V Congreso Chileno de Ingeniería Geotécnica

RESIDUAL SOILS – WITH SPECIAL REFERENCE TO ALLOPHANE CLAYS

L.D. Wesley
University of Auckland

INTRODUCTION

Soil mechanics grew up in northern Europe and North America, and most of its concepts
regarding soil behaviour developed from the study of sedimentary soils. In fact, most of the early
concepts came from the study of remoulded sedimentary soils and involved investigating the
influence of stress history on their behaviour, in the belief that this was simulating the influence
of stresses which soils may be subject to during their formation processes. Most text books on
soil mechanics and university courses on the subject place considerable emphasis on stress
history – soils tend to be divided into normally consolidated and over-consolidated on this basis,
and behavioural frameworks are developed around this stress history concept.
This might be all very well if all soils were sedimentary soils. This of course is clearly not the
case. Large areas of the earth (including large areas in the North Island of New Zealand) consist
of residual soils, and the application of concepts coming from sedimentary soils may or may not
be relevant to these soils. It is interesting to note that very few text books, and probably very few
university courses on soil mechanics, even mention residual soils, let alone give an adequate
account of their properties.

RESIDUAL SOIL:
-produced by physical and chemical
weathering of underlying rock.

Erosion by rainfall
and run-off.
Soil Transport by stream Re-depositoin in layers
and river. in lakes or the ocean.

Delta deposits
Rock Sea or lake level

SEDIMENTARY SOIL
- later tectonic movement may
raise this above sea level

Fig. 1 Diagrammatic representation of soil formation processes.

FORMATION PROCESSES

Fig.1 shows diagrammatically the physical processes that to the formation of sedimentary and
residual soils. Residual soils are formed directly from the physical and chemical weathering of
the parent material, normally rock of some sort. Sedimentary soils are formed by a depositional
process, normally in a marine or lake environment. Fig. 2 is an attempt to summarise the factors
involved in the formation processes that influence the properties of the two soil types.
Sedimentary soils are seen to undergo a various additional processes beyond the initial physical
 V Congreso Chileno de Ingeniería Geotécnica

and chemical weathering of the

Parent Material
parent rock. It might appear from
(normally rock)
this diagram that the factors involved
in the formation of sedimentary soils
are more complex than those
Physical and chemical weathering
involved in forming residual soils.
There is some truth in this, but in
Possible unloading due to erosion
practice two important factors lead
to a degree of homogeneity and
predictability with sedimentary soils
that is absent from residual soils. Erosion and transportation
These factors are Residual soil
Re-deposition
(a) The sorting process which take
place during erosion,
transportation and deposition SEDIMENTARY SOIL
of sedimentary soils tend to (Young)
produce homogeneous
deposits. Self loading, causing compression (consolidation)
(b) Stress history is a prominent
factor in determining the Possible erosion, causing unloading
behavioural characteristics of
sedimentary soils, and leads to Possible further chemical & physical processes
the convenient division of eg secondary consolidation, cementation, hardening,
these soils into normally and leaching, producing structural effects such as
over consolidated materials.
apparent over-consolidation, sensitivity.
The absence of these factors with
residual soils means that they are
generally more complex and less SEDIMENTARY SOIL
capable of being divided into tidy (Old)
categories or groups.
Fig. 2 Soil formation factors influencing
It is perhaps helpful to consider that soil behaviour.
the behaviour of a soil, whether
residual or sedimentary, is dependent
on two factors, or two groups of factors. These are, firstly the nature of the soil particles
themselves (i.e. their size, shape, and mineralogical composition) and secondly, the particular
state in which these particles exist in the ground. For convenience, these factors can be referred
to respectively as composition and structure.

With sedimentary clays, the influence of composition is well known – kaolinite group clays are
relatively “inert” with consequent low shrinkage/swell characteristics, relatively low
compressibility etc, while montmorillinite clays are highly active and of opposite characteristics
to the kaolinite group. Notwithstanding the influence of mineralogy, by far the most important
“attribute” of sedimentary clays in their undisturbed state (at least according to conventional soil
mechanics) is their stress history i.e. whether they are normally consolidated or over-
consolidated. This is generally given greater importance in the literature than either mineralogy
or structure.

With residual, soils, mineralogy remains an important influence, but stress history is not a
concept which has much if any relevance. The physical and chemical weathering processes that
 V Congreso Chileno de Ingeniería Geotécnica

form these soils produce particular types of clay minerals, and particular “structures” i.e.
particular arrangements of the particles, and possibly bonding or cementing effects between
particles. These influences are infinitely more important than stress history. The terms normally
consolidated and overconsolidated are therefore not directly relevant to residual soils.

GROUPING AND CLASSIFICATION OF RESIDUAL SOILS

Various attempts have been made to group or classify residual soils, but none are particularly
useful. Some, such as that of the British Geological Society (1990) make use of soil science
classifications and are not very useful for engineering purposes. Terms such as vertisols,
andosols, etc are not normally meaningful to engineers, and the variation in properties within
these groups is likely to be so large as to make the grouping of little relevance.

Focussing on the two factors discussed above, namely mineralogical composition and structure,
provides a basis for dividing residual soils into groups that can be expected to have fairly similar
engineering properties. Starting with mineralogy, the following groups can be established:

(a) Soils without a strong mineralogical influence

(ie those containing low activity clays)
Many residual soils fall into this category, especially those derived from the weathering of
sandstones, or igneous rocks such as granite. These soils are likely to be fairly coarse
grained with a small clay fraction. Structure is likely to be an important concept in
understanding the behaviour of these soils. The weathered granite soils of Hong Kong and
Malaysia fall into this group.

(b) Soils with a strong mineralogical influence, from “conventional” clay minerals
(ie those containing high activity clays)
One very important worldwide group comes into this category – the “black cotton” soils or
"vertisols", also called Houston Black Clay in Texas, Tropical Black Earths of Australia,
“Tirs” of Morocco etc. The predominant clay mineral is smectite, a group of which
montmorillinite is a member.
These black cotton soils are highly plastic, highly compressible and of high shrink/swell
potential. Structural effects are almost zero with these soils. They normally form in poorly
drained areas, and have poor engineering properties.

(c) Soils with a strong mineralogical influence, coming from special clay minerals not found
in sedimentary clays.
The two most important clay minerals found only in certain residual soils (especially
tropical residual soils of volcanic origin) are halloysite and allophane. These are both
silicate clay minerals. Apart from silicate minerals, tropical soils may contain non-silicate
minerals (or “oxide” minerals), in particular the hydrated forms of aluminium and iron
oxide, gibbsite and goethite. The most unusual of these minerals, in terms of understanding
soil behaviour is allophane.

Soils of Group (c) which contain these unusual minerals include:

(i) Tropical red clays – predominant mineral is halloysite but may also contain kaolinite, with
gibbsite and goethite.
Halloysite particles are generally very small in size but are of low activity, and soils
containing halloysite as the predominant mineral generally have good engineering
properties.
V Congreso Chileno de Ingeniería Geotécnica
V Congreso Chileno de Ingeniería Geotécnica
 V Congreso Chileno de Ingeniería Geotécnica

Red clays generally form in well drained areas in a tropical climate having a wet and dry
season. Red clays may be referred to as lateritic soils or as latosols. There is a wide range
of engineering properties found in red clays, but they should not be confused with laterite
itself.

