Sie sind auf Seite 1von 16

WALLACE,K. B. (1973).

Structural

Geotechnique23, No. 2, 203-218.

behaviour of residual soils of the continually


wet Highlands of Papua New Guinea
K. B. WALLACE*

The Paper illustrates how high porosity, a strongly


cemented structure and highly hydrated kandite clay
minerals control the geotechnical properties of residual sandy to silty clay soils of the Papua New
Guinea highlands. These soils are characterized by
high field moisture contents, relatively high shear
strengths and low sensitivity, and are highly compressible when the applied pressure in an oedometer
test exceeds a certain critical pressure. Measured
values of critical pressure and compression index are
compared with those reported for a wide variety of
soils, and a linear relationship between volumetric
strain and porosity is established for porous soils.
The natural inhomogeneity
of residual soils is
emphasized.
Examination of relationships between
geotechnical properties and porosity and moisture
content shows that porosity or dry density is a more
reasonable basis for description and grouping of
these soils for engineering purposes.
The behaviour of the soils is interpreted in terms of
an idealized model of the soil structure.
The soil is
considered to consist of a coarse open skeleton of
cemented rock mineral and aggregated clay particles
surrounded
by a viscous gel. At low applied
stresses, when the cemented bonds in the soil
skeleton are intact, it is the soil skeleton that determines structural behaviour. At high applied stresses
the porous soil skeleton still has a strong influence on
structural behaviour.

Cette communication
montre comment la porosite
BlevBe, la structure
fortement
cimentCe et les
min6raux argileux trbs hydra&
du type kandite,
rdgissent les propri&s
gkotechniques de sols r&iduels de type sableux & silt argileux dans les hautes
terres de Panouasie-Nouvelle
GuinBe. Ces sols
sent caract&&s
par des teneurs en eau en place
dlevkes, des rCsistances au cisaillement relativement
6levees et une faible sensibilit6, et sent t&s compressibles lorsque la pression appliqu6e depasse une
certaine valeur critique lors de lessai oedom&rique.
On compare les valeurs mesurees pour la pression
critique et lindice de compression B celles qui ont
dt6 publiees pour une grande vari& de sols, et on
etablit une relation liSaire entre la dkformation
volumetrique et la porosit6, pour les sols poreux.
On insiste sur lhCt&ogCn&t6 des sols r6siduels.
Lexamen des relations entre les propriCtCs geetechniques et la porosit6 et la teneur en eau montre

que la porositk ou la densite B set forme une base


raisonnable de description et de groupement de ces
sols pour le genie civil.
On interprkte le comportement de ces sols en fonction dun modkle idkalisd de la structure du sol. On
considkre que le sol se compose dun squellette
grossier ouvert de minkraux cimentks et daggrtgats
de particules dargile entourtes dun gel visqueux.
Aux faibles contraintes, les liaisons cimentkes dans
le squelette du sol restent intactes et le squelette du
sol determine le comportement
de la structure.
Aux contraintes tlevtes, le squelette poreux du sol
a encore une grande influence sur le comportement
de la structure.

residual soils described in this Paper occur at well-drained


sites in the Western and
Southern Highlands Districts of Papua New Guinea.
The Highlands Districts are situated
close to the equator (63, 14OE) and consist of undulating
valleys (4000-6000 ft above sea
level) separated by extremely rugged mountainous
terrain.
Although these valleys contain
about 40% of the countrys population,
they are mainly undeveloped with the population
depending on a simple subsistence agriculture.
The climate is continually
wet with an average annual rainfall of about 100 inches, evenly
distributed throughout
the year.
The mean monthly rainfall nearly always exceeds the estimated mean evapo-transpiration
and rain may be expected on an average of between 70 and

The

* Research Fellow, Department of Engineering, James Cook University of North Queensland, Townsville, Australia.

204

K. B. WALLACE

90% of all days.


At 5000 ft elevation the temperature
usually varies from 55F at night to
75F during the day and there is very little monthly fluctuation in temperature.
This warm,
continually
wet climate is the major factor determining
the unusual properties of soils in this
region.
The zonal nature of these soils has been recognized by earlier pedological studies (Haantjens
and Rutherford,
1964), which showed similarities in the morphology and clay mineralogy of
soils on such dissimilar parent materials as limestone, siltstone, basalt and andesitic ash. The
Authors studies (Wallace, 1971) have confirmed this zonality and have highlighted one difference of engineering consequence.
Soil depth on volcanics is generally greater than 20-30 ft
whereas in many areas, only 3-6 ft of soil rests abruptly on limestone.
Pleistocene to recent volcanic eruptions have covered large areas of the Highlands with
layers of andesitic ash. The combination
of volcanic ash and continually
wet climate have
produced soils which are, in many respects, similar to the New Zealand andesitic ash soils
(Birrell, 1952) and to Kanto Loam of Japan (Miki, 1960).
Basic data on these residual soils of the Papua New Guinea Highlands is presented in the
next section of this Paper, while the succeeding sections describe their geotechnical properties
and go on to discuss the relationship
between these properties and the soil structure.
THE SOILS

