You are on page 1of 45

CHAPTER 4

The pre-existing capabilities of ICFEP

4.1. Introduction

ICFEP is a finite element computer program, developed by Professor David Potts at Imperial
College, specifically for geotechnical engineering analysis. The capabilities of the program have
been progressively updated and expanded. At the commencement of this project, version 8.0 of
ICFEP was in use. Development work for this project and a number of other simultaneous
research projects introduced several new capabilities to ICFEP, the most relevant being an ability
to undertake unsaturated soil analysis. Implementation and validation of these new capabilities
led to the formal designation of ICFEP version 9.0.

Before presenting the development work that was undertaken as part of this research project, it is
necessary to outline the pre-existing capabilities of ICFEP version 8.0.

4.2. ICFEP version 8.0 general capabilities

ICFEP uses four or eight nodal quadrilateral elements to resolve finite element analyses, whether
two-dimensional plane strain, two-dimensional plane stress, or three-dimensional axi-symmetric
problems. Three-dimensional problems may also be analysed through the application of Fourier
series aided analysis.

Analysis can be drained, undrained or partly drained ( i.e. coupled ), and a large range of
constitutive soil models is available. However, within version 8.0, all such models are for
saturated soils. A number of consolidation / permeability options and a variety of boundary
conditions, several of which are relevant to this project, are available in version 8.0 of the
program.

4.3. Consolidation and permeability models

A notable strength of ICFEP is its ability to undertake fully coupled consolidation.

Chapter 4 88
To calculate the consolidation behaviour of a soil, it is necessary to simulate its permeability
behaviour. Prior to this project, ICFEP had the capability to simulate the permeability of a soil to
water flow through a number of models. Options existed for isotropic or anisotropic permeability,
with or without spatial variation. Two methods of generating stress-dependent permeability were
available, as was a voids ratio dependent model, and perhaps most relevantly, suction-dependent
permeability behaviour could also be modelled. This ‘suction switch’ may be combined with
either of the stress-dependent models.

4.3.1. The suction switch

The suction dependent permeability model –‘the suction switch’ – within ICFEP is a relatively
simple one, and is illustrated in Figure 4.1.

When using the suction switch with otherwise homogeneous isotropic permeability, a constant
( fully saturated ) permeability is maintained for compressive values of pore water pressure, and
up to a specified tensile pore water pressure ( p1 in the figure ). Permeability then drops
logarithmically with a linear increase in the tensile PWP, until some predetermined residual
permeability is obtained ( kmin at p2 in the figure ), at which point permeability again become
constant regardless of further changes in the PWP.

If the suction switch is combined with a stress-dependent model, then the initial constant
permeability is actually determined from the stress-dependent behaviour, with the residual
permeability affected proportionally, but otherwise the model works as described above.

A number of other researchers, for example Li and Griffiths (1988), and Giorda and Desideri
(1988) have used similar suction-permeability relationships to the suction switch within ICFEP.
The use of a constant residual permeability is also consistent with the ideas of Fredlund (1998).

4.4. Boundary conditions

ICFEP permits a wide variety of boundary conditions to be applied. All the boundary conditions
may be specified for particular increments of the analysis, and hence are not fixed for the entire
analysis, but may be varied throughout the problem.

Chapter 4 89
Such boundary conditions include the option to specify forces, moments, rotations or
displacements of boundary nodes. Normal and / or tangential stresses may also be applied, and
body forces, such as the self-weight of elements, may be specified.

The boundary conditions particularly applicable to infiltration problems are those concerned with
flow and consolidation. Options include a fixed pore water pressure ( variable flow ) boundary, or
a specified flow ( variable pore water pressure ) boundary, which includes the possibility of a no-
flow boundary. Non-zero specified flows may be in or out of the analysis mesh. Additionally, any
node within the mesh may be specified as a source or sink, adding or removing water
( respectively ). Most relevantly to this project, ICFEP contains a precipitation boundary
condition.

4.4.1. The precipitation boundary condition

The precipitation boundary condition within ICFEP v8.0 enables the simulation of rainfall on a
ground surface. It acts either as an infiltration ( specified flow ) condition, or as a constant
pressure ( variable flow ) condition.

The operation of the precipitation boundary condition is illustrated in Figure 4.2. The boundary
condition requires that an infiltration rate ( i.e. the rainfall rate at the ground surface ) be
specified, along with some maximum threshold value ( THV ) of the pore water pressure at the
surface boundary.

If at the start of an increment of the analysis, the pore water pressure at the surface boundary is
below ( that is, is more tensile than ) the THV, then ICFEP applies an infiltration boundary, using
the specified infiltration rate.

If the pore water pressure at the boundary exceeds ( that is, is more compressive than ) the THV
at the start of the increment, ICFEP re-sets the boundary pore water pressure to be equal to the
THV, and maintains a constant pressure equal to this value.

The constant boundary pressure is maintained by applying an inflow that is some portion of the
specified infiltration rate. The ‘excess’ portion of the specified infiltration is disregarded. If the
specified infiltration rate is reduced after the boundary has switched to a constant pressure
boundary, then it may switch back to being an infiltration boundary if the new maximum inflow
rate is insufficient to maintain the THV pressure.

Chapter 4 90
In applying the precipitation condition, the specified infiltration rate is normally taken as the
actual rainfall for the site under analysis. If allowance is required for canopy intercept, this must
be done by inputting a reduced rainfall rate. However, no allowance needs to be made for run-off:
the boundary condition automatically determines the portion of the specified inflow that enters
the mesh and treats the remainder as run-off, based on the THV chosen.