(ii) Volcanic ash soils (or andosols or andisols)

These are found in many tropical and sub-tropical countries (including New Zealand) and
are formed by the weathering of volcanic “glass”. The predominant clay mineral is
allophane (frequently associated with another mineral called imogolite).

(iii) Laterites
The term laterite is used very loosely, but should refer to deposits in which weathering has
reached an advanced stage and has resulted in a concentration of iron and aluminium
oxides (the sesquioxides gibbsite and goethite), which act as cementing agents. Laterials
therefore tend to consist of hard granules formed by this cementing action; they may range
from sandy clays to gravels, and are used for road sub-bases or bases.

Table 1 shows this grouping system for residuals soils, and Table 2 attempts to list some of the
more distinctive characteristics of these soil groups and indicates the means by which they may
possibly be identified.

Following on from mineralogy, the next characteristic which should be considered is structure,
which refers to specific characteristics of the soil in its undisturbed (in situ) state. Structure can
be divided into two categories:
(a) Macro-structure, or discernible structure.
This includes all features discernible to the naked eye, such as layering, discontinuities,
fissures, pores, presence of unweathered or partially weathered rock and other relict
structures inherited from the parent rock mass.
(b) Micro-structure, or non-discernible structure.
This includes fabric, inter-particle bonding or cementation, aggregations of particles, pores
etc. Micro-structure is more difficult to identify than macro-structure, although it can be
inferred indirectly from other behavioural characteristics such as sensitivity. High sensitivity
indicates the presence of some form of bonds between particles which are destroyed by
remoulding.

This grouping system is intended to help geotechnical engineers find their way around residual
soils, and to draw attention to the properties likely to be of most significance for geotechnical
engineering. It is not intended to perform a function as a rigorous classification system. Some
comments on local or Southeast Asian soils may be helpful at this stage.
Weathered Waitemata clays (Auckland, NZ) : This is an example of a group which
doesn't fit comfortably in any one category and this in itself tells us something about these
clays. Some Waitemata "clays" are essentially silts, and are not strongly influenced by clay
minerals - they belong to Group A. Others are very highly plastic clays, resulting from the
presence of smectite (montmorillonite) minerals - and belong in Group B. The two types
may occur in quite close proximity ie in interbedded layers. It appears that the weathering
process in this case is not actually creating the clay minerals; it is simply destroying the
weak bonds which "lock" the clay minerals into the parent material. Waitemata clays may or
not exhibit macro-structure as well as micro-structural effects.
Weathered greywacke soils (Wellington, NZ): These probably belong in Group A, as their
properties are not strongly influenced by their mineralogical content. They are likely to
exhibit significant macro-structure effects, dependent on their degree of weathering.
 V Congreso Chileno de Ingeniería Geotécnica

Weathered granite soils (Worldwide): These also belong to Group A, and exhibit macro-
structural effects - from joints etc and presence of "floating" un-weathered rock boulders.
Volcanic ash (allophane) soils (Worldwide): These clearly belong to Group C. They are
very strongly influenced by their mineral composition. They are unlikely to exhibit
significant macro-structure, but may exhibit some micro-structure - significant sensitivity for
example.
Tropical red clays (Many tropical countries): These also belong to Group C. Those found
in the island of Java, Indonesia (with which the author is familiar) are rather unusual in that
they exhibit neither macro-structure nor micro- structure, except when the weathering is not
far advanced. In this case they may show traces of the structure of their parent material.

GEOTECHNICAL ENGINEERING IN RESIDUAL SOILS

In the following sections some comments will be made on issues of direct relevance to
geotechnical engineers, namely foundation design, slope stability and compaction. They are not
comprehensive and should not be taken as generalisations applicable to all residual soils.

FOUNDATION DESIGN

Consolidation behaviour
1.6
(a) Magnitude
(stress/deformation curves)

Fig. 3 show typical consolidation test

Void ratio

results from one residual soil type - 1.4

the tropical red clay found in Java,
Indonesia. Although it is standard
practice to plot consolidation test
results as void ratio versus log
pressure graphs, it is often informative 1.2
to also plot them as direct (a) log scale
compression graphs using linear
10 100 1000
scales. The lower part of Fig 3 shows Pressure (kPa)
the linear plots. 0 100 200 300

The results show the following

points:
Compression %

(i) Conventional graphs (e-logp)

2
suggest the clays behave as
moderately over-consolidated
soils, although there is no clearly
defined "pre-consolidation" or
"vertical yield" pressure. It 4
appears to be somewhere (b) linear scale
between 100 kPa and 500 kPa.
5
(ii) When plotted using a linear scale,
the picture is quite different. The Fig 3 Oedometer test results from a
curves are reasonably close to tropical red clay
linear, especially over the
 V Congreso Chileno de Ingeniería Geotécnica

pressure range likely to be of engineering interest, generally about 0 to 200 kPa. The
evidence of a "yield" stress has largely disappeared.

It is not suggested that the curves in Fig. 3 are representative of residual soils in general. They
are presented primarily to illustrate that the standard e-log(p) graph can be quite misleading and
may imply the existence of "pre-consolidation" or "yield" pressures when no such pressure
exists. With residual soils (and possibly also with sedimentary soils) it is generally desirable to
plot consolidation test results using a linear scale for pressure as well as the normal log scale
before drawing any conclusions about the behaviour of the soil. Further examples will be
presented during this seminar which show that some residual soils show quite distinct "yield"
pressures, while others show steadily increasing stiffness with stress level, and some demonstrate
almost linear behaviour.

Fig. 4 is presented to show the influence of 5.0

remoulding on compression behaviour for Allophane clay
(Indonesia)
three different residual soils. These are
4.5
respectively an allophane clay, a tropical
red clay, and a silt derived from weathered
Waitemata sandstone. Consolidation 4.0
Void ratio

curves are given for the soil in its Undisturbed

Remoulded
undisturbed state, its remoulded state, and Slurry
after mixing it with water to form a slurry. 3.5
These last curves can be regarded as the
"virgin" consolidation lines for the soil in
its completely remoulded state. It is seen 3.0
that with the allophane clay and the
Waitemata silt, remoulding results in a 2.5
very significant change in the compression 10 100 1000 5000
curve. These soils clearly have a relatively Pressure (kPa)
stiff structure in their undisturbed state 10 100 1000 5000
1.8
which is destroyed by re-moulding (or "de- Undisturbed
Remoulded
structuring" to use the in vogue term for Tropical red clay
Slurry
(Indonesia)
this effect). The red clay on the other hand 1.6
shows almost no change in behaviour after
remoulding. This is often the case with red
1.4
clays. They appear to exist naturally in a
dense unstructured state close to their
Void ratio

Plastic Limit, and remoulding thus has 1.2

little or no effect on them.
Silt (weathered
With regard to the estimation of settlement Waitematas)
1.0
magnitude, there are two procedures
commonly used in soil mechanics. The
first is to use the parameters Cc and Cs 0.8
which are obtained from the e – log (p)
plot, and the second method is to use mv
values. For soils which give an 0.6
approximately straight line on a linear
stress/compression plot the use of mv
Fig. 4 Influence of remoulding
seems most appropriate. The choice of
method is a matter for individual on e-log(p) graphs
judgement, based primarily on the actual
 V Congreso Chileno de Ingeniería Geotécnica

soil behaviour in consolidation tests. With residual soils the mv parameter often seems more
appropriate than the Cc or Cs parameters.

