M04 Permeability and Seepage

CVNG 2002 Soil Mechanics
M04 Permeability and Seepage
Page 1 of 54
THE UNIVERSITY OF THE WEST INDIES

ST AUGUSTINE, TRINIDAD AND TOBAGO, WEST INDIES
FACULTY OF ENGINEERING
DEPARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING
CVNG 2002 Soil Mechanics Module 4. Permeability and Seepage

Lectures by Richard Dean, Semester 2, 2006
Flownet for embankment dam Casagrande solution (Sarsby, 2000)
Flownet for flow into an excavation (www.ejge.com/iGEM/f-excav.gif)
Falling head permeameter for laboratory testing, UWI
Analysis of acquifer and well performance in map view (Ferris et al, 1992; http://en.wikipedia.org/wiki/Flownet )
\\Civilserver02\temp\course notes 2005\CVNG 2002 (SoilMechanics)\CE22B RDEAN 2005\M04 Permeability and Seepage.doc
Page 2 of 54
Contents
What is permeability and seepage?.............................................................................................................. 3 How water flows through soil ........................................................................................................................ 6 The concepts of pressure head, elevation head, and total head .................................................................. 9 Darcy's Law................................................................................................................................................. 11 Values of permeability................................................................................................................................. 13 Wells and boreholes.................................................................................................................................... 15 Fluid flow and effective stress ..................................................................................................................... 18 Permeability for layered soils ...................................................................................................................... 21 Re-visiting the constant head permeameter ............................................................................................... 23 The falling head permeameter .................................................................................................................... 25 Permeability and Poisson's equation .......................................................................................................... 26 Flowlines, equipotentials, and flownets....................................................................................................... 28 Constructing a flownet................................................................................................................................. 30 Extracting engineering information from flownets....................................................................................... 32 Flow through earth dams ............................................................................................................................ 40 Teton dam ................................................................................................................................................... 46 Further aspects of flownets ......................................................................................................................... 48 Review Questions ....................................................................................................................................... 50 Previous Exam Questions........................................................................................................................... 51 References and further reading .................................................................................................................. 53 Software ...................................................................................................................................................... 54
Page 3 of 54
What is permeability and seepage?

Seepage is the flow of fluid through soil. Usually the fluid we are interested in is water, and this module will focus mainly on the flow of water through soils. Permeability is a measure of the ease with which this flow or seepage can occur. Sometimes engineers need to consider other fluids, such as oil, and sometimes engineers are called to deal with problems associated with contaminants that get into groundwater. This module will touch on these subjects but only briefly. Let us start at large scale, by considering the Water Cycle. Water is continually moving around, with evaporation from streams and storage basin like likes, seas, and oceans; with evapo-transpiration from plants. Water is stored in the atmosphere as humidity, and condenses into cloud formations. Winds transport the water, which will eventually precipitate out as rain, hail, or snow. Precipitation that falls on high ground may simply run-off into streams, ponds, or rivers, or may seep or infiltrate into the ground. As groundwater, water may flow downhill through the soil. There may be some discharge upwards as springs or seepage into river beds, and there can be further infiltration by seepage from riverbeds back into the ground. The water will eventually return to lakes, seas, or oceans and the cycle will continue.
Page 4 of 54
Various processes can happened to water as it travels down from where it is precipitated to where it is stored again in lakes, seas, or oceans. In the sketch below, there is some ponding on the higher ground because the rate of rainfall is greater than the rate at which the water can infiltrate or seep into the soil. There is some runoff, and some of the runoff later infiltrates into the soil. There is also some flow through a permeable layer of sand or gravel. The water takes some time to seep or permeate through the layer. People may sink wells lower down to extract the water for purification and drinking water supplies, or for other uses such as cooling water in industrial processes. precipitation
infiltration
Run-off
Seepage flow
well
Civil engineers may be called up to direct the flow of water in some way. For example, engineers may be asked to divert a river, by constructing new riverbanks. The following picture shows Wolf Creek dam, Nashville which is a combination rolled earth fill and concrete gravity structure. It has a maximum height of 258 feet above founding level. The concrete section is part of a hydro-electric plant.
Picture from www.lrn.usace.army.mil/pao/issues/WOLcommo/seepage.htm
Page 5 of 54
Engineers need to be able to assess the effectiveness of a dam before it is ever built. Amongst other things, engineers may be called upon: to investigate the soils around and beneath a proposed dam site to select the materials to be used for an earth dam, to determine the compaction requirements for the earth fill to predict the amount of water that seeps through the dam.
Another example is a sheet pile wall which may be used as a dam, as shown (simplified) at left in the following sketch, or as part of the supporting walls of an excavation, as shown at right. Sheetpile wall Water table
Water table
For the excavation, there will be a pumping system to remove water from the base of the excavation and hold the water table a short distance below the base. There will be some seepage underneath the wall, which the engineer must assess in order to provide advice about selecting the right size of pump. Too much seepage, or seepage that is not controlled properly, can cause problems for construction operations inside the excavation. It can also seriously jeopardize the stability of the wall, as well as the safety of people working in the excavated area. Civil engineers are also involved in tasks relating to the quality of people's lives. The following picture shows a driveway and part of the front garden of a house that is located at the bottom of a hill. During the rainstorm water soaked into the ground in the hill above the driveway. It flowed downward and horizontally along permeable layers, downhill toward a creek at the bottom of the hill. Unfortunately, the driveway had been dug deep enough into the ground so that it cut into the permeable layer of soil that carried the underground water downhill. Consequently, the ground-water seepage came to the surface.
Page 6 of 54
Picture and information from www.ga.water.usgs.gov/edu/gwdriveway.html
How water flows through soil

Water can flow through soil because of the presence of voids between particles, and because the voids between different particles are generally connected. In the following sketch, the water is flowing on average into the paper. The particles A, B, and C shapes that do not fit tightly together, and a void is created even though the particles are in contact at points a, b, and c. Water can flow through the void.
c Particle A void b
Particle B
Particle C
The situation is three-dimensional, and the water will flow from one void out and straight into another which may have a different shape. In the following sketch, the average flow is from left to right. If we consider water molecules that start at points 1, 2, and 3, they follow tortuous paths through the voids formed by the matrix of soil particles.
Page 7 of 54
Tortuous paths of water flowing through the connected pores of a soil matrix; the particle are in contact but this is a plane section through the soil; the plane section intersects only a few contacts; the actual path of the water molecules will also be three-dimensional If the flow is sufficiently slow, then laminar conditions occur, and the flow has some similarities with laminar flow through pipes. This means that the rate of flow between two points is proportional to the pressure difference between those points, taking account of differences in elevation. The flow will be faster for soils with less tortuous paths, and for soils with larger voids. Thus the sands is much greater than the permeability of clays, and the permeability of a loose sand, with a large void ratio, is greater than the permeability of the same sand after it has been compacted to a lower void ratio. Because of tortuous paths, the actual length of the path travelled by any given water molecule can be considerably longer that the straight line distance. Different tortuous paths may have different lengths, so molecules that start off at the same time at points 1, 2, and 3 on the left may emerge on the right at quite different times. The dimensions of the void spaces along different tortuous paths may be different, so the speeds that the molecules travel may be different. And different tortuous paths may emerge at different places. All these features mean that molecules that start off close together may emerge separated by distance and time. Thus there is dispersive flow. In many situations, engineers are simply interested in average flows for instance average flow rates so we often ignore dispersion. In special situations we cannot do this for example if we are looking at the way contaminants in the water flow through soil.
Page 8 of 54
What drives the average flow? If we think of a small region around a single tortuous path, it is a little like a long, windy tube. We know from fluid mechanics that the flow through a tube is affected by the viscocity of the fluid, which provides resistance to the flow. The flow is driven by a pressure difference between the ends of the tube, from high pressure to low pressure. In engineering we need to be careful about hat we mean by "pressure". In the following situations, there is no flow from A to B, even though the pressure at A is higher than at B. Water table
situation of a tube in a water tank
situation of an imaginary tube in soil
In soil mechanics we think of the pressure u in the pore water as composed of two parts. One part is the equilibrium pore water pressure ueq. The second part is the excess pore pressure uxs: u = ueq + uxs (1)
The equilibrium pore pressure is the pore pressure that would occur in a situation of no flow. The excess pore pressure is the pressure over and above the equilibrium pressure. Flow can occur through soil whenever there is (a) a difference between the excess pore pressures at two points, and (b) a hydraulic connection between those points. In the above pictures, the equilibrium pressure is higher at A than at B, but this does not cause flow. There is a hydraulic connection between A and B, because water molecules could in principle flow through the void spaces in the soil from A to B. But the excess pore pressures at A and B are zero, so there is nothing to drive the flow. Excess pore pressures are important in consolidation theory, but for seepage we normally think in terms of pressure heads. These are explained in the next section, where we also look at Darcy's law for seepage through porous media such as soils.
Page 9 of 54
The concepts of pressure head, elevation head, and total head