The proportion of the infiltration that becomes run-off is not, however, explicitly modelled.
Rather, it is simply discounted from the analysis, since this flow occurs outside of the analysis
mesh.

Typically for a slope analysis, the THV would be set to 0 kPa. Thus the soil could develop an all-
compressive ( ‘fully saturated’ ) pore water pressure profile, but a compressive pore water
pressure greater than zero could not build up at the ground surface. Non-zero THVs may also be
specified: compressive pore pressures greater than zero may be specified for the THV, to allow
surface ponding to occur. The maximum depth of ponding that can be achieved will thus be
determined by the value of the THV specified. Alternately, a tensile THV may be specified,
which prevents total loss of suctions at the ground surface.

It should be noted, however, that care is required in dealing with ponding. The precipitation
boundary condition applies the specified infiltration at the soil surface, and the resulting changes
in the pore water pressure reflect this. In the ‘real’ field condition, once ponding occurs, rainfall
enters into a free water surface, which will be some height above the soil surface.

Within ICFEP, if ponding is permitted, the increase in pore pressure at the ground surface will
imply a certain depth of ponded water ( for example, a THV of –9.81 kPa would imply a
maximum depth of ponding of 1m ). From the infiltration ( rainfall ) rate specified it is possible
to determine the time that should be required for such a depth of ponding to build up, assuming
that all the rainfall ponds and none enters the soil. Since some of the rainfall does penetrate into
the soil, the actual time required to achieve a given depth of ponding will be somewhat greater
than this. However, ICFEP generates ponding water pressures much quicker than this would
indicate, and the surface pore water pressure, once above ( more compressive than ) 0 kPa, is
inconsistent with the total volume of water applied as rainfall. That is, while the combination of
precipitation rate and elapsed time indicate that, say, 50mm of water has fallen on the ground
surface, the calculated rise in PWP at the ground surface ( when starting from a fully saturated

Chapter 4 91
condition with the phreatic surface at the ground surface ) is invariably greater than 50mm water-
equivalent pressure.

This is a reflection of the fact that the rainfall is applied straight to the surface of the soil, and not
to the ponded water surface. Inflow into the soil, which is what ICFEP models, involves adding
water into a volume of space already occupied fully by a mixture of soil particles and water. Thus
the PWP responds rapidly to even small volumes of infiltration. To account for ponding, the
infiltration should be into a free water surface in some imaginary element above the soil surface,
which is empty of soil particles or ( prior to ponding ) any water, and in which only hydrostatic
water pressures can develop.

However, as long as there is no requirement for surface ponding to be accurately modelled, the
precipitation boundary condition works accurately.

Since this project is concerned with rainfall into slopes, the assumption will always be made that
no ponding is possible, and the maximum pore water pressure at the surface will be limited to
0 kPa ( that is, the phreatic surface can be co-located with the soil surface, but no free standing
water can exist on top of the soil ). As such, this inconsistency between rainfall rates and depths
of ponding is not significant. However, further development of ICFEP is required to address this
issue before any analyses involving ponding is attempted.

In investigating the capabilities of ICFEP v8.0, a further problem was determined with the
precipitation boundary condition, which did need to be dealt with as part of this project. As stated
above, the boundary pore pressure is adjusted back to the THV if at the start of the increment the
pressure exceeds the THV as a result of the previous increment’s infiltration. Where inflows are
relatively small and the increment time step is short, the amount by which the pore pressure
exceeds the THV is likely to be small, and this method of operation is reasonable.

However, it was found than that if the infiltration rate was high ( relatively ) and the time step was
large, very high compressive pore pressures could be generated at the surface on the last
increment in which an inflow boundary condition exists. This is illustrated by Figure 4.3.
‘Increment 0’ represents some pre-existing pore water pressure distribution. Precipitation is
applied from increment 1, and the precipitation rate is high relative to the permeability of the soil,
while the time step of the increment is relatively long. The pore water pressure distribution at the
end of increment 1 is as shown in the figure. While on the next increment the surface pore water
pressure is corrected, the shallow sub-surface pore pressure distribution is in error: this is shown

Chapter 4 92
as the ‘increment 2’ plot line in the figure. The increment 2 PWP distribution shown is obviously
unrealistic given that the surface PWP should not be able to exceed 0 kPa. Clearly, it was
necessary to modify the boundary condition to limit the amount by which the THV could be
exceeded before the condition switched from inflow to constant pressure. This was done as part of
the development work, through the use of an automatic incrementation procedure, and is
described in Chapter 6.

4.5. Modelling of precipitation on to saturated soil

While ICFEP v8.0 is unable to model unsaturated behaviour, the combination of the precipitation
boundary condition and the suction switch does enable some study to be made of the pore water
pressure response to rainfall.

While the unsaturated soil behaviour model was under development ( as detailed in Chapters 5
and 6 ), a limited study was undertaken of infiltration into a tension-saturated ( tensile PWP, but
with a fully saturated soil model ) soil, principally through a series of column analyses.

The initial set of analyses looked at how differences in the specified infiltration rate ( within the
precipitation boundary condition ), the saturated permeability and the parameters on the suction
switch affected the resulting pore water pressure distribution within the column.