It should be noted that for settlement estimates with sedimentary soils, there are various
empirical constructions or corrections for improving the accuracy of estimates. The best known
are probably the Schmertman construction and the Skempton and Bjerrum method. Both these
methods are based primarily on stress history concepts and are not intended for residual soils.
Therefore the use of these methods with residual soils is highly questionable. There are no
established procedures available for correcting consolidation curves for residual soils to allow
for sample disturbance (such as the Schmertman method for sedimentary soils) and hence it is
very important to obtain good quality undisturbed samples for consolidation tests.

One further factor which should be PORE WATER PRESSUE

appreciated when attempting to predict Negative Positive
settlement magnitudes of foundations on
residual soils is that the initial stress state
in the ground is likely to be unknown if Equilibrium pore water
tension for a sealed surface
the water table is at some depth below
the surface. The pore pressures above Possible seasonal
variation Wet weather value
the water table will be negative (ie in a
suction or "tension" state), and likely to
vary between winter and summer. During Dry weather value
ZONE OF NEGATIVE
prolonged dry periods the suction value PORE PRESSURE
may be quite large. This means that the
Water table
initial effective stress in the ground is not
know and likely to vary between winter ZONE OF POSITVE
and summer. This is a fact commonly PORE PRESSURE
ignored in routing settlement effects.
This situation is illustrated in Fig. 5.
Hydrostatic pore
(b) Consolidation Rate pressure
Consolidation rates with residual soils
tend to be rather faster than with
sedimentary soils; as evidenced by their
behaviour, both in the laboratory and in
the field. This appears to be due to Fig.5 Pore water pressure state above
higher permeability associated with their and below the water table
undisturbed structure. In consolidation
tests the rate of pore pressure dissipation
may be too fast to allow reliable determination of the coefficient of consolidation. This is
demonstrated in Fig.6 which shows standard graphs of compression versus root time for the
loading increment 100 kPa to 200 kPa for three residual soils. The normal straight line section,
which is used to determine t90 is not clearly defined. Hence, the estimation of cv is problematical.
It is usually found that at higher stresses the graphs become more linear; the higher stress tends
to destroy the original structure and lower the permeability.

It should be appreciated that there is an upper limit to the value of coefficient of consolidation
which can be measured in a conventional consolidation test. Analysis shows that the highest
value of cv which can be reliably measured with a 19mm thick sample is about 0.1m2/day
(=0.012cm2/sec.). Soils with cv values greater than this will not show distinct straight lines on a
conventional compression versus root time plot. If reliable values of cv are required for soils
 V Congreso Chileno de Ingeniería Geotécnica

√ time √ min. time min

0 2 4 6 0 1 2 3 4 5 6
Auckland clay
(Weathered sandstone)
Volcanic ash soil
Weathered Waitemata clay 1

Compression (%)
Tropical red clay
2 Undisturbed

2
Compression (%)

Remoulded

4
3

time min
0 1 2 3 4 5 6
6 Volcanic ash soil
(allophanic)
Undisturbed
1

Compression (%)
8

2 Remoulded

10
3

Fig 6 Typical root time graphs from

residual soils Fig. 7 Influence of remoulding
on consolidation rate.

which behave in this way, it is probably best to use a different method of measurement, such as a
pore pressure dissipation test in a triaxial cell.

Values of cv for the three soil types in Fig 6 cover a wide range as follows:

Waitemata silts and clays: 0.01 to 10m2/day

Indonesian red clays: 0.07 to 0.7m2/day
Volcanic ash soils: 0.01 to 200m2/day

These values lie above and below the value of 0.1m2/day that can be measured in the standard
consolidation test.

Fig. 7 illustrates the influence which remoulding may have on consolidation rate. The two curves
are for the same stress increment, from 100kPa to 200kPa. Remoulding destroys the soil
structure responsible for its high permeability and the much slower rate of consolidation
produces the normal straight line on the root time plot.

Shear strength
It is not possible to make many categorical statements regarding the shear strength of residual
soils; the following observations are generalisations and should be treated with some caution. It
is reasonably true to assert (excluding montmorillonite "black cotton" soils) that the shear
strength of residual soils, whether expressed as undrained shear strength or effective strength
parameters, is generally higher than that of sedimentary soils. It is rare for the undrained strength
to be less than about 75kPa, and is generally between 100 and 200 kPa. Their φ′ values are
 V Congreso Chileno de Ingeniería Geotécnica

generally above 30o, and they have significant values of the cohesion intercept c′. In the case of
some allophane rich volcanic ash soils both the peak φ′ and residual φ′ values may be higher than
35o. Figure 8 shows the results of triaxial tests on two residual soils; the first is for volcanic ash
soils and the second for a clay (known as Middle clay) derived from weathered sandstone.

300 o
o
VOLCANIC ASH CLAYS 4
=3
5 / =3
/
φ φ
a Pa
kP 14k
34 / =
c= c
/
Shear strength (kPa)

200

100

0 100 200 300 400 500

Normal stress (kPa)

600
“MIDDLE CLAY”
- FROM WEATHERED SANDSTONE
Shear strength (kPa)

o o
4 2 5
=3 φ=
/
400 /
φ a
k P
Pa /
=5
54k c
/
c =

200

0 200 400 600 800 1000

Normal stress (kPa)

Figure 8. Triaxial test results from two types of residual soils.

The results from volcanic ash soil in the upper figure show a relatively small variation in the
shear strength; this is not surprising since volcanic ash soils are generally free of discontinuities
and are of reasonably uniform composition. The lower figure shows the influence of structural
defects (macro- structure) in the parent rock that are still present in the soil. It is clear that in the
latter case it would be almost impossible to infer reasonable design parameters from results of
this sort.
 V Congreso Chileno de Ingeniería Geotécnica

Bearing Capacity
As mentioned above, the permeability and consolidation rates with residual soils are generally
high, and in situations where residual soils are subject to external loading by the construction of
foundations it is likely that generated pore pressures will dissipate almost immediately and the
soils will remain in the drained state. This means that design using undrained strength will be
conservative, as there will be some increase in strength as the load on the foundation increases.

However, this is not an argument against the use of undrained strength to estimate the bearing
capacity of the soil for foundation design purposes. During rapid load application, such as during
seismic loading, the soil will still behave in an undrained manner, and for this reason especially,
design should be based on undrained strength. There are also strong practical arguments in
favour of using undrained strength, as this can be measured relatively easily and reliably. Both
field methods (e.g. Dutch penetrometer) and laboratory methods (unconfined compression or
vane test) can be used to obtain reliable undrained strength values, whereas the measurement of
drained strength parameters c′ and φ′ is more difficult and less certain.