Engineers find it useful to express water pressure in terms of "head". For a particular location in a soil body, the pressure head is the water pressure divided by the unit weight of water. It equals the height to which a column of still water would rise above the location. The elevation head is simply the height of the point above some datum. The total head is the sum of the pressure head and the elevation head. In the situation shown below, the gravel layer is sealed above and below by clay. Let us suppose that the gravel is filled with water. If we put tall pipes at A and B, water would rise to the same levels at the two points. If we define a datum level C, then the elevation heads at the two points are different, and the pressure heads are different, but the total heads are the same. There is no flow of water.
Pressure head, hA, at A
Elevation head at A
Pressure head, hB, at B
C
Clay
Heads in a situation of no flow Gravel
Elevation head at B
The total head at A is the height of the water surface above A, above the datum level C. It is the same as the total head at point B. Note that the water pressures at points A and B are the unit weight of water, w, times the pressure heads hA and hB. The situation of no flow would occur whether hA was greater than hB, or less. So the difference in pore pressures between A and B do not cause fluid flow. Now let us alter the situation, by drilling a borehole through the clay at point B. If there was no friction or viscosity of the water, water could flow through the gravel and up the borehole, reaching a height equal to the original pressure head above point B. However, the gravel offers a lot of resistance to the flow. In the following situation, the total head at A is the same as before, represented by the height of
Page 10 of 54
the upper dotted line above level C. But the total head at B is now smaller, represented by the height of level D above level C. Hence there is now a difference between the total heads at A and B.
Pressure head hA at A
A Excess head at A
hAD
Level D
C
Heads in a situation of flow
Pressure head hB at B Elevation head at B
The difference in total heads at A and B is called the excess head. In the present situation we can either say that there is a positive excess head at A, meaning that the upper dotted line is higher than the lower one, or we can say there is a negative excess head at B, meaning that the lower dotted line is lower than the upper one. Both viewpoints result in the same conclusion: flow occurs from A to B. How can we reconcile this physical understanding with our previous conclusion that flow through pipes is driven by differences in pressure? One way is like this. First, we define an equilibrium pore pressure. The equilibrium pore pressure at some point is the pore pressure that is due to the depth of that point below some reference level. For example, if the reference level is D, then the equilibrium pore pressure at point B would be simply the unit weight of water w times the pressure head at point B; Equilibrium pore pressure at B: ueq,B = w.hB (2)
However, the equilibrium pore pressure at A would be negative since point A is above level D. It would be w times the negative of the height hAD of A above D: Equilibrium pore pressure at A: ueq,A = w.hAD (3)
We might think of this as the notional suction at A that would apply if enough water had flowed out of the gravel to re-establish a new equilibrium situation with the water surface everywhere at level D. Now the actual pore pressures at A and B are w.hA and w.hB respectively, so using our previous equation 1:
Page 11 of 54
Excess pore pressure at A: Excess pore pressure at B:
uxs,A uxs,B
= =
uA ueq,A uB ueq,B
= = 0
w.(hA+hAD)
(4) (5)
Comparing these two results, we can see that the difference between the excess pore pressure is just w times the difference of levels between the dotted lines in the sketch, ie. it is w times the excess head at A. This shows that the two viewpoints are equivalent we can say that flow is driven by difference in total heads, or that flow is driven by difference in excess pore pressures.
Darcy's Law
Henry Darcy lived in France in the 19th century and was responsible for developing several empirical equations for fluid flow, and some devices such as the Pitot tube. In soil mechanics, Darcy's law describes seepage flow through soils. The sketch below shows the essential features of Darcy's experiments, which are today reproduced in a laboratory permeameters. We normally set up a permeameter vertically, but it does not have to be vertical. Darcy experiments encompassed many different arrangements, and by this he was able to develop a law that was true for all set ups. Tap
Total head at A, HA
Flowpath length L Total head at B, HB Porous stone Measurement cylinder
Cylinder of soil, cross-sectional area A Porous stone datum
Key features of the measurement of permeability with a constant head
Page 12 of 54
On the left, there is a water reservoir that is kept topped up by a tap and an overflow pipe. Water can flow freely down a tube to one end of a cylinder of soil that is held between two porous stones. The water flows out across a porous stone at the other end of the tube, and into a second reservoir whose level is kept constant by means of an overflow pipe. The overflow falls into a measurement cylinder. Using this and a stopwatch, we can measure the quantity of water passing through the soil per unit time. The difference in total head across the soil body is the difference in levels between the two reservoirs. Let this difference be h = HA - HB. We can measure the rate of volume flow Q, as a volume per unit time, and relate it to h. By doing experiments with tubes of different cross sectional areas A and different lengths L, we find that the flow rate can be related to the total head difference h approximately as follows:
Q A
k.
h L
(6)
where k is a constant for a particular soil sample. The approximation gets better and better if we use porous stones that are thinner and thinner. So, we deduce that the above relation would hold exactly if the porous stones were infinitely thin. That means the above equation describes the flow through the soil, and the approximations are die to effects of the porous stones (which we look at later). The constant k is called the permeability of the soil. It is not affected by the orientation of the measuring cylinder, or the area or length of the cylinder. To calculate its dimensions, we note that Q/A has units of volume / time divided by area, ie.velocity, and h/L is dimensionless. Hence permeability has units of velocity. Q/A is called the seepage velocity v, and h/L is called the hydraulic gradient i. We can re-write Darcy's equation as: v = k.i (7)
The seepage velocity is less than the actual velocity of the water molecules, because the cross sectional area A includes cross-section through particles as well as through voids. If the voids ratio is e, then the average interstitial velocity is (1+e) / e times the seepage velocity: average interstitial velocity of a water molecule = v.
1+ e e
v n
(8)
where n = e/(1+e) is the porosity of the soil. We use the interstitial velocity if we want to know how fast molecules travel through soil. For example, if some contaminant is dumped onto soil, we would use interstitial velocity to calculate how fast the contaminant would spread through the soil. We would also
Page 13 of 54
need to consider the effects of dispersive flow, which would men that some contaminant molecules will travel faster than others. But for most situations, we are interested in average, volume flows, so we use the seepage velocity. Darcy's Law can also be written in terms of excess pore pressures. In the sketch on page 11, a condition of no flow would occur if the tap for the inlet tank on the left was turned off, and water was allowed to flow through the system until equilibrium was reached at the level of the top of the overflow pipe for the outflow tank on the right. With reference to this level, the equilibrium pore pressures at the left ends of the cylindrical soil sample are: Left: Right ueq ueq = = w.(HB height of left end above datum) w.(HB height of right end above datum) (9) (10)
In the actual experiment, the pore pressure at the right end of the soil sample is the same as the equilibrium pore pressure there, so the excess pore pressure at the right end is zero. On the left, the total pore pressure u will be: Left: u = w.(HA height of left end above datum) (11)
The excess pore pressure is then the total pore pressure less the equilibrium pore pressure, so: Left: uxs = u ueq = w.(HA HB) = w.h (12)
Now this value is the difference between the excess pore pressures at the left and right ends of the soil sample, so we should strictly call it uxs. Using this to substitute for the excess head in equation 6 gives: v =
k u xs . w L
(13)
We call uxs/L the excess pore pressure gradient. This form of Darcy's law is useful for consolidation calculations, and also to examine how the flow of fluids in soils affects the effective stresses. We look at the effect on effective stress later in this module. We look at consolidation in Module 6.
Values of permeability
The permeability of the soil does not depend on the orientation of the apparatus, but it does depend on some properties of the soil and fluid. If we test two different samples of the same soil, with the same fluid, but with one sample packed with a low voids ratio and the other not packed so much, so with
Page 14 of 54
a higher voids ratio, then we find that the permeability is less for the same with the lower voids ratio. If we test two samples of the same soil with the same void ratio, but use water in one test and a viscous oil in the second test, then the permeability is less for the oil, due to its higher viscocity. Craig (2005) defined an "absolute permeability" K, by the equation: k =
K. w
(14)
where is the viscocity of the pore fluid. K has units of length-squared. In principle it depends only on the soil, and not on the fluid. Other people have proposed different quantities, such as "intrinsic" permeability. Pereability is sometimes called hydraulic conductivity (eg.Winterkorn and Fang, 1991). It is a measure of the ease with which a fluid can flow through soil. Typical values for the flow of water through different soils are listed below. Permeability, m/sec 1 101 102 103 104 105 106 107 108 109 1010
Clean gravel
Clean sands, sand-gravel mixes
Very fine sands, silts, clay-silt mixes Sands Silty sands, coarse silts
Un-fissured clays and clay-silts with >20% clay Silts, clayey sands, loess Intact clays, sandy clays
Gravels, coarse clean sands
Fissured materials (dominated by flow through fissures rather than flow through soil per se) High permeability Medium permeability Low permeability Very low permeability Practically impermeable
Ranges of permeabilities, and descriptive terms (BS 8004, Craig 1996, and Lambe and Whitman 1979) The values increase logarithmic from clays to silts to sand to gravels. This is expected since the sizes of the voids through which the fluid passes is roughly proportional to the sizes of particles. Hazen's law for sands is that permeability is proportional to the square of the effective size D10 as follows:
Page 15 of 54
2 D10 m/s 100
(15)
where k is in m/sec and D10 is in mm. Hazen's equation can give a good first estimate for "clean" sands. For silty or clayey sands the law breaks down because the voids between sand particles get filled with silt or clay particles which restrict the flow. To get an idea of what these numbers mean, consider a hydraulic gradient of 1. Typical values of hydraulic gradient are not usually greater than 1 in the field, for reasons described later. Then, using Darcy's law and the table: For a clean sand of permeability 103 m/sec, the seepage velocity is 1mm/sec when the hydraulic gradient is 1. If the porosity is 50%, a molecule will take on average about 8.3 minutes to travel through a metre of sand at this hydraulic gradient For an unfissured clay of permeability 109 m/sec, the seepage velocity is 1 nanometre/sec when the hydraulic gradient is 1. If the porosity is 50%, a molecule will take on average about 109/2 seconds to travel a metre, equivalent to about 16 years, at this hydraulic gradient.
Wells and boreholes

