Draw a trial flow line and then draw in other flow lines to define all
the flow tubes; some trial-and-error sketching may be necessary
6. Where flow tubes constrict, higher head gradients (more closely
spaced equipotentials) are needed to move the same quantity of
water through the flow tube.
7. Fit together the curvilinear squares by drawing in the
equipotentials.
8. As you do this, you may have to revise the positions of some of the
flow lines. Trial-and-error is the order of the day.
Seepage flow into a trench Cross section diagram of trench, which
describes flow. This drawing is called a flownet.
Calculation of Flow
 Consider an element from a flow channel of length L
between equipotentials which indicate a fall in total
head h and between flow lines b apart.
∆
 The average hydraulic gradient is
and for unit width of flow net the volume flow rate is
∆

 There is an advantage in displaying or sketching

flownets in the form of curvilinear 'squares' so that
a circle can be inscribed within each four-sided
figure bounded by two equipotentials and two flow
lines.
 Then and ∆
so the flow rate through the flow channel is the
permeability multiplied by the uniform interval h
between equipotentials.
 Calculation of Total Flow
 For a complete problem, the flownet has been drawn with the overall
head drop h divided by the number of head drop Nd ∆
 Determine the head at the left-most and right-most equipotentials and
subtract them to get h, the total head difference across the net.
 Then for each channel
 With the number of flow channels Nf ,the total flow rate per unit
width is


 It is usually convenient in sketching flownets to make Nd an integer.

 Nf then may or may not be an integer.

It does not matter how many Nf you get, as Nd will be adjusted

accordingly if you follow the rules and the ratio of Nf / Nd will
remain unchanged.
 Draw a very simple flow net

Flow
h1 h2

Your classmates may get a different Nf and Nd values,

but the ratio of Nf / Nd should be similar.
Flow Beneath a Sheet Pile Coffer Dam

h1 h9

q5

q4
h2 q3 h8

h3 q2 h7
h4 h5 h6
q1
1. Determine boundary equipotentials and flow lines:
a. lines kb (h1) and hl (h9) are equipotentials because they are at constant
elevation and have constant water depth above them.
b. lines mn (q1), be (q5) and he (q5) are flow lines because they are
impermeable surfaces
2. Determine boundary heads: head along lines kb and hl are equal to
elevation of boundary above the datum plus depth of water above the
boundary: 1 60 30 90 9 60 5 65
3. Sketch equipotentials and flow lines
4. Find total ∆h along flow lines (e.g., along q2):
Δ 1– 9 90 – 65 25
5. Determine Nd by counting number of equipotential drops (squares) from h1
to h9: 8 squares are traversed going along q2 from h1 to h9 so 8
6. Determine hd, the head drop across a square :
Δ / 25 /8 3.125
Thus 2 86.875 3 83.750 4 80.625 5 77.500 6
74.375 7 71.250 8 68.125
7. To determine pressure at point e:
elevation of e: ze = 30 ft
point e lies on equipotential h5
5 / so 5– where 62.4 / 3
62.4 / 77.5 – 30 2964 / 2
Example 3.3: Past Paper 2010 Fall Q2a
 An impermeable dam impounds water over a soil with properties as shown.
The width of the dam across the water flow is 300 m. Estimate the quantity
of seepage under the dam by means of flow net construction.

1
Example 3.3 Wrong Answer (1)
Example 3.3 Wrong Answer (2)
Example 3.3 Wrong Answer (3)
Example 3.3 Solution (1)

 The flownet shows three flow tubes ( 3) and 11 head drops

( 11). The entire flow is
 /
 2 10 6 18.1 3.1 3/11 300
 2.45 10 3 3/
Example 3.4 Uplift pressure
The section through a dam spillway is shown. Determine the quantity of
seepage under the dam and plot the distribution of uplift pressure on
the base of the dam, and the net distribution of water pressure on the
cut-off wall at the upstream end of the spillway. The coefficient of
permeability of the foundation soil is 2.5x10-5 m/s.
 Example 3.4 Solutions (1)
 The ground surface is selected as datum.
 The total head loss is 5 m.
 3 flow channels and 10 equipotential drops

 2.5 10 5
 2.5 10 3/
Example 3.4 Solutions (2)

 The pore water pressures acting on the base of the

spillway are calculated at the points of intersection of
the equipotentials with the base of the spillway.
 The total head at each point is obtained from the
flownet, and the elevation head from the section.
Example 3.4 Solutions (3)

 The water pressures acting on the cut-off wall are

calculated on both the back (hb) and the front (hf) of
the wall at the points of intersection of the
equipotentials with the wall.
 The net pressure acting on the back face of the wall is
 )
Quick Condition and Piping
 If the flow is upward then the water pressure tends to lift the soil element.
 In the regions of strong upward flow – for example in front of the retaining walls
– the soil may fluidize or boil if the uplift force due to seepage exceeds the
weight of the soil.
 This is the quicksand condition if it occurs over a large area and is associated
with localized piping instabilities around excavations.
 Fluidization takes place when the upward hydraulic gradient exceeds a critical
value, icri, the critical hydraulic gradient.
 Considered the forces acting on a block of soil, uplift will just occur when the
upward force due to the pore water pressure acting on the base
begins to exceed the weight of the block of soil
 So
 γ
 Or / /
 For soils with 20 / 3 2 , 1
 Factor of safety, /
1) initiation and first deterioration 2) regressive erosion

3) formation of flow channel 4) liquefaction and collapse

Piping occurs when weight of clay is just balanced by water pressure

Example 3.5
Use the flownet below to investigate the instability of the floor of the
excavation. If the is a potential of instability, suggest two ways to
overcome it.
Example 3.5 Solutions
 Does the upward hydraulic gradient i exceed icrit (assume here
1) below the floor of the excavation?
 In the flownet shown, the embedment of the sheet pile below
ground level is 7m.
 The head drop between the toe of the sheet pile and the floor of
excavation is 8m.
 An average hydraulic gradient Δ / Δ in front of the wall =
8 /7 1.14
 So fluidization is quite likely to happen.
 The problem could be overcome by
 Driving longer sheet piles into the soil to form a complete cut-
off
 Lowering the groundwater level inside the cofferdam