SLOPE STABILITY
There are several aspects of the stability of residual soil slopes that are of particular interest to
the geotechnical engineer. These include the following:

(a) slopes in residual soils (excluding "black cotton" soils) generally remain stable at much
steeper angles than those in most sedimentary soils. Slopes of 450 or steeper are not
uncommon, and cuts can often be made as steep as 600 without danger of slip failure.
(b) slope failures in residual soils, especially when steep slopes are involved are unlikely to be
deep seated circular failures. They are more likely to be relatively shallow, with fairly
planar failure surfaces. However, the volume of material involved may still be very large.
(c) slips and landslides in residual soils generally occur during periods of heavy rainfall, and
are the result of temporary increases in the pore water pressure in the slope.
(d) the value of c′ is usually significant and is considered to be due to some form of weak
bonds between particles
(e) the residual strength is likely to be closer to the peak strength than is the case with many
sedimentary soils, especially in clays continuing allophane or halloysite
(f) with some (possibly the majority) residual soils, the presence of discontinuities may be the
governing factor.

Factors (c) and (f) are very important with respect to the use of analytical (slip circle) methods
for assessing stability. Factor (c) is particularly important; with sedimentary clays of low
permeability the pore pressures can be measured and the assumption made that they will remain
approximately the same for a long time. With residual soils, any measurement of pore water
pressure in the slope is valid only at the time it is made and is not relevant to long term stability
estimates. For such estimates it is the worst condition likely to occur in the future which is of
importance. Factor (f) is likely to dominate the behaviour of many cut slopes in residual soils,
and rule out the use of analytical methods. Fig.8 shows an example of such a soil. Only in very
rare situations is it likely to be possible to determine the location, orientation, and strength of
discontinuities with the degree of reliability needed for the use of analytical methods.

The rapid changes in pore water pressure that occur with residual soils mean that stability
analysis must be carried out in terms of effective stresses. The only exception to this might be
when an embankment is constructed on a residual soil; this situation is similar to a foundation
situation and undrained strength could be used.
 V Congreso Chileno de Ingeniería Geotécnica

Note: There is some evidence that pore water pressure in a slope will
only change significantly as a result of periods of heavy rainfall if the cv
value is greater than about 0.1m2/day. See Kenney and Lau: “Temporal
changes of groundwater pressure in a natural clay slope” Canadian
Geotechnical Journal Vol 21, 1984.

COMPACTION OF RESIDUAL SOILS

One last property of residual soils that has caused difficulties to engineers relates to their
compaction behaviour. There are two problems, as follows:
(a) The variability of residual soils may mean a large and rapid variation in optimum water
content within short distances in any borrow pit.
(b) Some compaction curves for residual soils, notably volcanic ash soils do not show peaks
indicating maximum dry densities and optimum water contents.

Neither of the above “problems” are real problems in the sense of indicating that residual soils
are more difficult to compact than sedimentary soils. If there is a problem, it is only in the
evaluation of the soils and the method to be adopted for specifying and controlling the
compaction. Many volcanic ash soils can be effectively compacted at water contents in the range
of 100% to 180%, a fact which geotechnical engineers are often reluctant to accept.

Industrial site Steel mill site

1.8 1.3
Pleistocene deposits Weathered basalt and ashes

1.2
Dry Density (gm/cm )
3

Ze
1.4 ro
Ze air
ro vo
air 1.1 id
sl
vo in
id e
sl
in
e
1.0 1.0

0.9

0.8

20 40 60 80 40 50 60 70
Water content (%) Water content (%)

Fig. 9 Compaction curves from residual soils on two sites near Auckland.

Figure 9 shows the results of compaction tests carried out on a number of different samples from
two sites involving residual soils. It is evident that there is a very wide range of optimum water
contents and maximum dry densities. Figure 10 shows the result of a compaction test on a
volcanic ash sample from Java, Indonesia. The test has first been carried out by drying the soil in
stages from its natural water content. The soil has then had water added to it after various
degrees of drying, and further compaction tests carried out. The results show the very flat
compaction curve obtained from the natural soil, and also the very significant influence which
drying has on the soil properties. Any value of optimum water content can be obtained by
varying the extent of pre-drying.
 V Congreso Chileno de Ingeniería Geotécnica

1.2
Natural
Air dried
Oven dried
Dry density g/cm3
1.0
Air dried to 65 %
Ze
ro
air
0.8 vo
ids

0.6

0.4
20 40 60 80 100 120 140 160 180
Water content (%)

Fig.10 Compaction test result from a volcanic as soil (Indonesia).

The behaviour illustrated in Figs. 9 and 10 means that the control of compaction by the
conventional method of specifying dry density and water content limits based on standard
compaction tests is very difficult. Alternative methods of compaction control have been
developed for such soils which overcome the above difficulties. The simplest method is that
which is based on undrained strength and air voids criteria and is described by Pickens (1980).

The principle of the method is to

specify a minimum value of shear
Sh
ea

strength (commonly 100 kPa to

rs

150 kPa) and a maximum value of Shear strength

tre

air voids (commonly 8 to 12%) for limit

ng

the compacted soil. These values

th

can be varied according to the

Shear strength
nature of the job and the soil or
weather conditions at the site.
Dry density

Ze
ro
air
Figure 11 illustrates the principle vo
ids
of the method in relation to the
conventional method based on
Dry density limit
water content and maximum dry
density. The requirement of a
minimum strength means that the Air
voi
soil must not be too wet, and the ds lim
Water content
requirement that the air voids not it
limits
exceed a certain value means that
the soil must not be too dry.
Water content
The method is easy to use and
Fig. 11. Compaction control limits using
control testing involves density and
shear strength and air voids criteria.
water content measurements in the
usual way. The values obtained are
not significant in themselves; they are simply used to calculate the value of the air voids. At each
control point, measurements are also made of shear strength. The simplest method of doing this
is by using a hand operated shear vane, such as the “Pilcon” vane.
 V Congreso Chileno de Ingeniería Geotécnica

The actual values of optimum water content and maximum dry density of the soil do not need to
be known, and it is not essential to carry out normal compaction tests at all. Such tests may
however be useful in order to know whether much drying of the soil will be needed in order to be
able to effectively compact it.

COMMENTS ON NORMAL IDENTIFICATION AND CLASSIFICATION TESTS

The tests normally used as a starting point in the evaluation and classification of soils are particle
size measurement and the Atterberg limits. The applicability of such tests to residual soils is a
matter of some contention within the profession; it is useful therefore, to examine the arguments
put forward to suggest that these tests are of less relevance to residual soils than to sedimentary
soils. The arguments are as follows:

(a) Classification test are carried out on the remoulded soil, and since remoulding destroys
the important structural features which dominate the behaviour of many residual soils the
tests indicate very little about undisturbed behaviour.
(b) Some residual soils contain a large proportion of coarse particles, and since Atterberg
limits are carried out on the fine fraction only, they do not give a good indication of the
properties of the soil as a whole.
(c) The particles of some residual soils are of a weak and fragile nature and are broken down
into smaller particles during testing.
(d) The results of these tests are influenced by pre-drying the soil, and the plasticity limits
are also dependent on the amount of mixing carried out prior to testing.
(e) Empirical relationships between either particle size or Atterberg limits and other
engineering properties have been developed from sedimentary soils and are not
necessarily valid for residual soils.