If a borehole is drilled into permeable, water-bearing strata, such as water-bearing gravels or sands, and if the borehole is pumped out, water will flow from the ground into the borehole. The rate of flow depends on the permeability of the ground and on geometrical features such as the depth of the water-bearing strata and the diameter of the borehole. Thus a knowledge of the permeability and geometry should allow predictions to be made about the rate at which water can be extracted from an aquifer. Conversely, by carrying out pumping tests, one can infer the permeability of the ground. Wells can also be used as part of a de-watering plan for an excavation. One of the common arrangements for well-pumping tests is shown in the following sketch. There is a water-bearing stratum that contains the water table, and which overlies an impermeable layer. A well is drilled and pumped out at a volume flow rate of Q m3/sec. Observation wells are placed at radii such as r1 and r2, and the heights h1 and h2 of the water levels in the observation wells are measured when steady-state conditions are achieved.
Page 16 of 54
Arrangement 1 for well pumping tests, after page 35 of Craig (2005). The layer of soil contains the water table (WT) and overlies an impermeable stratum. To analyze this arrangement, an assumption is made that the flow of water towards the well is radial, with no vertical component of seepage velocity. The hydraulic gradient at a general radius r is taken to be the gradient dh/dr, where h is the height to which water would rise above the impermeable stratus if a standpipe was placed at radius r. This assumption is called the Dupuit assumption, and is reasonably accurate except at points near the well. Consider the internal area of a cylinder of radius r and height h above the impermeable layer. The area is 2r.h. The volume flow rate Q across this area is given by the product of the seepage velocity and the area. Using Darcy's law together with the Dupuit assumption, the volume flow rate is: Q = 2r.h.k.
dh dr
(16)
Under steady state conditions, the same rate of flow will occur across the areas at al radii, so Q will not be a function of radius. One can the solve the equation by taking dr /r over to the left, giving: Q.
dr r
2.k.h.dh
(17)
Integrating both sides between radii r1 and r2 where the heads are h1 and h2 gives: Q =
2 h2 2 h1 .k. ln(r2 / r1 )
(18)
Page 17 of 54
Thus a field measurement of permeability can be made, by measuring the flow rate Q under steady state conditions, and the heads h1 and h2 at radii r1 and r2. In principle we might expect that the calculation could also be done the other way round prediction of the volume flow rate given the radii and heads. In practice the equation does not give good answers for volume flow rates. However, it is worth doing an exploratory calculation. Suppose that the material is clean sand with a permeability of 103 m/sec. Suppose also that r1 corresponds to a borehole radius of 20cm. Suppose that in the absence of pumping, the water table is at a height of 5m above the impermeable stratum, and that when pumping starts, it is desired to draw the water down so that h1 = 2m. If we suppose that h2 is negligibly different from 5m at radius greater than r2=10m, say, then: Q =
.(103m/s).
(5 2 0.2 2 )m 2 ln(10 / 0.2)
20 liters/sec
(19)
This is a high rate of pumping. If we had assumed that h2=5m at radius r2=5m instead of 10m, the calculation would have given: Q =
.(103m/s).
(5 2 0.2 2 )m 2 ln(5 / 0.2)
24 liters/sec
(20)
Hence there is not much difference between the results assuming h2=5m or h2=10m. For this reason, the analysis using Dupuit's assumption is often used in practice, at least as a first estimate. A slightly different approach is also needed if the water-bearing stratum does not extend to the soil surface. In the following sketch, the water bearing stratum is of height H, and is located between two impermeable strata. Other conditions are similar to previously. In this case, an equation similar to equation 8 applies, but the height of the water-saturated soil is not h but H. Hence equation 8 becomes: Q = 2r.H.k.
dh dr
(21)
Re-arrangement gives: Q.
dr r
2.k.H.dh
(22)
Integrating both sides between radii r1 and r2 where the heads are h1 and h2 gives: Q = 2.k.
H.(h 2 h1 )
ln(r2 / r1 )
(23)
This result gives a way of estimating permeability in the field, or of estimating volume flow rate if the permeability and geometry of the field layers and water table is known.
Page 18 of 54
Arrangement 2 for well pumping tests, after page 35 of Craig (2005). The layer of soil of height H is below the water table (WT) and underlies and overlies an impermeable stratum.
We can also use these equations to obtain a very approximate estimate of the time needed for the water level in a borehole to return to the level of the water table. This is important because, when drlling a borehole in a site investigation, the drillers will use pumps or other methods to control the water level in the borehole. To get an idea of where the water table is under normal conditions, it is a unusual procedure to return to the borehole 24 hours after drilling and pumping operations have ceased, and to measure the level of the water table then. In the above cases, the pumping rate to achieve the drawdown to h1=2m was around 20 litres/second, or 20 x 103 m3/sec. For a borehole of diameter 0.2m, the crosssectional area is .0.12 = 0.03 m2. Hence the velocity flow rate up the borehole required to sustain that drawdown is of the order of 20 x 103 / 0.03 m/sec = 0.6m/sec. Thus for a drawdown of 3m (from h2=5m to h1=2m), the water table would be re-established I the borehole in a matter of half a minute or so when the pump was switched off. However, if the material was silt, with a permeability 1000 times lower, the flow rate would be 1000 times slower, and the water table might not be re-established till next day.
Fluid flow and effective stress