The column modelled consisted of a 100m high column in plane strain, with no-flow boundaries
along the sides and base, and a hydrostatic PWP profile with an initial phreatic surface 20m
below the ground surface ( see Figure 4.4 ). Analysis was carried out normally using 100
increments, each of 12 seconds. Soil behaviour was modelled as linear elastic, with a Young’s
modulus of 25000 kPa and a Poisson’s ratio of 0.25. Thus the volume of the column was able to
expand elastically as water entered the surface.

4.5.1. No suction switch

Precipitation occurring at various rates was applied to the surface of the soil column. The soil
permeability was set to be independent of the pore water pressure ( no suction switch applied ) at
either 3.4E-6 m/sec or 3.4E-5m/sec, and was isotropic and homogeneous.

Where the inflow rate was set less than the permeability of the soil, there was no significant
change in the pore water pressures. For both permeabilities, with the inflow rate set at two orders

Chapter 4 93
of magnitude less than the permeability, the change in the pore water pressure at the column
surface after 50 increments of analysis ( at which point the analysis was terminated ) amounted to
less than 1 kPa ( or less than 0.5% of the initial tensile pressure ).

The more permeable column was also analysed with an inflow of only 1 order of magnitude less
than the permeability, and this produced a change in the surface PWP of about 9 kPa ( or less than
5% of the original tensile PWP ).

The depth of the slight variation in the pore pressure that did occur seemed to be dependent on the
permeability of the soil, but independent of the inflow rate. For the less permeable soil, a small
reduction in pore water tension occurred over the top 6m of the column, whereas in the more
permeable soil, this zone of effect extended to 30m below the column surface. Where the
specified inflow rate was varied, the higher the inflow rate the greater the variation in pore
pressure, but the depth of the zone effect appeared to remain the same. It is believed that this
behaviour is due to an elastic change in the soil structure caused by inflow, since no change in the
PWP would be expected for the case where the inflow rate is less than the soil permeability in a
rigid soil.

When the inflow rate was increased to be exactly equal to the permeability of the soil, there was a
clear, gentle reduction in the pore water tensions within the soil, as shown in Figure 4.5.

The figure shows the response of the more permeable column, but both columns exhibited similar
behaviour. It may be noted that while the tensile value of the pore water reduces, the highest
tension remains at the soil surface throughout. Thus, the pore water pressure always becomes
more compressive with depth.

This behaviour was also sensitive to the permeability of the soil, with the changes in pore water
pressure being more rapid in the higher permeability soil. Obviously, since inflow was set equal
to permeability, this may simply be a reflection of the fact that the inflow rate is higher.

For the lower permeability soil, the effects were much the same, just of a smaller magnitude.

The modelled behaviour was felt to be slightly surprising. With inflow equal to permeability, it
was assumed inflow would pass straight through the soil. With no-flow side and bottom
boundaries, it was expected that under these conditions, there would be a uniform increase in
water pressure, with the PWP profile remaining hydrostatic, but with the phreatic surface steadily

Chapter 4 94
rising. Again, it is believed that the divergence between expected and actual behaviour is a
reflection of the ability of the soil to deform elastically.

While not shown in the figure, when the inflow ( precipitation ) was stopped but the analysis
continued on, internal drainage occurred within the mesh, leading to the development of the
expected profile. The PWP profile became approximately linear but slightly steeper than
hydrostatic quite rapidly after cessation of infiltration, then slowly reverted towards the
hydrostatic state, but with pore pressures more compressive than the initial pre-infiltration
condition.

Once the inflow rate specified exceeded the soil permeability, it was expected that noticeable
effects would occur. This was found to be the case, and there was a significant difference between
this behaviour and that described immediately above.

With inflow exceeding permeability, there was a rapid change in the surface pore water pressure,
with it becoming much more compressive. The precipitation threshold value ( specified as 0 kPa )
was rapidly reached, with the boundary condition then switching to the constant pressure
condition. Figure 4.6 illustrates the modelled behaviour. As can be seen, unlike the case where
inflow equals the permeability, in this case the pore pressure does not always become more
compressive with depth. Rather, the initial tensile pressures tend to remain at shallow depth in the
column, despite the compressive changes at the surface.

It was found that increasing the rainfall inflow ( while keeping the soil permeability constant )
had the effect of increasing the rate at which the tensile pore pressure became compressive.

Similarly, for a given inflow rate, increasing the permeability of the soil leads to a more rapid loss
of pore water tension. The influence of the soil permeability seemed to be greater than that of the
specified inflow rate.

These points are illustrated by the results presented in Figures 4.6, 4.7 and 4.8.

It can be seen from these results that ( when using a tension-saturated model ) without some form
of suction-dependent permeability it is not possible to produce a true wetting front effect.
Certainly, it seems impossible to reproduce the situation where precipitation reduces the tensile
water pressure, while still preserving some degree of tension in the pore water. It was believed
after this work that the application of the suction switch would enable this form of behaviour to
be reproduced.

Chapter 4 95
4.5.2. Tension-saturated column with suction switch

The analyses undertaken without a suction-dependent permeability ( as detailed above ) were then
repeated with the suction switch applied.

It was found that the application of a suction-dependent permeability had a significant effect on
the pore water pressure response to infiltration. This is illustrated by Figures 4.9, 4.10, 4.11 and
4.12. These present the results of analyses using the more permeable soil column, with a suction
switch set to give a kmax / kmin ratio of 100, between tensile PWPs of 10 kPa ( ‘p1’ ) and 100 kPa
( ‘p2’ ) ( as defined in Figure 4.1 ). It may be noted that Figure 4.11 is the equivalent to Figure 4.7,
and Figure 4.12 relates to Figure 4.8, in respect to flow rates.