The results presented in this Paper are taken from a more general study of field occurrence
and geotechnical properties of these soils (Wallace, 1971). Because of the relatively remote
and undeveloped
nature of the Highlands the field work consisted of sampling from a few
representative
sites which had been chosen after extensive terrain description and examination
of road cuttings in the region.
Although the data are inadequate for detailed statistical analysis and correlation, it is thought that the careful selection of sampling sites has produced data
that are characteristic of the nature of soils on the parent materials within the region.
Soils from four well-drained locations are described in Table 1, which also gives details of
field moisture content, liquid limit, plasticity index, percentage sand (as measured by wet
sieving undried soil on a 75 pm sieve), clay mineralogy and soil chemistry.
Three of the soils were formed from intermediate to basic volcanics and one from limestone.
As all the sites were adjacent to Mt Giluwe, a Pleistocene volcano which is known to have
covered the area with andesitic ash showers (Blake and Liiffler, 1971), it is extremely difficult
to determine the relative influence of volcanic ash at each site. Examination
of quarry
exposures, sand mineralogy and local geomorphology
indicates that the allophanic sandy clay
(Soils No. 2 and No. 3) has been strongly influenced by andesitic ash whereas the halloysitic
silty clay (Soil No. 1) is formed from a vesicular basic lava flow with possibly some minor ash
contamination
from above.
The limestone soil (Soil No. 4) has been only slightly contaminated by ash (Wallace, 1971).
Clay mineral analysis was done by X-ray diffraction and differential thermal analysis (Amdel,
1971) after free iron removal by a dithionite-citrate
system buffered with sodium bicarbonate
(Mehra and Jackson, 1960). Table 1 shows that the predominant
clay minerals were the
highly hydrated kandite minerals, halloysite and allophane with gibbsite occurring in large
proportions,
close to the surface, especially on limestone.
The presence of allophane can
remain undetected
by available tests on soils which contain both hydrated halloysite and
gibbsite, as do Soils No. 1 and No. 4.
The low silica/alumina
ratios in Table 1 illustrate the severe leaching of silica during rock
weathering.
The silty clay, Soil 1, has a silica/alumina
ratio of 1.15 compared with a value of

No. 4
3 depth at
Site No. 4 on
Limestone

No. 3
5-6 depth at
Site No. 3 on
Volcanics

No. 2
5-6 depth at
Site No. 2 on
Volcanics

7-6 depth at
Site No. 1 on
Volcanics

No. 1

location

Firm, moist,
olive-brown
:;;tred,
sandy

Firm, moist,
olive-brown,
fissured, sandy
,clay

Firm, moist,
yellow-brown
fissured silty clay

Firm, moist,
yellow-brown
slightly friable,
silty clay

I-

l-

Field description

Description of soils

Soi~~~~dnce

Table 1.

90

130

85

130

Typical field
moisture
content
(105G), %

___110

160

145

135

_____

Liquid limtit

-___

35

85

70

55

Plasticity
index

15
(mainly aggregated clay
minerals)

45
(rock forming
minerals)

35
(rock forming
minerals)

15
forming
minerals)

I[rock

Approximate
per;;;ztge

Gibbsite predominant with


moderate amount: i
of hydrated
halloysite

Allophane and
vermiculite

Allophane and
vermiculite

Hydrated halloysite predominant


with moderate
quantities of
gibbsite

Principal clay
minerals

1
pH = 5.6

pH = 5.4

2:

3
pH = 5.5

pH = 6.2

14

% Organic
matter
PH

0.6

1.3

1.65

1.15

SiOz
Al,03

206

K. B. WALLACE

2.8 for the parent materials. Ruxton (1968) has concluded that silica/alumina ratio is a good
index of the degree of rock weathering in free-draining weathering environments in humid
regions. This conclusion is supported by the results of chemical analysis of the present soils
(Wallace, 1971). In their chemical classification of lateritic soils, Martin and Doyne (1927)
described soils with a silica/alumina ratio between 1.33 and 2.00 as lateritic soils. Soils with a
silica/alumina ratio less than 1.33 were called laterite soils, while soils with a silica/alumina
ratio in excess of 2.00 were considered to be non-Iateritic.
GEOTECHNICAL