We saw that Darcy's law relates seepage velocity to excess pore pressure gradient. We also saw that excess pore pressure is part of the net pore pressure u, and we know from Terzaghi's principle of Effective Stress that the net pore pressure is part of the total stress. For instance, the vertical total stress
v is the sum of the vertical effective stress and the pore pressure, so using equation 1:
Page 19 of 54
/ v + (ueq + uxs)
(24)
In many situations, the total vertical stress is known as a function of the weight of soil above a particular level. Then, if the excess pore pressure increases, the effective stresses will decrease. In general this will reduce the factor of safety if the soil is sand. An example is shown in the following sketch. Fluid flows from the high water table at the left of the sheet pile wall, to the lower water table at the right of the wall.
Sand stratum P Flowpath
h z
At the top left, the water percolates vertically downwards. It then moves under the wall, and upwards towards the surface on the right. Consider some point P in the soil, at a depth z below the surface on the right. Then the total vertical stress there is w.h + bulk.z where bulk is the bulk unit weight of the soil. The equilibrium pore pressure is the value due to the pressure head (h+z). Hence the vertical effective stress at that point is:
/ v
( v ueq ) uxs
(bulk w).z uxs
(25)
The quantity bulk w is called the submerged unit weight, or the buoyant unit weight, denoted as . Thus if there is no flow, the vertical effective stress is just z. However, suppose flow occurs, and that there is an upwards hydraulic gradient i in the vicinity of the point. This means that the excess pore pressure
Page 20 of 54
gradient is w.i, and so the excess pore pressure at a depth z below the surface is w.i.z. Hence the vertical effective stress is:
/ v
(bulk w).z uxs
( i.w).z
(26)
This means that the vertical effective stresses in the soil would reduce to zero if the hydraulic gradient reached the value /w. This is called the critical hydraulic gradient. For fully saturated soils, it equals (Gs1)/(1+e) where e is the void ratio. If bulk unit weight is in the range 1.6 to 2.1 times the unit weight of water, the critical hydraulic gradient is in the range 0.6 to 1.1. What happens if the water left on the left is initially low, and is then raised? The rate of flow underneath the sheet pile wall will obviously increase. The seepage velocity will increase, so from Darcy's law the hydraulic gradient will increase. The effective stresses at the point P will therefore decrease. The Mohr's Circle will therefore generally move leftwards, towards the failure line corresponding to the angle of friction of the sand. At some stage, a little before the critical hydraulic gradient is reached, there will be local failures within the soil near P. Pipes and channels develop where the soil happens to be looser than average, and as this occur water tends to flow towards these channels. The soil has the appearance of "boiling". The phenomenon is called "piping" or "fluidization". Ground surface
Piping; chaotic upwards flow through channels and pipes that continually move Fluidized soils are sometimes called quicksands. Because the effective stresses are very low, the strength of the soil is also very low. One implication of this is that if you stand on a fluidized bed, you will sink about half-way into it. As you sink, there will be a buoyancy effect. You only sink half-way because you are sinking into a fluidized material that is roughly twice the density of water. Because of the chaotic nature of the piping phenomenon, with pipes continually moving, fluidized beds are good ways of mixing things. They are used in the chemical industry for this purpose.
Page 21 of 54
Another implication of the low strength is that the wall may fail. This is because the wall is held in place by the soil at left and right. If the soil at the right loses its strength, the wall can fail by moving or toppling rightwards. We look at the failure of sheet pile walls in Module 7 Earth Pressures.
Permeability for layered soils

In many engineering situations, we need to calculate flow involving several soil layers with different permeabilities. Two examples are parallel flow and series or normal flow.
LAYER 1 LAYER 2 LAYER 3 LAYER 4 (a) parallel flow
LAYER 1 LAYER 2 LAYER 3 LAYER 4 (b) series (normal) flow
To analyse parallel flow, it is convenient to consider the simplified situation shown below. There are N soil layers, with the nth layer having a height hn. The layers are held between thin porous stones, and each layer experienced the same water pressure difference left-to-right across it. The layers are all the same length L, so they all experience the same hydraulic gradient i.
h
impermeable boundary
n-th layer, permeability kn thickness hn
Simplified situation for the analysis of parallel flow
Page 22 of 54
Suppose the layers all have the same width w in the direction into the paper. Then for the nth layer, the cross-sectional area is w.hn. If the volume flow rate through the layer is Qn, and if the permeability of the layer is kn, then from Darcy's law, neglecting the effects of the porous stones:
Qn w.h n
kn.
h L
kn.i
(27)
Taking the areas to the right side, then adding the volume flow rates for all layers together to get the net flow rate Q = Qn, gives: Q = w
N k n .h n .i n=1
(28)
Dividing by the net cross-sectional area A = (w.hn) then gives:
Q A
kparallel.i
with:
kparallel
k n .hn
n=1 N
hn
n=1
(29)
This shows that in parallel flow, the overall permeability kparallel is the weighted average of the permeabilities of the individual layers. The quantity Q/A is the average seepage velocity. Its relationship to the interstitial velocities in the individual layers is more complicated than for a single layer. To analyse series flow, it is convenient to consider the different simplified situation shown in the next sketch. There are N soil layers, with the nth layer having a thickness hn. We consider a section of width w and unit depth into the paper. Because the flow is normal to the layers, the same flow rate Q/A occurs for each layer, through this width w and unit depth. Suppose the excess pore pressure difference across the nth layer is uxs,n. Applying Darcy's law gives:
Q A
k 1 .u xs,1 w .h1
k n .u xs,n w .h n
(30)
Solving each equation for the excess pore pressure difference in terms of the volume flow rate gives: uxs,n = w.
Q hn A kn
(31)
Adding up all the excess pore pressure differences gives the net excess pore pressure difference across the N layers:
Page 23 of 54
uxs
(uxs,n)
w.
Q A
kn n
n=1
(32)
Re-arranging this into the same format as Darcy's Law gives: w
u xs,n w
thickness hn
n-th layer, permeability kn
Simplified situation for series flow
Q A
k series u xs . h n w
with:
1 k series
kn n
n=1 N
hn
n=1
(33)
This shows that in series flow, the inverse of the overall permeability kseries is the weighted average of the inverses of the permeabilities of the individual layers. The seepage velocity Q/A is the same for each layer. The interstitial velocities can be different, depending on the porosities of the layers.
Re-visiting the constant head permeameter

We can now do an exact analysis of the constant head permeability test in the sketch on page 11. Let us suppose that the porous stones each have individual thicknesses Lp, and each have permeabilities kp. If the soil has permeability k and the cylinder has a length L, then using the equation 33 for series flow:
Page 24 of 54
Q A
u xs / w
(L / k ) + 2.(L p / k p )
(34)
Putting v = Q/A as the seepage velocity, and re-arranging slightly, gives:
Q A
k u xs . L w
1 k.L p . 1 + 2 . k p .L
(35)
Now if the length of the porous stone is small compared to the length of the soil sample, and if the permeability of the porous stone is high compared to the permeability of the soil, then the factor in large brackets will be close to 1. However, if the porous stone is too thick, or if its permeability is not high compared to that of the soil, the factor can be significantly smaller than 1. This explains why, in the original analysis, equation 6 was only approximately correct. The lengths Lp of the porous stones are easy to determine. One way to measure their permeabilities is to carry out the constant head test with no soil in the cylinder. We would the n use equation 35 to analyses the no-soil data, but with k replaced by a near infinite value to represent the permeability of the cylinder with no soil in it. Putting k=infinity in the equation gives:
Q A
kp w
u xs 2.L p
(36)
Having found the permeability of the porous stones, it is then possible to solve equation 35 to obtain the permeability of the soil based on the experimental data. Another way to infer the permeability of the porous stones can be developed by using equation 35 to calculate the quantity uxs /(w.v). This gives:
u xs / w v
Lp L + 2 k kp
(37)
Therefore, if we do several tests using different lengths L of soil, taking care to prepare the soil the same way each time, then we can plot the quantity on the left versus the length L. The plot should be a straight line that intersects the uxs /(w.v) axis at a positive value. Using the above equation, we can deduce the soil permeability from the slope of the straight line, and the value of 2Lp/kp from the intersection.
Page 25 of 54
The falling head permeameter