As can be seen, a distinct wetting front profile is developed when a suction-dependent


permeability is applied. Notable also is that where the inflow exceeds the maximum permeability,
the infiltration tends to reduce pore water tension within the ‘wetted zone’ ( the area above the
wetting front ) to zero ( as shown in Figures 4.11 and 4.12 ). This can be compared to the case
where the infiltration rate is specified as being equal to the maximum permeability, as shown in
Figure 4.10, or where infiltration is less than the maximum permeability, as in Figure 4.9.

Where inflow equals the maximum permeability, the pore water tension is rapidly reduced by
infiltration to a tensile value of 10 kPa. This corresponds to the ‘p 1’ value set on the suction
switch. Hence the soil permeability at the surface of the column is now equal to its specified
maximum value, and remains constant unless the infiltration rate drops allowing the pore pressure
to become more tensile again.

However, deeper within the column, the water tensions are higher, and permeability is
correspondingly lower. Thus, while the wetting front continues to develop deeper in the soil, the
near surface behaviour is flow through a constant permeability soil, and becomes as described for
the no-suction switch case, with a constant pore water pressure at the soil surface. The
permeability of the soil at the surface is now equal to the imposed inflow, so flow in equals flow
out, and an approximately steady condition is achieved within this zone.

The case where the infiltration rate was less than the maximum permeability is illustrated by
Figure 4.9. Here, the effect of the suction switch is to give a suction-dependent permeability at the
surface that is initially less than the inflow rate.

Chapter 4 96
As infiltration proceeds, the tensions in the pore water are reduced, and the permeability of the
soil increases correspondingly, until it matches the inflow rate. At this point, a pseudo-steady
state is achieved. Due to the side and base boundary conditions, the soil column tends to ‘fill up’
with water, leading to a gradual rise in the phreatic surface later in the analysis.

It may be noted from Figure 4.9 that the steady pore tension is 55 kPa; reference to Figure 4.1,
and the application of the specified suction switch parameters, shows that the permeability rate
which corresponds to 55kPa suction is 3.4E-6 m/min. Thus it can be seen that the steady state is
achieved once the suction dependent permeability has risen to a value equal to the applied
infiltration rate.

The behaviour of the soil column with the suction switch applied was generally similar regardless
of the specified maximum permeability, although the effect of lowering the maximum
permeability was to increase the reduction in tensile pore pressure at the column surface. That is,
the compressive change in pore water pressure that occurred at the column surface was less for
the more permeable column than for the less permeable column.

It was also found that changing the ratio of permeabilities in the suction switch ( i.e. kmax / kmin )
had some small effect. The smaller the ratio used the smaller the change in pore water pressure
that occurred at the column surface. This is consistent with the influence of permeability. A small
suction switch ratio will result in the soil permeability at the surface being slightly reduced from
the specified maximum, whereas a high suction switch ratio will produce a much-reduced near
surface permeability. The soil will respond to infiltration accordingly.

Clearly, then, the application of the suction switch produces significant effects when modelling
infiltration into a tension-saturated soil. For the range of values used, the exact parameters applied
within the switch seem to be less significant than the act of applying the switch, although this may
not be the case generally. More over, while it was possible to reproduce the type of wetting front
that is believed to occur in unsaturated soil using the suction switch combined with a tension-
saturated soil, this is not sufficient to claim that unsaturated soils may be modelled. For example,
the tension-saturated approach does not allow for the storage of water and variation in volumetric
water content that occurs in an unsaturated soil. The development work set out in chapters 5 and 6
is still required.

This tension-saturated work does show that some of the response of unsaturated soil to infiltration
is likely due to the variations in permeability, rather than the direct effects of variable water

Chapter 4 97
content and any related volume change. For example, the resulting pore water pressure
distributions shown in Figures 4.9 to 4.12 are comparable to the work of Rubin et al (1964),
Kasim et al (1998) and Sun et al (1998).

4.5.3. Element size sensitivity

It is notable that in Figures 4.9 to 4.12, the plotted pore water pressure distribution is fairly
angular for many of the cases shown.

As described by Chapuis et al (2001) and Forsyth (1988), there is likely to be some element size
dependency in the results of a finite element analysis of infiltration into unsaturated ( or tension
saturated ) soil, and it was believed that the angular distribution plots were due to this aspect.

Thus, while the development work for unsaturated behaviour was underway, a small study was
undertaken into the effects of element size on the resulting pore water pressure distribution
generated by ICFEP, for the case of infiltration into a tension saturated soil column.

The element size sensitivity study was carried out using the same basic 100m high tension-
saturated column as previously described, except that the number and size of elements in the top
20m of the column were varied.

For this study, the maximum permeability of the soil was specified as 3.4E-9 m/s, with inflow
equal to 3.4E-7 m/s. One increment of the analysis represented 1200 seconds. The suction switch
was applied, with permeability varying between water tension values of 1 kPa and 1000 kPa. The
permeability ratio was varied, being either 10, 1000, or 1000,000, to ensure any effects observed
were due to element size, and were not simply due to an extreme permeability ratio.

Selected results of some of the sensitivity study analyses ( all with a permeability ratio of
1000,000 ) are presented in Figures 4.13 to 4.16. It should be noted that these figures only show
the PWP distribution for the top 40m of the soil column, since this is the region were the water
pressure was responding to infiltration.