There is some validity in all of these arguments, but we should be careful in our evaluation of
them; they are certainly not valid for all residual soils on a general basis. In the case of one
important residual soil group, namely the “vertisols” (or Black Cotton soils) it is likely that none
of these arguments is of any relevance at all.

Arguments (a) and (b) above are not peculiar to residual soils; they frequently apply also to
sedimentary soils, and in any case classification tests are frequently used for the evaluation of fill
materials in which case it is the properties of the remoulded soil which are required.

Argument (d), at least with respect to the influence of pre-drying the soil, is not a valid argument
against the use of classification tests, since there is no difficulty at all in carrying out the tests
without pre-drying the soil.

Argument (e) above is perhaps the most important question to be considered, especially with
respect to the Atterberg limits. It has been the author’s experience that with residual soils the
position which a soil occupies on the conventional Plasticity Chart provides a good indication of
properties - probably just as good as with sedimentary soils. Soils which plot well below the A-
line behave as silts while those which plot well above the A-line behave as clays. Fig 11 show
the position on the Plasticity Chart of the three most distinctive residual soils - the "Black
Cotton" soils, the tropical red clays, and the allophane clays.

Problems arise when attempts are made to relate specific soil properties, or classification
boundaries to one or other of the liquid and plastic limits. For example, the British classification
system (BS 5903: 1981) divides soils up into a number of categories based on the liquid limit.
 V Congreso Chileno de Ingeniería Geotécnica

150
e
- Lin
A
CLAY
100 SILTY CLAY
Plasticity Index

ils ite
)
” so ) ys SILT
on it e
llo
c ott tillon (ha
50 k o s soils
lac tm l ay a nic ash e)
“B mon red c lc
Vo lophan
( l (al
pi ca
o
Tr

0 50 100 150 200 250

Liquid Limit

Fig. 12 The Plasticity Chart and residual soils.

Such a division is not very relevant to residual soils. It is the position above or below the A-line
which is of most significance, especially with tropical residual soils.

Rather than a subdivision based on the liquid limit, a subdivision along the lines shown in Fig.
12 would be most relevant to residual soils. The lines drawn parallel to the A-line divide soils
into three types labelled clay, silty clay, and silt. Many residual soils behave as silty clays for
engineering purposes, and rightly fall into the category of silty clay on this chart. The more
distinctive residual soil types, such as "Black Cotton" soils, and allophane clays, would rightly
be classified as clays and silts respectively.

It should be noted that the influence of increased mixing (or even drying) of the soil on the
Atterberg limits is to move the point on the plasticity chart parallel to the A-line; hence if we use
distance above or below the A-line as our main criteria for evaluating soils this movement is not
of great significance. Hence argument (d) above is not very important.

Empirical Relationships based on particle size or Atterberg Limits.

There are some rather vague general relationships involving particle size and Atterberg limits,
and there are specific empirical relationships.

Among the general relationships is the understanding that as particle size decreases (or possibly
as L.L. increases) the properties of a soil become less favourable for engineering purposes. This
is generally true (or held to be true) if a particular soil type is being considered. This
understanding may well apply to many residual soils, but there is very considerable evidence that
it does not apply to halloysite or allophane soils. Especially with allophane soils, there is no
evidence of decrease in strength or increase in compressibility with either decreasing particle
size or increasing L.L.

Specific empirical relationships would be those such as

Cc = 0.009 (L.L. – 10)

 V Congreso Chileno de Ingeniería Geotécnica

This relationship is for remoulded N.C. soils and thus has no relevance to engineering situations
in residual soils. In general, these types of relationships should hold for materials of conventional
clay mineralogy. For residual soils containing allophane or even halloysite they may not be
valid.

GENERAL REMARKS ON RESIDUAL SOILS

If there are lessons to be learnt from geotechnical engineering in residual soils, they are probably
the following:
1. Geotechnical engineers ought to have open minds about how soils may behave, and not
assume they will conform to preconceived patterns, especially when working with
residual soils.
2. In evaluating the engineering properties of soils we ought to first observe carefully their
behaviour in the field, before looking at their behaviour in laboratory tests.
3. While every effort should be made to develop theoretical or behavioural frameworks to
assist us in understanding and interpreting soil behaviour, we ought to recognise the
limitations of such frameworks, and not seek to make all soils fit into these frameworks.
4. Some well established procedures, such as the use of the e-log p plot for analysing
consolidation behaviour, are not necessarily appropriate for all soils, especially residual
soils.
5. With residual soils, the mode of formation is so varied that it is unrealistic to expect them
to fit into a single behavioural pattern.

THE SPECIAL PROPERTIES OF ALLOPHANE (VOLANIC ASH) CLAYS

Occurrence

There are substantial areas in the North Island where clays derived from the weathering of
volcanic ash occur. These clays tend to be rich in the clay mineral allophane, which gives them
rather unusual and unique properties. They are often referred to as “brown ash” by local
engineers. Whether all clays referred to as brown ash contain allophane is not known to the
writer; the term is used rather loosely and in some cases may be applied to clays that do not
contain allophane. The clays described here are those whose properties are influenced primarily
by their allophane content, and will be referred to as allophane clays. Similar clays occur in
many parts of the world, including Indonesia, The Philippines, Japan, Central and South
America, and Africa.

Formation

The formation and composition of allophane clay is complex, and most of the research and
literature on the subject comes from the discipline of soil science rather than soil mechanics.
This research and literature has grown enormously in the last two or three decades since the term
allophane first found its way into geotechnical literature, and it shows a number of new and
interesting findings. Firstly, it shows that allophane seldom occurs by itself. Instead, it is almost
invariably found with other clay minerals, especially a mineral called imogolite. It seems to be
almost inseparably linked to imogolite, and many papers on allophane are in fact on "allophane
and imogolite" rather than on allophane alone. Secondly, it shows that allophane is not strictly
amorphous, as early literature asserted. Both allophane and imogolite have some crystalline
structure, albeit of a very different nature to other clay minerals.
 V Congreso Chileno de Ingeniería Geotécnica

Allophane clays are derived primarily from the in situ weathering of volcanic ash, although they
may be derived from other volcanic material. This parent material may be either basic or acidic
in nature. It appears that the primary condition for allophane formation is that the parent material
be of non-crystalline (or poorly ordered structure) composition. Volcanic ash meets this criteria;
it is formed by the rapid cooling of relatively fine-grained pyroclastic material, the cooling
process being too rapid for the formation of well ordered crystalline structures. In the author’s
experience, the parent volcanic ash from which allophane clays are formed is generally in the
coarse silt to fine sand particle size range.

In addition to the above requirement of non-crystalline parent material, it appears that the
weathering environment must be well drained, with water seeping vertically downward through
the ash deposit. High temperatures also appear to favour or accelerate the formation of allophane
clays. Allophane clays may be very deep; in Indonesia the writer has encountered cuts in these
materials up to about 30m deep, while site investigation drilling has shown depths of up to
almost 40 metres. This thickness results from successive eruptions and associated ash showers,
with weathering progressing as the thickness grows. Examination of cut exposures in West Java,
Indonesia, shows the individual layer thickness to vary generally between about 100 and 300mm.