In the falling head permeameter, the upstream side is replaces by a burette or tube. If the tube is marked with volume gradations, a separate measuring cylinder is not needed at the outlet end of the device. The following sketch shows a falling head permeameter arranged vertically.
Burette, crosssectional area Ab h
Porous stone
Soil sample, crosssectional area A
Porous stone Falling head permeameter arranged vertically The head difference across the soil sample and porous stones is represented by the height h of the meniscus in the burette above the outlet pipe. As water flows through the device, this height reduces, so the head across the stones and pipe falls. The stones and soil are in series, and equation 37 applies. uxs is equal to w times the excess head difference h across the porous stones and sample. v is the seepage velocity through the sample whose cross-sectional area is A, so the surface of the water in the tube falls at a rate of A/Ab times this. Hence:
dh dt
1 A h k.L p k . . Ab L 1 + 2. k p .L
(38)
Taking dt over to the right, and h over to the left, gives:
dh h
1 A k k.L p . 1 + 2 . Ab L k p .L
.dt
(39)
Integrating the left side gives ln(h). Integrating the right side gives a linear function of time. Re-arranging
Page 26 of 54
the results then gives:
Lp L + 2 k kp
A.(t 1 t 2 ) / A b ln(h1 / h 2 )
A.(t1 t 2 ) / A b 2.3log 10 (h1 / h 2 )
(40)
where t1 and t2 were the times at which the heads across the sample were h1 and h2 respectively, and ln is the Napierian logarithm. Hence by measuring the heads and times and dimensions, we can either do tests at different lengths L to infer the soil permeability k, or we can infer 2Lp/kp from a test without any soil, and then infer L/k from the test with the soil.
Permeability and Poisson's equation

In practical situations, the in-situ permeability of a soil is usually larger in the horizontal direction than the vertical. This anisotropic permeability is thought to be due to the way that particles pack together during the process of deposition and subsequent consolidation under load. However, if we take a sample from the field and transport it to the lab, we usually destroy the "fabric" of the way that particles are arranged because the sample will normally be a bag sample. One way to avoid too much disturbance is to freeze the ground before cutting a sample. In two dimensions, x horizontally and z vertically, we can often write separate Darcy equations involving horizontal permeability kx and vertical permeability kz. In differential form, these equations are: Vx Vz = =
u xs x u kz. xs z
kx.
(41) (42)
where vx is the seepage velocity in the +x direction, and vz is the seepage velocity in the +z direction. In the following sketch, the rectangular element of soil lies between coordinates x and x+x laterally, where
x is a small distance, and between z and z+z vertically, where z is a small distance. The seepage
velocities into and out of the element are shown. The net inflow in the x direction can be calculated by subtracting the outflow across the boundary on the right, from the inflow across the boundary on the left: Net inflow along vertical boundaries =
v x x x
times boundary dimension z
Similarly, the net inflow in the z direction can be calculated by subtracting the outflow across the low boundary from the inflow across the upper boundary
Page 27 of 54
x vz z
x+x
vx z+z
vx +
vx .x x
Vz +
vz .z z
Inflows and outflows for an element of size x by z and unit depth into the paper
Net inflow through horizontal boundaries =
v z z x boundary dimension x z
The total inflow, per unit depth into the paper, is the sum of the net inflows: Total inflow per unit depth =
v x v z + z. x z x
If the element is fully saturated and if its total volume remains constant, the total inflow must be zero. Then, using the Darcy equations 41 and 42 to substitute fro the seepage velocities, and assuming the horizontal permeability kx does not vary with horizontal coordinate x, and that the vertical permeability kz does not vary with vertical coordinate z, gives: kx.
2 u xs x2
+ kz.
2 u xs z2
(43)
This is Poisson's equation for an anisotropic medium in the x, z directions, with permeabilities kx and kz which do not vary with position (x,z). Poisson's equation also appears in other branches of physics, such as electricity and heat conduction. The equation has been very extensively studied, and there are many solutions and solution methods available in the literature. One of the solution methods is the method of flownets, which we will
Page 28 of 54
look at shortly. Since the equation applies in several branches of physics, we can also use experimental methods from one branch to solve practical problems in another branch. For instance, flow of electrons through metal plates satisfies Poisson's equation, so we can do electrical experiments to find empirical solutions to soil mechanics problems. Poisson's equation also provides a way of resolving the problem of anisotropy of permeability. Suppose we define a new variable x such that x = x. k x / k y . Putting this into the above equation gives, after simplification:
2 u xs ( x ) 2
2 u xs z2
(44)
This is the isotropic version of Poisson's equation, using coordinates {x,z} instead of {x,z}. It means that, if we find solutions to the isotropic equation, they can generally be applied to the real, anisotropic situation by means of a transformation of coordinates.
Flowlines, equipotentials, and flownets

A flowline is the path that an average water molecule will take as it passes through a body of soil. The following sketch shows some flowlines for the case of flow underneath a sheet pile wall.
Flowline
IN-1
OUT-1
Equipotential
Flow underneath a sheet pile wall, from Lambe and Whitman (1979) page 269
Page 29 of 54
If we looked at a single molecule, it would not flow like this, because of the dispersive nature of flow at microscopic scale. But on average, a molecule that enters the soil at position IN-1 will exit at position OUT-1, for example. The flowlines are curved, and they do not intersect each other. They can be measured in a laboratory by placing a small model in a glass tank. We can then release dye at positions like IN-1, and we can watch the dye flow though the soil. The dye spreads a bit as it passes along a flowline, but it also contracts in regions where the flowlines are close together. The dye-front moves faster through these regions compared to regions where the flowlines are widely spaced. In the method of flownets, we also use equipotentials. An equipotential is a curve or surface where the excess head has a unique value. It is a contour of excess head or of excess pore pressure. The following sketch shows some of the equipotentials that are involved for the sheet pile wall. If we put a series of standpipes along a given equipotential, the water columns would all rise to the same level.
Standpipes along an equipotential Darcy's law implies that fluid flows from places of high excess pore pressure to places of low excess pore pressure. If the permeability is isotropic, then a flowline is essentially the shortest path from one equipotential to another. This means equipotentials are at right angles to flowlines.
Page 30 of 54
The shape formed by a set of flowlines and equipotentials is a flownet. There are an infinite number of flowlines and an infinite number of equipotentials, but we normally draw only a few, and in a special way. In the above diagram, many of the individual elements of the flownet are approximately square. It turns out that the process of drawing curved squares has some advantages. The next section goes through an example of how it can be done, and then uses the flow net for certain calculations.
Constructing a flownet
The following figure shows steps in the construction of a flownet for flow underneath a weir with a sheet pile cutoff. The first step in the construction is to draw the extreme flowlines and the extreme equipotentials. The extreme upper equipotential is the base of the water reservoir. The extreme lower equipotential is the soil surface downstream. The extreme flowline close to the barrier follows the base of the barrier and goes around the cutoff wall. Another extreme flowline starts theoretically at infinity leftwards, comes along the top of the impermeable stratum, and then rises at theoretically infinite distance rightwards. The next step is generally to sketch in one or two trial intermediate flowlines, remembering that flowlines cannot cross each other, and must intersect equipotentials at right angles. The third step is to sketch in some trial intermediate equipotentials, crossing the flowlines at right angles, and forming approximate squares with the flowlines. The next step is to make judgments about where improvements are needed. In the example shown, curved squares are judged to be good if a circle drawn inside them can ouch all four edges. All intersections between flowlines and equipotentials need to be checked to see if they are at 90. The method then continues by modifying the first flownet to make some of the improvements. Then there is another judgment stage, and so on. Once an approximate flownet is obtained with a small number of flowlines and equipotentials, it is feasible to sketch in flowlines between the flowlines that have been drawn, and equipotentials between the equipotentials that have been drawn. If the coarse flownet is reasonable good, then the finer flownet will also be good.
Page 31 of 54
Steps in the construction of a flownet for a barrier with an in-ground cutoff wall (a) extreme lines (b) intermediate flowlines (c) equipotentials and judgement stage 1 (from p.137 of Sarsby, 2000)
Page 32 of 54
Extracting engineering information from flownets