Figure 4.13 shows the results obtained when the top 20m of the column was composed of 2m
thick elements (in all cases, the elements below 20m depth were 10m thick ). As can be seen, the
effects of infiltration are limited to near the surface of the column, and the resulting PWP
distribution is not smooth, but demonstrates a ‘spike’ just under the wetting front.

Chapter 4 98
Figure 4.14 shows the results when the near surface element thickness was increased to 10m. the
apparent response to infiltration shows a less immediate reaction at the surface, but with a deeper
area of effect. The ‘blocky’ nature of the PWP distribution clearly corresponds to the element size
chosen, and demonstrates that there is element size sensitivity.

Figure 4.15 shows the results when the top 20m of the column was modelled using 0.5m thick
elements, while Figure 4.16 shows 0.25m thick elements. In both cases, the resulting PWP
distribution profile is relatively smooth, and the two plots are basically the same. The implication
is that for this particular infiltration problem, a surface element size of 0.5m is necessary to avoid
element size dependency in the results. Further reduction in the element size does not appear to
provide any further benefit, and by increasing the number of elements in the mesh, increases the
computational effort required to obtain a solution.

This conclusion does not seem to be affected by the choice of the permeability ratio ( although the
results for the other ratios are not presented here ).

‘Good practice’ in finite element analysis is to use smaller elements where the greatest change is
occurring, and it is clear from the results so far presented from tension-saturated ICFEP analyses,
and from the literature ( see Chapter 3 ), that the near surface elements tend to experience the
greatest change in PWP. Thus the conclusion of the element size study – that relatively small
elements should be placed in the near surface position – is consistent with what might be
expected.

Experience gained during this project through extensive use of ICFEP and the precipitation
boundary condition suggests that for analysis of ‘field-scale’ problems, the surface element layer
should generally be less than 1m thick.

4.5.4. Two-dimensional tension-saturated analysis

This project is principally concerned with the effects of precipitation on unsaturated soil slopes.
The one-dimensional ( column ) analysis of a tension-saturated soil gave results that conform to
some degree to the expected behaviour for unsaturated soils ( based on the literature review ).
From this, it was felt that a full two-dimensional analysis of rainfall into a tension-saturated slope
might prove useful to understand the likely behaviour of an unsaturated soil slope. The analysis
undertaken was carried out using ICFP v8.0, but with the automatic incrementation procedure

Chapter 4 99
( described in Chapter 6 ) adopted to deal with the precipitation boundary condition problem
detailed in section 4.4.

The FE mesh used for the analyses is shown in Figure 4.17. A range of permeabilities were
applied, but the results presented here are all for a maximum permeability of 3.4E-7 m/s unless
stated otherwise. The suction switch was applied, with the parameters used varying according to
the analysis, as did the specified precipitation rate. The incremental time step was set at 300
seconds. The sides of the mesh were given no-flow boundary conditions, while the base had a
fixed pore water pressure boundary condition. The precipitation boundary condition was applied
across the whole of the upper surface of the mesh, including the horizontal sections as well as the
actual slope.

Figures 4.18 and 4.19 show the results of an analysis, with precipitation at 6.8E-8 m/s. The
suction switch parameters had permeability varying between water tension values of 1 kPa and
10 kPa, with a permeability ratio of 100.

Figure 4.18 ( a to d ) shows the evolution of the contours of PWP within the slope. As can be
seen, the effect of the rainfall is to quite rapidly reduce the pore water tension in the near surface
soil, but within the core of the slope there appears initially to be minimal effects. Additionally, the
effects of rainfall seem to be greater lower down the slope, with a broad zone of soil wetting up.
In contrast, the soil at the crest shows a much narrower wetting band.

With time, the drop in tensile PWP percolates deeper into the mass of the soil, but it is notable
that significant compressive water pressures do not build up. Also, the effect of gravity drainage
appears to be visible, in that slight tensions are retained in the upper portion of the slope. This is
more clearly illustrated by Figure 4.19, which presents the pore water pressure along a horizontal
cross section through the slope, 4m below the crest ( as shown in Figure 4.17 ).

The initial pressure along this section is a constant tensile value, consistent with the section’s
height above the initial phreatic surface.

The effect of rainfall is to produce a ‘wetted band’ of soil on the surface of the slope, which
gradually migrates in towards the core. The ‘wetted band’ retains some slight pore water tension.
Given the suction switch parameters, a tension value of 1 kPa would be logical, based on the one-
dimensional analyses described earlier, but as can be seen, slightly higher tension values are
retained, suggesting that some two-dimensional flow effect may be present. It may also be noted

Chapter 4 100
that while the inflow rate is greater than the initial ( tension-dependent ) permeability of the soil
across most of the mesh surface, it is less than the maximum permeability of the soil. Hence some
amount of pore water tension beyond the ‘p1’ value of 1 kPa is not entirely illogical.

Increasing the specified infiltration rate had the logical effect of creating a more rapid reduction
in the pore water tensions, with a sharp wetting front, as shown in Figure 4.20.

Two ‘spot-value’ contour plots for this case are presented in Figure 4.21. As can be seen in Figure
4.21a, the effect of heavier rainfall is to destroy the tensile pore pressures along the entire surface
of the mesh, with no zone of slightly tensile pore pressures being retained towards the top of the
slope.