Structure

The precise structure of allophane clays is somewhat problematic. Their extraordinarily high
natural water contents and void ratios (described in the next section) clearly indicate an unusual
material, and call for an explanation in terms of either structure or chemical composition (or
both). Various explanations have been offered over the years. As mentioned above, allophane
has been described in the past as non-crystalline or amorphous, and “gell-like”. However,
electron microscopy studies over the past 10 years or so (Wada, 1989 and Jacquet, 1990) show
that the material in its natural state does have an ordered structure – consisting of aggregations of
spherical allophane particles with imogolite threads "weaving" among them, or forming
“bridges” between them.

50nm

Figure 13. Electron micrograph of allophane and immogolite (after Wada, 1989).
 V Congreso Chileno de Ingeniería Geotécnica

Figure 13 shows an electron micrograph of the material in its undisturbed state, (taken from
Wada, 1989). The concept of approximately spherical particles with thread-like structures
spanning between them appears to explain both the very high natural water content, and the
changes the material undergoes on remoulding. Remoulding appears to break up the
aggregations of particles and threads spanning between them and turns the material into a
homogeneous unstructured mass. This is generally accompanied by some loss of strength and an
increase in compressibility, as well as a reduction in permeability.

General comments on engineering properties

Before describing particular properties the point should be made that allophone clays are not
problem soils. There is still a belief among some geotechnical engineers that the presence of
allophone in a soil is something to fear or be concerned about. This should not be the case.
Observation of these clays in their natural environment shows them to perform remarkably well.
For example, terraced ricefields in allophone clay areas in many countries exist on slopes as
steep as 35o and almost up to 40o. They are permanently saturated by irrigation water flowing
from terrace to terrace. Many water retaining structures have been successfully constructed from
allophone clays. While they ought not to be a cause for concern, it is important that their special
properties be understood and taken account of in planning engineering projects.

Natural water content, void ratio, and Atterberg Limits

The natural water content of allophane clay covers a very wide range, from about 50% to 300%.
This corresponds to void ratios from about 1.5 to 8. It appears that water content is a reasonable
indication of allophane content – the higher the water content the greater the allophane content.
Atterberg Limits similarly cover a wide range, and when plotted on the conventional Plasticity
Chart invariably lie well below the A-line. This means that according to the Unified Soil
Classification System they are silts. However they do not display the characteristics normally
associated with silt – the tendency to become “quick” when vibrated and to dilate when
deformed. At the same time they are not highly plastic like true clays, so they do not fit
comfortably into conventional classification systems. Figure 14 shows a plot of the Atterberg
limits on the Plasticity Chart.

100

80
Plasticity Index

60 e
- lin
A
40

20

0 40 80 120 160 200 240

Liquid Limit

Figure 14. Atterberg limits of Allophane clays on the Plasticity Chart.

 V Congreso Chileno de Ingeniería Geotécnica

Influence of drying

Drying has a very important effect on allophane clays. Frost (1967) gave the first systematic
account of this effect for both air and oven drying on tropical soils belonging to the allophane
and halloysite group. He showed that clays from the mountainous districts of Papua New Guinea
with values of Plasticity Index ranging from about 30 to 80 in their natural state become non-
plastic when air or oven dried. Wesley (1973) describes similar effects from the allophane clays
of Java, Indonesia. The properties of the clay described in this paper apply to the clay in its
natural state, ie without air or oven drying, unless otherwise stated.

Identification of allophane clays

There are various techniques used by soil scientists to identify allophane: these are primarily X-
ray diffraction and electron microscopy. Such methods are not readily available to geotechnical
engineers For engineering purposes, sufficient indicators of the presence of allophane are the
following:
1) Volcanic parent material
2) Very high water contents
3) Very high liquid and plastic limits lying well below the A-line on the Plasticity
Chart
4) Irreversible changes on air or oven drying - from a plastic to a non-plastic material.

If all of these apply then the soil almost certainly contains a significant allophone content.

Stiffness and compressibility

Typical results from oedometer tests on undisturbed samples from Indonesia and New Zealand
are shown in Figures 15 and 16. Details of the samples are given in Table 1.

Sample Location Information Depth VoidWater Atterberg Limits

Number (m) RatioContent
(%) L.L. P.L. P.I.
I.3 Cipanunjang, Indonesia 1.0 4.76 161.2 187 149 38
I.4 Cipanunjang, Indonesia 3.0 4.72 156.1 179 139 40
I.6 Kamojang, Indonesia 12.2 5.34 190.7 *
NZ.4 Omata, New Zealand 7.5 2.58 81.3 *
NZ.6 Omata, New Zealand 7.0 3.20 114.1 *
NZ.7 Omata, New Zealand 5.1 2.49 88.5 *
* Atterberg limits not measured on these samples

Table 1. Details of samples used for oedometer tests.

Figure 15 shows the results as conventional e-log(p) graphs and Figure16 as compression versus
stress on a linear scale. The e-log(p) curves suggest that all the samples have similar
compressibility characteristics with “pre-consolidation” pressures of varying magnitude.
However, when plotted using a linear pressure scale this is no longer the case: only some of the
samples show an apparent pre-consolidation pressure. This arises from the structure of the soil
created by the weathering process, and is perhaps best described as a vertical yield pressure.
Why some samples show a yield pressure and others do not is not known, though it may be
related to the original denseness of the parent material.
 V Congreso Chileno de Ingeniería Geotécnica

Pressure (kPa)
10 100 1000 5000

5
Pressure (kPa)
0 500 1000 1500 2000

Sample I.3
Sample I.4
4 Sample I.6
Sample N.Z.1

Compression (Vertical strain %)

Sample N.Z.6 10
Void Ratio

Sample N.Z.7

20
3

30
Sample I.3
2 Sample I.4
Sample I.6
Sample N.Z.1
Sample N.Z.6
40 Sample N.Z.7

1
Figure 16. Oedometer tests showing
Figure 15. Oedometer test results as compression versus pressure on a linear scale.
e-log(p) plots.

These graphs illustrate two important points. Firstly, to gain a clear picture of the consolidation
behaviour it is necessary to plot the results using a linear scale as well as a log scale. Secondly,
the portion of the graph of interest in foundation design is often close to linear with respect to
pressure, and favours the use of the linear parameter mv (or constrained modulus D) for
settlement calculations rather than the log parameters Cc and Cs.
20
(a) 0 -200 kPa

15
Constrained modulus D (MPa)

10

0
15
(b) 1600 - 2000 kPa

10

5
1 2 3 4 5 6
Initial void ratio
 V Congreso Chileno de Ingeniería Geotécnica

Figure 17. Constrained modulus (D) versus initial void ratio.

It is of interest to note that for these clays there does not appear to be any relationship between
the initial void ratio and compressibility. Fig. 17 shows the constrained modulus D measured
when the sample is loaded from 0 to 200 kPa, and again between 1600 kPa and 2000 kPa, plotted
against the initial void ratio. The data shows considerable scatter, but there is no clear trend
towards higher compressibility with increase in void ratio from 2 to nearly 6.