We normally get several types of information from flownets. We can calculate rates of flow of water through the net. We can calculate excess pore pressures, and so make estimates relating to effective stress and the possibility of either settlement or failure. We look for possible locations of quicksand conditions. We can also use flownets to help improve a design, and to identify possible design flaws or hidden assumptions.
Calculating volume rates of flow

Engineers need to calculate volume flow rates through soil fro several reasons. For a dam, the engineer needs to know how much water will flow through and under the dam in order to verify that the dam will indeed hold back water, and also to determine the size of drainage ditches that may be needed downstream to take the seepage water away. For an excavation, the engineer will need to know what capacity of de-watering pump is needed, and this will depend partly on the precipitation, and partly on the rate of flow through the soil surrounding the excavation, under the walls, and up through the soil at the base of the excavation. To do this calculation, we use a special property of square flownets. The sketch below shows the curved square of one such flownet. The average width between different flowlines, w, is the same as the average distance between the two equipotentials.
Head loss h1
w equipotential flowline
equipotential
flowline
Curved square of a flownet
Page 33 of 54
Suppose that the head loss between the two equipotentials is h1. Then the hydraulic gradient for flow through the element of the net is h1/w. Hence the average seepage velocity through the element is k.h1/w. Notice that this means that the seepage velocity will be highest where the spacing w between flowlines is smallest. This agrees with experimental observations of flow along flowlines. Considering a unit depth into the paper, the cross sectional area across which the fluid flows is (w x 1). The volume follow rate Q1 for the single element of the flownet is the product, so: Q1 = k. h1 (volume flow per flow channel per unit depth into the paper) (45)
This is independent of the size w of the element. Hence all elements between any two neighboring equipotentials have the same volume flow going through them. If there are Nf flowlines, there will be Nf1 flow channels, and each of them will have the same volume flow rate, given by the above equation. Hence the total volume flow rate Q is just Nf1 times the flow rate through one of the flow channels. Q = k.h1.( Nf1) (volume flow per unit depth into the paper) (46)
Consider now the flow-channels between next pair of equipotentials, as shown in the following sketch. Head loss h1 Head loss h2
equipotential
flowline
equipotential
equipotential flowline The fluid that flows out of the previous flow-channels flows straight into these new flow channels. Hence the total volume flow must be the same, which means from the above equation that the head difference
Page 34 of 54
h2 between the new pair of equipotentials must be the same as the head difference h1 between the previous pair. So, if there are Ne-1 equipotential intervals, between Ne equipotentials, then the head loss h1 across any neighbouring equipotentials is just the total head loss h divided by Ne-1. Hence: Q = k.h.
Nf 1 Ne 1
(volume flow per unit depth into the paper)
(47)
This means that we can calculate the volume flow if we know the permeability, the total head loss, and the numbers of flowlines and equipotentials. As an example, let us calculate the volume flow rate underneath the barrier sketched on page x, supposing that the barrier holds back a 5m head of water. The third flownet still needs some improvement, but it is good enough for a first estimate. There are 3 flow-channels, so Nf=4. There are 9 equipotential intervals, so Ne=10. If the material is ands with a permeability of, say 103 m/sec (page 14), then the volume flow rate will be 103 x 5 x 3 / 9 m3/sec per metre length into the paper = 1.7 litres/sec per metre length into the paper. If the barrier was 1km long, then the engineer would need to design drains big enough to carry away 1700 litres of water per second from the downstream, "dry" side of the barrier. On the front cover of this module, the second flownet is for an excavation, which might be an excavation for a basement, say. The head loss across each of the retaining walls is 2m. There are 3 flow channels and 6 equipotential drops around each wall. If the material is sand with a permeability of 10-3 m/sec, the total inflow per wall is 10-3 m/s x 2m x 3 / 6 = 1 litres per second. If this was a 10m x 10m basement excavation, the total circumference is 40m, and a first estimate of the net inflow rate would be 40 x 1 = 40 litres per second. In practice this would indicate either that the sheet pile walls should be continues deeper, blocking off some of the flow and so reducing the volume flow rate to more manageable amounts, or that a substantial de-watering scheme would be needed during construction.
Calculating pore pressures under steady flow conditions

Engineers need to be able to estimate pore pressures because they need to be able to estimate effective stresses. Effective stresses are needed in order to be able to estimate frictional resistance through the Mohr-Coulomb failure equation, which is used to analyse the factor of safety against failure. Since a flownet contains equipotentials, it allows us to estimate the excess pore pressures everywhere, and so to deduce the net pore pressures. An easy way to do this is to "count down" from the upper extreme equipotential, or "count up" from the lower extreme equipotential. In this instance we have
Page 35 of 54
to do both, because the flownet has not been drawn quite right. The nets left and right of the centerline are fine, but the "squares" between the equipotentials from points D and E are not square at all. If we count the total number of equipotential drops from the upstream to the downstream side, we get 13. To take account of the rectangles in the centre, Lambe and Whitman (p.271) estimated that there were an equivalent of only 12.6 drops, with the drop between D and E being 0.6 of the other drops.
Illustration of a calculation for the pore pressure heads along the base of a concrete gravity dam, from page 271 of Lambe and Whitman (1979) The upstream water elevation is 28.2m above datum, and the upstream water level is at elevation 20.4m. Hence there is a head loss of 28.2 20.4 = 7.8m across the dam. If there are 12.6 equipotential drops in the flownet, each drop is 7.8 / 12.6 = 0.62m head. The upper extreme equipotential is the soil surface at the upstream side of the dam, which has elevation 19.2m. Counting down from this to point B,
Page 36 of 54
there are about 4 equipotential drops. Hence there is a head loss of 4.5 x 0.62 = 2.8m between the upper extreme equipotential and B. Hence if a standpipe was to be place at point B, the water would rise to an elevation of 28.2 2.8 = 25.4m. Point B is at an elevation of 18m, so the pressure head at B is 25.4 18.0 = 7.4m. This value is marked on Lambe and Whitman's figure. If we start from the least equipotential, we would have to add equipotential drops. For instance, consider point H. It is three equipotential drops away from the downsteram soil surface, where the downstream water level is at elevation 20.4m. Hence if a standpipe was placed at H, the water would rise to elevation 20.4 + 3x0.62 = 22.3m. The point H is at elevation 12.9m. Hence the pressure head at H is 22.3 12.9 = 9.4m. This is marked on Lambe and Wnitman's figure. To illustrate how these calculations can be useful, suppose that the strength of the cutoff walls is quite small, and can be neglected to a first approximation. Let us calculate the force equilibrium of the dam, ignoring for simplicity the 1.2m embedment of the dam (which can be quite significant however).
Thrust on dam face from upstream reservoir , 245 kN/m Mobilized angle of friction, mob
Weight of dam, say 2200 kN/m length Uplift from water pressures beneath dam, 1581 kN/m
Simplified force equilibrium for the dam (not to scale) The lateral force from the water pressure on the upstream face of the dam is equal to the force due to a head of 28.2 29.2 = 9m of water, or approximately 90 x 9 / 2 = 245 kN per metre length of dam.
Page 37 of 54
Suppose the dam weighs 2200 kN per metre length into the paper. The average pressure head beneath the dam is about 6.2m head, so the net uplift force over the 25.5m width is 6.2m x 10 kN/m3 x 25.5 = 1581 kN per metre length. Hence the total effective force is 22001581 =619 kN per metre length. We can then calculate the mobilized angle of friction of tan1(245/619) = 21.6. If the friction angle of the soil is 30, then the factor of safety is tan30/tan21.6 = 1.46. This factor of safety is a lttle low, though can be acceptable provided other information is know with confidence. One way to increase the factor of safety would be to make the concrete dam more substantial, simply by adding weight to it. Another way would be to embed the dam indeed, we ignored the effects of embedment (these will be covered in module 7 Earth Pressures). We could also strength the sheet pile cutoffs so that they provided some additional contribution to sliding resistance.
Identifying locations of potential quicksand (piping) conditions