PROPERTIES

Field moisture content


In contrast to their firm moist appearance, these soils have extremely high field moisture
contents. This is illustrated by Soil 3 (Table 1) which contains 45% of sand particles but has a
moisture content of 130%. This is very much higher than the values of 20-70% moisture contents that are typical of a wide range of sedimentary clays of low to quite high plasticity.
Generally the moisture content of the present soils varies between 80-190%, and there is no
consistent pattern of variation of moisture content with depth or between sites. Field dry
density of the sub-soil is correspondingly low, varying between 35 and 55 lb/cu. ft.
The soils are not saturated. The measured degree of saturation of undisturbed samples was
generally between 85 and 95%.
The field moisture content of most samples is between 80 and 100% of the liquid limit.
Many of the samples tested had a moisture content which was about 10% lower than the liquid
limit. When the consistency limits are plotted on a Casagrande chart all soils are well below
the A-line and would therefore be classified as highly plastic silts.
Shear strength
The immediate, undrained shear strength has been determined using a Torvane. Generally
the Torvane shear strength of undisturbed samples tested at field moisture content ranged from
800 to 1200 Ib/sq. ft. The strength of the allophanic sandy clay (Soils 2 and 3) was slightly
greater than that of the halloysitic silty clay (Soil 1). Several samples of Soil 2 had Torvane
shear strengths of 2500 lb/sq. ft but other samples from the same location gave values of 1100
Ib/sq. ft. The undrained shear strength is much higher than the values of 300 lb/sq. ft which
are common for equivalent, normally consolidated, sedimentary soils (Skempton, 1957 and
Osterman, 1959). The reIatively high shear strength is also indicated by the existence of
stable, nearly vertical road cuttings up to 30 feet high.
There was general agreement between the Torvane shear strength measurements and the
cohesive component of shear strength as measured in drained shear box tests on soaked samples
of similar soil. From this it was concluded that moisture suction does not contribute to the
comparatively high undrained shear strength. In a later section of this Paper the high
undrained shear strength is attributed to cemented bonds between the soil particles.
Stress-deformation
curves for drained shear box tests which were carried out at two
different applied pressures on soaked, undisturbed samples of Soil 2, are shown in Fig. 1. It
can be seen that the behaviour of the soil was similar to that of an overconsolidated clay, in
that at low applied pressure the soil dilated during shearing and there was a pronounced peak
shear strength. At higher applied pressure the sample was compressed during shearing and
there was no sharp peak shear strength, the maximum shear strength occurring at a large
deformation.

STRUCTURAL

BEHAVIOUR

OF RESIDUAL

SOILS

OF THE

CONTINUALLY

WET

wN=

15 Ib/rq.

Fig. 1.

movemenr

OF PAPUA

NEW

GUINEA

207

in.

--o---l
Ib/rq. in.

UN=5

Horizontal

HIGHLANDS

in inches

Direct shear box test results for loads of 5 and 15 lb/sq. in. on Soil No. 2

Table 2 illustrates how the undrained shear strength is affected by soaking and remoulding.
The strength of undisturbed samples was not appreciably altered by soaking, but soaking
greatly reduced the strength of remoulded samples.
Both Table 2 and Fig. 1 indicate that, for soils of such high porosity, these soils are relatively
insensitive, having a sensitivity of about 2. This is very much less than the sensitivity of the
cemented Canadian sedimentary clays (Sangrey, 1972b) but not much lower than that of many
residual soils. Lohnes et al. (1971) report low sensitivity for a range of Puerto Rican lateritic
soils.
Residual values of the angle of internal friction obtained from drained shear box tests on
soaked, undisturbed samples of sandy clay were:
Soil 2

+d = 38

Soil 3 fjd = 29
Comparison of the descriptions of the two soils given in Table 1 shows that the higher strength
of Soil 2 may be attributed to the lesser degree of weathering as indicated by the lower field
moisture content and higher silica/alumina ratio of Soil 2.
Table 2.

Torvane shear strength before and after remoulding and soaking

Soil

Test condition
Undisturbed
1 Unsoaked,

No. 1

No. 3

640
980

lb/sq. ft

Soaked, lb/sq. ft
600
1040

Remoulded

/ Unsoaked,
300
420

at field moisture content

lb/sq. ft

Soaked, Ib/sq. ft
100
300

208

K. B. WALLACE

Compressibility

Typical compressibility characteristics for these soils are shown in Figure 2 which presents
the e-logp curves for oedometer tests on undisturbed and remoulded specimens of Soil 2.
At low applied pressure the compressibility is low but when the applied pressure exceeds a
certain critical pressure (pc) the compressibility of the soil is very high. The remoulded
compressibility is also very high but is lower than that of the undisturbed soil at pressures
exceeding the critical pressure. On removal of the load the soil does not swell very much.
The resultant rebound curve, for the least disturbed samples, was roughly parallel to the compression curve for low applied pressures.
The relative compressibility and rebound of soils may be described by the following indices :

and

C, the compression index of the soil in its natural state (that is the slope of the e-logp
curve at pressures exceeding the critical pressure, pJ
C, the compression index of the soil after remoulding
C,, the swelling index (the slope of the rebound curve).