Time min 50
0 2 4 6 7
New Zealand samples
- undisturbed

10

Coefficient of consolidation (cm2/sec)

20
12
Total compression (%)

80
-2

1
56
0k

40
Indonesian samples
Pa

16 - undisturbed
0-
32 0.1
60 0k
Pa

0.01
80
16-32 kPa
0.001
100
New Zealand samples
- remoulded
0.0004
20 100 1000 2000
Pressure (kPa)
Figure 18. Typical root time plots
from oedometer tests. Figure19. Summary of cv values from
pore pressure dissipation tests.

Figure 18 shows typical root time plots from oedometer tests. At low stress increments the
consolidation rate is clearly very rapid but becomes progressively slower as the stress level rises.
To investigate this effect in more detail, pore pressure dissipation tests were carried out using a
triaxial cell. Two samples from New Zealand and two from Indonesia were tested

A summary of the cv values obtained from these dissipation tests is shown in Fig.19. It is seen
that the cv value decreases by approximately four orders of magnitude as the stress increases
from 50 to 1000 kPa. With the New Zealand samples, the tests were repeated after remoulding
the soil. It is seen that the cv value is then consistently low and close to the end value from the
undisturbed samples. With the Indonesian samples, permeability measurements were also made
between each consolidation stage; the results showed an identical trend to the cv values. Figure
19 shows that remoulding the soil apparently destroys the open structure of the undisturbed soil,
which is believed to account for the high permeability.

As noted earlier, with clays of this type it is not possible to determine reliable cv values from
conventional oedometer tests. The drainage path length is too short for pore pressure dissipation
to control the deformation rate. The upper limit of the cv value which can be measured with a
 V Congreso Chileno de Ingeniería Geotécnica

conventional oedometer is about 0.01cm2/sec. At the relatively low stress levels relevant to
engineering situations, the cv value of allophane clays is normally much higher than this.

Undrained strength

Figure 20 shows cone penetrometer test (CPT) results from two sites, one in Indonesia and one
in New Zealand.

Cone resistance (MPa) Cone resistance (MPa)

0 5 10 15 20 0 5 10 15 20

10 10
Depth (m)

Depth (m)

20 20

30 30
Kamojang Geothermal Omata oil storage tank,
Power Station, Indonesia New Zealand

Figure 20. Cone penetrometer tests from allophone clay sites in Indonesia and New Zealand.

These are fairly similar. They show that while the in situ strength is reasonably uniform, it does
have small fluctuations over the full profile, and there are some zones with considerably higher
values. These are believed to be zones of coarser material within the fine clay. The cone
resistance varies between about 1 and 3 MPa. Using a correlation factor (Nk) of 15 this
corresponds to an undrained shear strength range of about 65 kPa to 200 kPa. Values of
undrained strength obtained from other methods at the Kamojang site ranged from about 50 kPa
to 170 kPa, confirming the trend indicated by the CPT tests.

Effective strength parameters

The effective strength parameters c′ and φ′ are surprisingly high for a soil of such fine grained
composition. This is perhaps not surprising; observation of field behaviour suggests that this
must be the case. As mentioned earlier, in Indonesia and other tropical countries, terraced rice
 V Congreso Chileno de Ingeniería Geotécnica

fields exist on remarkably steep slopes in areas of allophane clay. These slopes remain stable
despite permanent saturation with irrigation water, which flows from terrace to terrace.

RESIDUAL STRENGTH FROM

500 RING SHEAR TESTS:
Indonesian samples o

40
New Zealand samples / =
φ
400 a,
kP
20
/ =
Shear stress (kPa)

c
ts
300 t es
ial
ax
tri
m
f ro
200 gth
en
k str
P ea
100

0 100 200 300 400 500 600

Normal effective stress (kPa)

Figure 21. Effective strength parameters for allophone clays.

Figure 21 summarises results from laboratory tests on samples of allophone clay from Indonesia
and New Zealand. Triaxial tests were carried out to obtain the peak values, and ring shear tests
to obtain the residual values. Both values are remarkably high and there is surprisingly little
difference between them. Rouse et al (1986) have obtained similar high values from allophane
soils in Dominica.

50

40
φ r (degrees)

30

20
/

10 Indonesia
New Zealand
0 20 40 60 80
Plasticity Index

Figure 22. Residual strength friction angle from allophone clays versus Plasticity Index.

Fig. 22 shows values of the residual angle φ′r plotted against Plasticity Index. It is seen that there
is no relationship between the two; φ′r does not steadily decrease with Plasticity Index as is the
case with sedimentary clays. With PI values above about 80, sedimentary soils would be
expected to have φ′r values of around 10o, whereas the allophane clay has values between 30o and
40o.
 V Congreso Chileno de Ingeniería Geotécnica

Compaction characteristics

The compaction behaviour of typical allophane clay was illustrated earlier in Fig. 10. The natural
water content was 166%, and the natural curve was obtained by drying back the soil in steps
from this initial water content. Fresh soil was used for each point. The test was then repeated
three times, firstly after oven drying, secondly after air drying, and finally after limited air drying
(to w = 65%). The material was then wetted up in stages, using fresh soil for each point. The
results show the dramatic changes caused by drying. When dried from its natural water content
the compaction curve is almost flat, with only a very poorly defined optimum water content. On
re-wetting, the behaviour becomes more conventional, with clearly defined optimum water
contents and peak dry densities. It is evident from this that almost any result can be obtained if
the test involves drying and re-wetting. This result is from an Indoneisan allophone clay. Local
(ie New Zealand) allophone clays may not show such a dramatic effect because of their lower
allophone content.

Fig. 23 (after Kuno et al 1978) shows the effect of repeated compaction on allophane soils. Some
allophone clays are of high sensitivity, and others are not: this is reflected in the curves in Fig.
23. The strength of the soil has been measured after compaction using a series of different (but
known) compactive efforts. The compactive effort is indicated by the number of hammer blows.
A cone has been pushed into the soil to obtain a measure of strength; this is the "cone index"
value shown in the figure. The graphs show that in general there is a marked decrease in strength
as the number of blows increases. Presumably the structure of the soil is being progressively
destroyed, releasing water and softening the soil, an effect sometimes referred to as “over-
compaction”.

Figure 23. Influence of compactive effort on strength of compacted alllophane clays.

The above behaviour illustrates that difficulties can arise in compacting allophane soils if their
properties are not understood and taken account of in planning and executing earthworks
operations. Specifications can be almost meaningless if excessive drying is allowed before
testing is carried out. In countries like Papua-New Guinea and Indonesia the wet climate in
which allophane clays occur means that significant drying during excavation and compaction is
not very practical. Difficulties during earthworks operations are described by Parton & Olsen
(1980), and Moore & Styles (1988).
 V Congreso Chileno de Ingeniería Geotécnica

These problems can be overcome to some extent in several ways. The first is to recognise that
soils can be satisfactorily compacted without recourse to the rigid control methods associated
with water content and dry density values. The second is to be clear what objective is aimed for
in compacting the soil. For example, the objective with a road embankment is very different
from that with a water retaining embankment. With a road embankment it is preferable to keep
the compactive effort to a minimum and “press” the soil together with quite light compaction. –
enough to get rid of any large voids, but insufficient to destroy the natural “structure” of the soil
and cause it to soften. In this way it is possible to retain much of the original strength of the
material. With water retaining embankments a rather more rigorous approach is needed, but even
for these it is desirable to carefully control the compactive effort. Compaction control, involving
control of compactive effort, together with shear strength and air voids testing is generally a
better approach than conventional water content and dry density methods.