Quicksand conditions are potentially possible wherever there is an upwards hydraulic gradient. For the case just examined, the upwards hydraulic gradient occurs around the downstream toe, in the region marked X. The following sketch shows this region in more detail.
El.19.2m
El.12.9m
The equipotential drop here is 0.62m between the penultimate and the lower extreme equipotential. The equipotential intersects the sheet pile cutoff at an elevation of about 16.6m, which is 2.6m below the downstream soil surface. Hence the hydraulic gradient near the wall is about 0.6/2.6 = 0.23. If the critical hydraulic gradient is 1 (depending on the soil density), there would be a factor of safety against piping in this region of about 1/0.23 = 4.
Page 38 of 54
The factor of safety would be adequate, but does indicate that there would be a significant reduction in the strength of the soil here. In fact, we would get a value slightly slightly less than 4 if we were to do the calculation over for the last three equipotential drops, from point H to the downstream soil surface. The reason is that the equipotentials are more closely spaced near H, so the hydraulic gradient is much higher there. In practice this means that, if a piping problem starts, it would start at H rather than at the surface. If the material near H fluidized first, that would affect both the flownet itself and the stability of the material in region X. Mitigating measures can be provided for potential quicksand conditions. For the sheet pile wall shown below, gravel could be dumped to provide extra vertical stress. For earth dams, one solution is to provide a permeable berm that stabilizes the soil by its weight. Similarly, a small permeable rock-dump at the downstream toe of the dam considered above would be a relatively cheap way of increasing the factor of safety against piping there. It would also have the advantage of providing additional sliding resistance.
Gravel or rock dump
Berm
Some solutions for possible piping problems
Improving designs, identifying hidden assumptions, and what if? scenarios

Once a flownet has been drawn and some consequent information and calculations performed, it may be that some improvements are needed. For instance, the flow rate may be excessive, or the factor factor of safety against sliding or piping may be too small. By looking at the original flownet. One can sometimes identify the best places to make changes to resolve the problem. The following sketch shows an initial dam design, without sheet cutoffs, followed by two experimental flownets designed to find out what the effect of sheet cutoffs would be on the flow rates and the uplift pressures on the dam.
Page 39 of 54
Exploring the design of a dam, from page 272 of Lambe and Whitman (1979)
Page 40 of 54
In this instance, the cutoff walls do not make much difference top the flow rate, because the three flownets actually have the same number of flowlines in each and about the same numbers of equipotential drops (12, 14, and 14 respectively). However, the option with an upstream cutoff increases the number of equipotential, drops between the upper extreme equipotential and the start of the base of the dam. This significant reduces the uplift pressures on the base of the dam, so increasing the factor of safety against sliding. The option with the downstream cutoff reduces the upwards hydraulic gradient at the toe of the dam, but actually increases the uplift pressures on the dam base. We can also use this kind of sketch to identify some hidden design assumptions, and to run "what if" scenarios. The above sketches all show impervious layers at some depth below the dam. Thus our design is critically dependent on the assumption that these layers exist, and we will need to check, for example, that these soils are not cracked fissured clays or fissured rocks. We must also check that the construction of the dam does not itself cause cracking of the impervious layer due to the dam's weight, or due to the weight of water than is being imposed on the foundations on the upstream side. Examples of other what if questions could include: What will happen if the sheet pile cutoff corrodes away? What control measures are needed during construction to ensure that good quality cutoffs are installed, to the appropriate depths? What if a gravel berm is provided by members of the public take the gravel away to make driveways for their houses? Should there be some way of protecting the berm? What if a colony of rabbits sets up burrows at the downstream side of the dam, and start to dig burrows in the upstream direction? What measures can be installed to prevent this?
Flow through earth dams

A simple type of earth dam consists of an approximately triangular section of soil that has been carefully selected, graded, and compacted. The soil will have been trucked in from some place nearby. A major part of the cost of an earth dam is the cost of the earth itself, and much of the cost of the earth is in the cost of transporting it from the borrow site to site of the dam. Sometimes there is a trade-off between technical quality and economic cost. Thus the engineer will not always be working with ideal materials. The following sketch shows a simple trapezoidal earth dam with a rock doe The toe is designed so that water flow through the dam an into the doe, rather than through the dam to the downstream face. If the downstream side is to be kept dry, there will be a graded filter below the toe, and a drain to take the seepage water away.
Page 41 of 54
Upstream water surface
Flow through an earth dam with rock toe (US Army Corps of Engineers, 1952)
At the upstream side of the dam, the upstream slope is an equipotential, and so the flowlines leave the slope at right angles to it. The upper flowline is actually the water table or "phreatic surface". It is the locus of points where the absolute water pressure equals atmospheric pressure. Before the flownet is drawn, the position of the equipotential is not yet known. This makes some difficulties for the first step in drawing a flownet drawing the extreme flowlines. Two clues are available. This first is due to the fact that the upper flowline is the phreatic surface. Hence if we put standpipes at points along this line, the standpipes would rise simply to the phreatic surface. This means that the head loss between neighboring equipotentials that intersect this flowline has to be equal to the elevation loss along the flowline. In the sketch, there are an estimated 9 equipotential drops. The total head loss is 12m. Hence the equipotentials have to cut the phreatic surface at intervals of 12/9 = 1.33m in height. The second clue is that, near the rock toe, the upper flowline is known to be always approximately parabolic. The method is known as the Casagrande solution, and is described in several text books (eg.Craig, 1976; pp.56 to 61). By using both of these clues together, it is often possible to construct an adequate first-estimate flownet without going into details. The following sketches show various types of embankment dam design.
Page 42 of 54
Embankment dam with downstream filter (Craig, 2005, p.60)
Some options for filter geometries (Craig, 2005, p.60)
Page 43 of 54
Embankment dams with (a) central clay core and chimney drain, (b) grout curtain to reduce underflow, (c) impermeable upstream blanket (Craig, 2005, p.61)
Page 44 of 54
From page 482 of Terzaghi et al (1996)
Page 45 of 54
Simple zoned embankment dam, with different materials in zones 1 and 2. In this case the permeability of the upstream material in Zone 1 is about 5 times less than the permeability of the downstream material in Zone 2 (Craig, 2005, p.65).
Complex zoned dam; built on top of original dam More complicated zoned dam, with staged construction, concret upstream face, upstream grout curtain, and downstream dike/min-berm, from page 495 of Terzaghi et al (1996)
Page 46 of 54
Teton dam
When a dam is constructed the upstream side of the dam may be left dry until the full height of the dam has been completed. Construction can take several years, and during this time there is usually some consolidation of the soil beneath the dam, and some consolidation of the soil that the dam is made of. During this time, the river will be diverted in some way. At an appointed time, the diverted flow is cut off, and the reservoir behind the dam is filled for the first time. For most dams, the first filling goes well, without any problem. In the case of Teton Dam, in Idaho USA, the dam failed shortly after the first filling in 1976. The reservoir was 270 feet deep (at the dam) and drained in less than 6 hours. The failure triggered over 200 landslides around the edges of the reservoir as water drained away (www.usbr.gov/pmts/sediment/ projects/TetonRiver/TetonRiver.htm . The following pictures are on www.americanwhitewater.org
Abutment
Valley wall Slope of downstream face of dam
The brown streak is a debris flow of water and soil that is being eroded through a small erosion tunnel that has opened up due to piping at the toe. According to Terzaghi et al (1996, p.478), the core was a wind-blown silt that was placed against open untreated joints in the dam abutment. Cracks may have been caused as a result of settlement during construction, or due to lateral water pressures on the upstream face of the dam. During first filled these cracks filled with water, totally altering the hydraulic conditions compared to the design conditions. This led to the piping failure which then led to the erosion of material from the core.
Page 47 of 54
A few tens of minutes later The works at the base of the dam are now flooded; the flow has taken on a violent character. Most observers must have known at this point that there would be no saving the new dam.
About an hour later The crest of the dam has now collapsed and the dam has been breached.
Page 48 of 54
A few hours after the breach. Teton dam has now failed.
Further aspects of flownets