Figure 4.21b shows what is probably approaching a steady state condition, with a phreatic surface
that is co-located with the slope surface, and no tensile pore pressures anywhere in the slope. The
resulting pore water pressure contours are certainly influenced by the prescribed boundary
conditions, with a fixed PWP along the base, and no lateral flow permitted through the sides of
the mesh. However, it is also clear that the slope has ‘wetted up’ to a much greater degree under
the heavier rainfall than it did in the first case presented.

In this case, the pore water tension at the surface is rapidly reduced to zero due to the rainfall rate,
which is generally very much greater than the tension-dependent permeability. However, since
the inflow rate is less than the maximum permeability, and particularly given the apparent two-
dimensional effects detailed for example 1 ( see Figure 4.18 ), it would seem likely that the true
long term steady state condition would see the redevelopment of a slight surface pore water
tension, such that the permeability associated with that tension ( through the suction switch )
equalled the inflow rate. Reference to the suction switch equation on Figure 4.1 indicates that this
would mean a tensile value of just over 1 kPa at the surface.

It may also be noted that the ‘-25 kPa’ contour shown in Figures 4.21a and 4.21b is apparently
unchanging where it lies under the horizontal ground surface at the base of the slope. However,
comparison of these contours with Figure 4.18a, which shows the initial condition for all the
slope analyses, shows that the pore pressure has become more compressive in this zone, as would
be expected.

Figure 4.22 shows the situation when the precipitation rate was not changed, but instead the
permeability ratio in the suction switch was increased by two orders of magnitude ( with the ‘p2’

Chapter 4 101
value in the suction switch also increased, by one order of magnitude ). The wetting front
becomes sharper, but penetrates into the soil more slowly. Also of note is that the ‘residual’
tension in the near-surface soil is higher. Figure 4.23 shows the related contour plots.

Clearly, increasing the permeability ratio results in a reduction in the actual permeability when
pore water tensions equal or exceed the ‘p1’ value in the suction switch. Since the initial condition
at the height of the section shown in Figure 4.22 is of a tension value greater than the specified
‘p2’ value ( which is more tensile than ‘p1’), the initial permeability in example 3 is considerably
less than in example 1. Thus the observed behaviour is a reflection of the actual permeability of
the soil, and is as might be expected, with a slower rate of penetration of the wetting front into the
soil.

It can be seen that if the specified precipitation rate of example 3 is applied to the suction switch
formula, there is a corresponding water pressure of approximately 18 kPa, which is consistent
with the ‘residual’ water tension shown in Figure 4.22.

If the permeability ratio is held constant, but the range of pore water pressure over which the
permeability varies is increased, the effect appears markedly different. This was the case in
example 4, presented in Figure 4.24, where there appears to be no proper wetting front. Instead, it
seems that the whole soil mass is wetting up evenly, such that the pore water pressure gradually
becomes more compressive, but the contours of PWP remain approximately horizontal ( as shown
in Figure 4.25 ).

As described for example 2, the contours of pressure shown ( in Figure 4.25 ) indicate little
change in the pore water pressure distribution under the horizontal ground surface at the base of
the slope. However, again, the figures show increments 40 and 120. Comparison between these
results and the initial condition, given in Figure 4.18a, shows that there has been a pore pressure
response to infiltration in this region of the mesh.

In this case, the range over which the suction switch applied was considerably larger than the
range of tensile pore water pressures within the slope. Thus, there is little variation in
permeability of the soil, so the soil’s behaviour tended towards that of a soil with a homogeneous
permeability ( i.e. a soil in which there was no suction-switch applicable ). Hence the behaviour is
reminiscent of the tension-saturated soil column with no suction switch shown in Figure 4.5.

Chapter 4 102
However, it appears that the soil permeability, infiltration rate and suction switch parameters all
interact. Figures 4.26 and 4.27 illustrate the development of the pore pressures when both the
range over which the suction switch operated and the permeability ratio were increased.
Additionally, however, both the imposed rainfall and the maximum permeability of the soil were
decreased ( as shown on the figure ). While there is still no clear wetting front developing, some
sensitivity to the source of the infiltration is revealed.

Although the range of tensile pore water pressures over which the soil permeability varies extends
considerably beyond the maximum tension in the soil, the permeability ratio is sufficiently large
that there is some significant pore pressure dependent variation in permeability within the slope
mesh.

This example illustrates that, at least in tension-saturated soils, it is incorrect to simply divide
examples into ‘wetting front’ and ‘non-wetting front’ cases. Rather, the development of a wetting
front is clearly dependent on a number of parameters, which results in a gradual variation in the
observed response to infiltration.

4.5.5. Conclusions of the tension-saturated analyses

From the results of the tension-saturated analyses, it appears that a significant variation in
permeability with direction of flow is required to produce a distinct wetting front. The more
homogeneous the permeability, the less distinct the wetting front becomes, until, when
permeability is more or less constant, there is no discernible wetting front, but just a general
wetting up.

Since this behaviour appears to be controlled by the permeability variation, it is logical to assume
that it could be mimicked through the use of a permeability model that treated permeability as a
function of depth or effective stress, thus avoiding the need for a dedicated suction switch.

However, it should now be clear that with pore water pressure dependent permeability, the
permeability and pore pressures interact. Changing one of these parameters produces a change in
the other, which affects the whole flow regime within the soil. Attempts to mimic this behaviour
with a simple depth dependent model would be unable to show this interactive behaviour. If the
total stress is known, then substituting an effective stress permeability model for the suction
switch would seem acceptable. However, as discussed later ( see Chapter 6 ), for an unsaturated
soil, changes in the volumetric water content of a unit element of soil will produce corresponding

Chapter 4 103
changes in the weight of the soil element, as well as the change in the pore pressure reflected by
the SWCC.