The Cipanunjang dam in West Java (Wesley, 1974) is an example of successful compaction of
allophone clay; compaction here was done using steel rimmed rollers. Some difficulties were
encountered due to wet weather and softening of the soil, but the job was completed
satisfactorily. The writer has been involved in the compaction of allophane clay at a geothermal
power station site (Kamojang) in West Java, Indonesia. Difficulties were encountered because
the very wet climate at the site made it difficult to dry the soil sufficiently to achieve the target
undrained shear strength of 150 kPa. The fill was required to form a level platform for a
electrical tansformer and switch yard. The strength requirement was lowered to 90 kPa and the
job successfully completed. The fill appeared to “harden” with time, presumably due to the
development of negative pore pressure in the soil.

Erosion resistance.

It is an interesting observation that both in their undisturbed and re-compacted state, allophane
clays are remarkably resistant to erosion. It is only when they are cultivated and allowed to
partially dry at the surface that they become susceptible to erosion. Observation of road cuttings
in Southeast Asia as well as in New Zealand (Taranaki and the central volcanic plateau) shows
that negligible erosion occurs from the cut faces. In Indonesia, the drying of the face appears to
result in the formation of a hard “crust” which is resistant to erosion. It is also evident in terraced
rice-fields that negligible erosion takes place as the irrigation water flows from one terrace to the
next terrace.

In relation to erodibility, the writer has investigated the question of the dispersivity of allophone
clays by carrying out pin-hole dispersion tests on allophane clays from Indonesia and New
Zealand. The results are described by Wesley and Chan (1991). None of these tests showed any
evidence of erosion or dispersivity.

Significant engineering projects in allophone clays.

A number of dams and related water retaining structures have been successfully undertaken
making use of allophone clays. An early example is the water supply dam Cipanunjang (formerly
spelt Tjipanundjang) in West Java, Indonesia, built in 1928 during the Dutch colonial period.
This is a homogeneous 30m high embankment with cut-off drains in the downstream slope. It is
described in detail elsewhere (Wesley, 1974), and is still a vital part of the municipal water
supply of the city of Bandung, the capital city of West Java. The Mangamahoe Dam in New
Plymouth, New Zealand, and the embankment supporting the supply canal at the Kuratau power
scheme (on the western shore of Lake Taupo, New Zealand) are further examples of
embankments of allophone clay forming water retaining structures. The Kamojang geothermal
power station in West Java, Indonesia, is supported by a raft foundation on about 35m of
 V Congreso Chileno de Ingeniería Geotécnica

allophone clay (Figure 20). There have been no problems with its performance. Wesley and
Matuschka (1988) describe these examples in greater detail.

REFERENCES AND BIBLIOGRAPHY

Frost, R.J. 1967, Importance of correct pre-testing preparation of some tropical soils. Proc. First
Southeast Asian Regional Conf. on Soil Engineering, Bangkok: 44-53.
Geological Society Engineering Working Party Report: Tropical Residual Soils(1990) Quarterly
Journal of Engineering Geology, Vol.23, No.1 pp 1-101.
Kenney and Lau (1984) Temporal changes of groundwater pressure in a natural clay slope.
Canadian Geotechnical Journal. Vol. 21, 1984.
Kuno, G., Shinoki, R., Kondo, T. & Tsuchiya, C. (1978). “On the construction methods of a
motorway embankment by a sensitive volcanic clay,” Proc. Conf. on Clay Fills, London, pp.
149-156.
Moore, P.J., & Styles, J.R. 1988. Some characteristics of volcanic ash soil . Proc. Second Int.
Conf. on Geomechanics in Tropical Soils. Singapore: 161-166.
Morin, W.J. & Todor, P.C. 1975. Laterite and lateritic soils and other problem soils of the
tropics. USAID Lyon Associates, Baltimore, U.S.A.
Parton, I. M. & Olsen, A.J. (1980). Properties of Bay of Plenty Volcanic Soils. Proc. 3rd
Australia New Zealand Conference on Geomechanics, Welllington. Vol.1: 165-169.
Rao, Sudhakar M. 1996. Role of apparent cohesion in the stability of Dominican allophane soil
slopes. Engineering Geology 43(4): 265 - 279.
Rao, Sudhakar M. 1995. Mechanistic approach to the behaviour of allophanic soils. Engineering
Geology 40(3-4): 215 - 221.
Rouse, W.C. 1990. The mechanics of small tropical flowslides in Dominica, West Indies.
Engineering Geology 29:227-239.
Rouse, W.C. Reading, A.J. & Walsh, R.P.D. 1986. Volcanic soil properties in Dominica, West
Indies. Engineering Geology 23:1-18.
Shimizu, M. 1998. Engineering properties of andosols or Kuroboku. Volcanic Ash Soils in Japan
- Properties and Practical Use, Japan Geotechnical Society: 6 - 15.
Susuki, A. & Kitazono, Y. 1998. Engineering classification of volcanic cohesive soils. Volcanic
Ash Soils in Japan - Properties and Practical Use, Japan Geotechnical Society: 39 - 43.
Wada, K. (1989). “Allophane and imogolite”. Chapter 21 of Minerals in Soil Environments (2nd
Edition) SSSA Book Series No 1, pp. 1051-1087.
Wesley, L.D. 1973, Some basic engineering properties of halloysite and allophane clays in Java,
Indonesia. Geotechnique 23, No 4: 471-494.
Wesley, L.D. 1974. Tjipanundjang Dam in West Java, Indonesia. Journal of the Geotechnical
Division ASCE 100/GT5: 503-522.
Wesley, L.D. 1977, Shear strength properties of halloysite and allophane clays in Java,
Indonesia. Geotechnique 27, No2: 125-136
Wesley, L.D. & Matuschka, T. 1988. Geotechnical engineering in volcanic ash soils. Proc.
Second Int. Conf. on Geomechanics in Tropical Soils, Singapore Dec. 1988. Vol.1: 333-340. .
Wesley, L.D. 1998, Some lessons from geotechnical engineering in volcanic soils. Proc. Int.
Symposium on Problematic Soils. Oct. 1998: Vol.1: 851-863. Sendai, Japan: Balkema
Wesley, L.D. and Chan S.Y. (1990) “The dispersivity of volcanic ash soils,” Proc. IPENZ
Conference 1991. Vol. 1, pp.67-76.
Wesley, L.D. (2002). “Geotechnical characterisation and properties of allophone clays”. Proc.
International Workshop on Characterisation and Engineering Properties of Natural Soils.
Singapore, Dec. 2002.