In modern times there are computer programs that can draw flownets, and do calculations for stability and piping conditions, see eg the software suppliers listed at the end of this module. However, such programs do not give the physical insights that are obtained from "hand-drawn" flownets, and are more cumbersome to use for design improvements. It was mentioned earlier that permeability is normally anisotropic, and that calculations can be done by coordinate transformation. The following sketch shows an example of this. The lower diagram would be the actual section fro anisotropic conditions. The flownet is constructed by transforming coordinates, drawing an isotropic flownet as in the upper diagram, and then reverse-transforming the flownet back to the real coordinates of the lower diagram. Further details are in Craig (2005). Permeability is not always uniform, because ground is not always uniform. However, the measurement of permeability in the laboratory has the disadvantage, for cohesionless soils such as sands and graves, the soil fabric is completely disturbed and the fabric that occurs in the laboratory permeameter may be radically different from the fabric in the field. Subsequent diagrams show some further examples of flownets.
Page 49 of 54
Transformation of coordinates for calculations with anisotropic permeability, after Craig (2005). Note that flowlines are not at right angles to equipotentials if the permeability is anisotropic
Flow under a barrier with an inclined impermeable surface (Craig, 2004)
Flow beneath a combined berm and sheet pile wall (Craig, 2004)
Page 50 of 54
Review Questions
1. As a consultant engineer, you have been approached by a software company that wants to create and market a computer program to do permeability and flow calculations. Write a brief report to the company, describing the kinds of calculations that engineers would be interested in, and the kinds of inputs and outputs that would be involved. 2. How does water flow through soil? Define the terms seepage velocity and interstitial velocity. If the seepage velocity is 1 cm/sec and the porosity is 40%, what is the interstitial velocity? 3. The following sketch shows a plan for a river-wall designed to hold back a flood tide of 5m. The concrete block is 15m wide and there is 5m deep cutoff wall at its left edge and a 1m cutoff wall at its right edge. Draw the geometry of the situation at large scale, and draw a flownet for the situation of the design flood. Based on the flownet: (a) Estimate the volume flow of water underneath the wall at that time. How will you deal with this water? (b) What is the value of the factor of safety of the wall against sliding failure, assuming a wall weight of 150 tons per metre length (in the direction into the paper), and a friction angle of 30 degrees between the base of the wall and the soil? (c) What is the factor of safety against piping? (d) Identify the main design assumptions that you have made, and describe any improvements or other measures that you may feel appropriate. 15m Water table at ground surface Flood tide 5m CONCRETE BOX, 150 tons/m length 5m deep cutoff wall 1m deep cutoff wall
10m
Fine sand, permeability 104 m/sec IMPERMEABLE BEDROCK
Page 51 of 54
Previous Exam Questions

2000 (July). Question 8
A falling head permeameter has a cross-sectional area of 2000 mm2 and a standpipe area of 200 mm2. Before the sample is placed in it the instrument is calibrated, and it is found that, due to the resistance of the screen, the head of water in the standpipe takes 5s to fall from 1000mm to 150mm above the datum level. When a bed of silt 15mm deep is placed on the screen the time taken for the head to fall between the same limits is 150s. Calculate from first principles the coefficient of permeability of the soil.
2000 (April). Question 8

(a) On the assumption that a continuous stratum of sandy soil overlies a horizontal impermeable bed, a test well was sunk to the bottom of the stratum in order to determine its in-situ permeability. Two observation boreholes were drilled at 12.5m and 25.0m respectively from the test well. Water was pumped from this well at a rate of 3 x 103 m3/s until the water level in the boreholes became steady at 4.25m and 6.25m respectively above the impermeable bed. Determine the mean permeability of the stratum in m/day, deriving any formulae used. (b) If in fact this upper stratum was not continuous and the steady state flow condition was due to a confined layer of sandy gravel to 3.6m above the impermeable bed, determine the mean permeability of this layer in m/day, deriving any formulae used.
1999, Question 2
A constant head permeameter of cross-sectional area 50cm2 is set up as shown in the figure overpage, with two saturated materials of coefficients of permeability k1 = 8 x 106 m/sec and k2 = 3 x 106 m/sec. (a) (b) What flow rate (in cm3/sec) is recorded? At what level, in relation to the water level in the constant head tank, does the water level stand in a tapping at point X? (c) Both soils have a saturated unit weight y = 1.9 yw; if the height of the overflow is reduced, at what level will it be below the constant head tank when piping first occurs? Either work from first principles, or derive any formulae used.
Page 52 of 54
Page 53 of 54
References and further reading

ASTM D2434-68(2000) Standard Test Method for Permeability of Granular Soils (Constant Head) ASTM D2435-96 Standard Test Method for One-Dimensional Consolidation Properties of Soils ASTM D4186-89(1998)e1 Standard Test Method for One-Dimensional Consolidation Properties of Soils Using Controlled-Strain Loading BS 1377. Methods of test of soils for civil engineering purposes. Part 6. Consolidation and permeability tests. British Standards Institution BS 8004. British Standard Code of Practice for Foundations (formerly CP2004). 1986. British Standards Institution Butterfield, R. (1979). A natural compression law for soils an advance on e-logp'. Geotechnique 29, pp.469-480 Craig, R.F. (2005). Craig's Soil Mechanics. 7th edition. E&F Spon Craig, R.F. (2004). Craig's Soil Mechanics. 7th edition, Solutions Manual. E&F Spon Ferris, J.G., D.B. Knowles, R.H. Brown & R.W. Stallman, 1962. Theory of Aquifer Tests. US Geological Survey Water-Supply Paper 1536-E. (available as a pdf, http://pubs.usgs.gov/wsp/wsp1536-E ) Florin, V.A., and Ivanov, P.L. (1961). Liquefaction of Saturated Sandy Soils. Proc., 5th. Int. Conf. Soil
Mechanics and Foundation Engineering., Paris, Vol. 1, 107-111

Head, K.H., Manual of Soil Laboratory Testing, Permeability, Shear Strength and Compressibility Tests, Pentech Press Lambe, P.W. (1961). Residual pore pressures in compacted clay. Proceedings of the 5th International
Conference on Soil Mechanics and Foundation Engineering (ICSMFE), Paris, Vol.1, pages 207ff
Lambe, P.W., and Whitman, R.V. (1979). Soil Mechanics SI version. Wiley Powers, P. (1991). Dewatering and groundwater control. In Foundation Engineering Handbook, 2nd edition, Chapman & Hall, Chapter 7, pp.236248 Sarsby, R. (2000). Environmental Geotechnics. Thomas Telford Ltd
Page 54 of 54
Taylor, D.W. (1948). Fundamentals of Soil Mechanics. Wiiey Terzaghi, K., Peck, R., and Mesri, G. (1996). Soil Mechanics in Engineering Practice. Wiley US Army Corps of Engineers (1952). Seepage Control, Soil Mechanics Design, Dept of the Army, Washington DC US Navy. (1962). Design Manual Soil Mechanics, Foundations, and Earth Structures. NAVDOCKS DM7 Winterkorn, H.F., and Fang, H-Y, (1991). Soil technology and engineering properties of soils. In
Foundation Engineering Handbook, 2nd edition, Chapman & Hall, Chapter 3, pp.88143
Software
Examples of software suppliers (endorsement or completeness not implied!) www.geotechnicaldirectory.com/Software.htm SEEP2D www.seepage-analysis.com SEEP2D www.geology-software.com/seep-model.html SVFLUX www.scisoftware.com GEOSLOPE www.geo-slope.com

M04 Permeability and Seepage

Uploaded by

Copyright:

Available Formats

M04 Permeability and Seepage

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

M04 Permeability and Seepage

Uploaded by

Copyright:

Available Formats

CVNG 2002 Soil Mechanics