Thus the total stress in an unsaturated soil subject to infiltration or other variation in water content
is not constant. It seems clear therefore, that pore water pressure dependent permeability should
be modelled explicitly if the correct flow regime is to be reproduced with any accuracy, and the
ICFEP permeability-suction switch provides a way in which this can be done.

The results of the tension-saturated slope analysis, while not using an accurate model of
unsaturated soil behaviour, and based also on a fictitious slope, provide a useful insight into how
variable the behaviour of an unsaturated slope is likely to be. The results show how several
factors may be expected to impact on the resulting behaviour.

Notably, it appears that the variation in permeability is critical, and it is through affecting
permeability that the controlling factors affect behaviour. However, further conclusions about the
behaviour of an unsaturated slope require a proper unsaturated soil model, since clearly, the
tension-saturated analyses cannot reproduce the variation in water content that will occur within
an unsaturated soil. To determine the significance of this factor, a true unsaturated approach must
be developed.

Chapter 4 104
Log k

K max

K min

P1 P2 Pore water pressure


(Tension +ve)

Log k = log kmax - {(p-p1)/(p2-p1)}log(kmax/kmin)

Figure 4.1: ICFEP permeability - suction switch

105
Low rain fall intensity: PWP distribution at start of
infiltration ( surface PWP more increment
tensile than THV )
THV
Tensile,
+ve

Compressive,
-ve

High rain fall intensity: surface


ponding (THV compressive) or run- PWP distribution at start of
off ( surface PWP = THV ) increment
THV
Tensile,
+ve

Compressive,
-ve

THV: Threshold Value as specified in precipitation boundary condition

Figure 4.2: Precipitation boundary condition

106
Increment 0

Increment 1

Increment 2

Pore Water Pressure

Compression
Tension

Depth, z THV

THV: Threshold Value as specified in


precipitation boundary condition

Figure 4.3: Precipitation boundary condition with large


timestep and inflow rate

107
5m
+196.2 kPa

Initial ( hydrostatic )
pore water pressure
distribution

100m

-784.8 kPa

Figure 4.4: Tension-saturated soil column for ICFEP v8.0


analysis

108
0 Increment 0
Increment 5
10 Increment 25
Increment 50
20
Increment 80

30 Increment 100

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid Stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-5 m/min

Figure 4.5: Precipitation into tension-saturated soil


column, no suction switch

109
0 Increment 25

10 Increment 50

20 Increment 80

30 Increment 100

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-6 m/min
Infiltration = 3.4E-4 m/min

Figure 4.6: Precipitation into tension-saturated soil


column, no suction switch.

110
0 Increment 25

10 Increment 50

20 Increment 80

30 Increment 100

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-4 m/min

Figure 4.7: Precipitation into tension-saturated soil


column, no suction switch.

111
Increment5
0
Increment10
10
Increment 25
20
Increment 50
30
Increment 80
40 Increment 100

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-3 m/min

Figure 4.8: Precipitation into tension-saturated soil


column, no suction switch.

112
0 Increment 0
Increment 100
10
Increment 300
Increment 600
20
Increment 800
Increment 1000
30

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-6 m/min
Suction switch ratio, kmax/kmin = 100

Figure 4.9: Precipitation into tension saturated column,


with suction switch.

113
0 Increment 0
Increment 25
10 Increment 50
Increment 80
20 Increment 100

30

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-5 m/min
Suction switch ratio, kmax/kmin = 100

Figure 4.10: Precipitation into tension saturated column,


with suction switch.

114
0 Increment 0
Increment 10
10 Increment 25
Increment 50
20 Increment 80
Increment 100
30

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-4 m/min
Suction switch ratio, kmax/kmin = 100

Figure 4.11: Precipitation into tension saturated column,


with suction switch.

115
0 Increment 0
Increment 10
10 Increment 25
Increment 50
20 Increment 80
Increment 100
30

40

50
Depth (m)
60

70

80

90

100
200 100 0 -100 -200 -300 -400 -500 -600 -700 -800
Accumulated Fluid stress ( kPa ) ( tension = positive )

ks = 3.4E-5 m/min
Infiltration = 3.4E-3 m/min
Suction switch ratio, kmax/kmin = 100

Figure 4.12: Precipitation into tension saturated column, with


suction switch.

116
100 Increment 0

Increment 1

Increment 6
92
Increment 18

84
Depth, m

76

68

60
200 100 0 -100 -200 -300
Accumulated fluid stress ( tension positive )

Figure 4.13: Element size sensitivity analysis, 2m elements.

117
100 Increment 0

Increment 1

Increment 6
92
Increment 18

84
Depth, m

76

68

60
200 100 0 -100 -200 -300
Accumulated fluid stress ( tension positive )

Figure 4.14: Element size sensitivity analysis, 10m elements.

118
100 Increment 0

Increment 1

Increment 6
92
Increment 18

84
Depth, m

76

68

60
200 100 0 -100 -200 -300
Accumulated fluid stress ( tension positive )

Figure 4.15: Element size sensitivity analysis, 0.5m elements.

119
100 Increment 0

Increment 1

Increment 6
92
Increment 18

84
Depth, m

76

68

60
200 100 0 -100 -200 -300
Accumulated fluid stress ( tension positive )

Figure 4.16: Element size sensitivity analysis, 0.25m elements.

120
4m

4m
Position of sections shown in later figures

16m

24m

121
24m

60m

Figure 4.17: Finite element mesh for tension-saturated slope


analyses.
Figure 4.18a: Increment 0
Contour values
-25 kPa +50 kPa
0 kPa +100 kPa

+25 kPa +150 kPa

Note: Tension positive


0k
Pa
co
unt ntour
il th a
is p t surf
oin ace
t

Figure 4.18b: Increment 40

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 10 kPa, with permeability ratio of 100
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )
NOTE: Tension positive

Figure 4.18: Contours of pore water pressure in tension-


saturated slope, subject to precipitation, example 1.

122
0k
Pa
co
unt ntour
il th a
is p t surf
oin ace
t

Figure 4.18c: Increment 80


Contour values
-25 kPa +50 kPa

0 kPa +100 kPa

+25 kPa +150 kPa


Note: Tension positive
0k
Pa
co
unt ntour
il th a
is p t surf
oin ace
t

Figure 4.18d: Increment 120

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 10 kPa, with permeability ratio of 100
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )
NOTE: Tension positive

Figure 4.18: Contours of pore water pressure in tension-


saturated slope, subject to precipitation, example 1.

123
Increment 0
Accumulated
Increment 40
fluid stress
( kPa ) -100 Increment 80
Increment 120
-70

-40

-10 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 Distance (m)

20

50

80

110

140

170

200

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 10 kPa, with permeability ratio of 100
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )

NOTE: Tension positive

Figure 4.19: PWP across section of tension-saturated slope,


subject to precipitation, example 1.

124
Increment 0
Accumulated
Increment 40
fluid stress
( kPa ) -100 Increment 80
Increment 120
-70

-40

-10 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 Distance (m)

20

50

80

110

140

170

200

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 10 kPa, with permeability ratio of 100
Precipitation rate 3.3E-7 m/s ( THV = 0.0 kPa )

NOTE: Tension positive

Figure 4.20: PWP across section of tension-saturated slope,


subject to precipitation, example 2.

125
Phr
eat
ic s
urf
ace
at g
r oun
d sur
fac
e

Figure 4.21a: Increment 40 Contour values


-25 kPa +50 kPa

0 kPa +100 kPa


+25 kPa +150 kPa
Phr
eat Note: Tension positive
ic s
urf
ace
at g
r oun
d sur
fac
e

Figure 4.21b: Increment 120

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 10 kPa, with permeability ratio of 100
Precipitation rate 3.3E-7 m/s ( THV = 0.0 kPa )
NOTE: Tension positive

Figure 4.21: Contours of pore water pressure in tension-


saturated slope, subject to precipitation, example 2.

126
Increment 0
Accumulated
Increment 40
fluid stress
( kPa ) -100 Increment 80
Increment 120
-70

-40

-10 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 Distance (m)

20

50

80

110

140

170

200

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 100 kPa, with permeability ratio of 10000
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )

NOTE: Tension positive

Figure 4.22: PWP across section of tension-saturated slope,


subject to precipitation, example 3.

127
Figure 4.23a: Increment 40 Contour values
-25 kPa +50 kPa
0 kPa +100 kPa
+25 kPa +150 kPa
Note: Tension positive

Figure 4.23b: Increment 120

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 100 kPa, with permeability ratio of 10000
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )
NOTE: Tension positive

Figure 4.23: Contours of pore water pressure in tension-


saturated slope, subject to precipitation, example 3.

128
Increment 0
Accumulated
Increment 40
fluid stress
( kPa ) -100 Increment 80
Increment 120
-70

-40

-10 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 Distance (m)

20

50

80

110

140

170

200

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 1000 kPa, with permeability ratio of 100
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )

NOTE: Tension positive

Figure 4.24: PWP across section of tension-saturated slope,


subject to precipitation, example 4.

129
Figure 4.25a: Increment 40
Contour values
-25 kPa +50 kPa
0 kPa +100 kPa
+25 kPa +150 kPa

Note: Tension positive

Figure 4.25b: Increment 120

kmax : 3.4E-7 m/s


Suction switch varies p1 = 1 kPa, p2 = 1000 kPa, with permeability ratio of 100
Precipitation rate 6.8E-8 m/s ( THV = 0.0 kPa )
NOTE: Tension positive

Figure 4.25: Contours of pore water pressure in tension-


saturated slope, subject to precipitation, example 4.

130
Increment 0
Accumulated
Increment 40
fluid stress
( kPa ) -100 Increment 80
Increment 120
-70

-40

-10 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 Distance (m)

20

50

80

110

140

170

200

kmax : 3.4E-8 m/s


Suction switch varies p1 = 1 kPa, p2 = 1000 kPa, with permeability ratio of 1000
Precipitation rate 3.3E-8 m/s ( THV = 0.0 kPa )

NOTE: Tension positive

Figure 4.26: PWP across section of tension-saturated slope,


subject to precipitation, example 5.

131
Figure 4.27a: Increment 40 Contour values
-25 kPa +50 kPa

0 kPa +100 kPa


+25 kPa +150 kPa
Note: Tension positive

Figure 4.27b: Increment 120

kmax : 3.4E-8 m/s


Suction switch varies p1 = 1 kPa, p2 = 1000 kPa, with permeability ratio of 1000
Precipitation rate 3.3E-8 m/s ( THV = 0.0 kPa )
NOTE: Tension positive

Figure 4.27: Contours of pore water pressure in tension-


saturated slope, subject to precipitation, example 5.

132

You might also like