Values of these indices are given on Fig. 2 and the values obtained from oedometer tests on
each of the four soils are summarized in Table 3. Table 3 also gives some indication of the
relative disturbance of undisturbed samples. This is a subjective assessment based on the
shape of the e-logp curves. Soils that show a sharp change in slope of the curve over a
narrow range of pressure around the critical pressure are described as least disturbed. This
assessment was found to agree well with the relative difficulty of preparing oedometer specimens from 4-inch tube samples.
It should be noted that in all cases the critical pressure was much greater than the present
overburden pressure. In this respect, the compressibility characteristics are similar to those of
overconsolidated soils but there is no geomorphological evidence to suggest that any of the
soils have ever experienced overburden pressures of similar magnitude to the critical pressure.
The continually wet climate and the presence of highly hydrated allophane and halloysite suggest that desiccation has not caused overconsolidation.
This critical pressure may therefore
be attributed to cemented bonds between the soil particles as is discussed in a later section of
this Paper.
The magnitude of the critical pressure measured on samples of the various soils ranged from
about 1 to 34 ton/sq. ft and this may be compared with values reported for other residual soils.
Gradwell and Birrell (1954) report values of 1.1 to 2.7 ton/sq. ft for a wide range of volcanic
clays. Vargas (1953) shows that the critical pressures for residual clays on gneiss, basalt and
sandstone in Southern Brazil are widely scattered between O-6 and 4.5 ton/sq. ft. Values for
residual soils in the South-eastern United States (Sowers, 1963) vary between 1 and 5.5 ton/sq.
ft. Finally, the Japanese soil, Kanto Loam (Koizumi and Ito, 1963), which is mineralogically
Table 3.

Compressibility characteristics

Soil

,
No. 1
No. 2
No. 3
No. 4

Critical
pressure,
ton/sq. ft
1.6, 2.3
3.5, 3.6
1.7-1.8
0.9-2.3

Compression

1.31, 1.93
079, 093
1.01-2.3
0.74, 1.19

/
i
1

Relative disturbance

index

0.10, 0.07
0.06, 0.07
0.06
0.04, 0.05

0.39-1.29
0.59

most
least
least
most

disturbed
disturbed
disturbed
disturbed

STRUCTURAL

BEHAVIOUR

OF RESIDUAL

I.0

I/

O!

02

SOILS OF THE

CONTINUALLY

05
Applied

Fig. 2.

I.0

WET

HIGHLANDS

OF PAPUA

NEW

209

GUINEA

50

2.0

pressure in ton.sq. fr

Typical e-logp curves for oedometer test on Soil No. 2

and geotechnically
similar to the present soils, has critical pressures within the range 2-5.5
ton/sq. ft. From these diverse results it can be seen that although there is general agreement
on the order of magnitude of critical pressures for a wide variety of residual soils, there is a
Large point to point variations in residual
large variation in values for any particular soil type.
soil properties are discussed in a later section of this Paper.
Compressibility correlations
On Fig. 3, the measured values of compression index are plotted against initial void ratio for
It can be seen that there is a good linear relationboth undisturbed
and remoulded samples.
When the applied pressure exceeds the
ship between compression index and initial void ratio.
critical pressure the compressibility
of the undisturbed
samples with initial void ratios greater
than 2 can be described by the relationship
C = ae,-b

(1)

where a=0*6 and b=0.7.


Similar correlations
have been obtained by others as a result of more extensive tests on a
Some of these other correlations between compression index and initial void
variety of soils.
indicate that for
ratio are compared with the present results on Fig. 4. These correlations
soils with an initial void ratio in excess of I.0 (or a saturation moisture content greater than
40%) there is a reasonably
linear relationship
between compression
index and initial void
ratio.
For comparison
of the compressibility
of various soils at loads in excess of the critical
pressure, the settlement, S, of a layer of thickness, H, due to an increase in pressure from p.
to p. + Ap may be described by the approximate
relationship
&$-olog,op~

(2)

210

K. B. WALLACE

lnltial void ratio, co

Fig. 3.

Relationship between initial void ratio and compression index of undisturbed and remoulded soils

(I)

The soils discussed in this Paper

(2) Sowers (1963):

residual soils of south-eastern

(3) Penta et al. (1961):

Fig. 4.

2
Initial Void

3
Ratlo,

(4) Arango (Lambe and Whitman.


USA and Columbia

e,

(5) Kay and Krizek

USA

Italian volcanic ash soils


1969): soils of western

(1971): sedimentary

roils:of

USA

Compression index correlations with initial void ratio

(This approximation appears to be reasonable when considered with respect to the relatively
low compressibility below the critical pressure and the uncertainty in estimating the critical
pressure.)
Combining equations (1) and (2) gives
S

ae,-b

-=l+elog,opc
H
0

PO+AP

Substituting initial porosity, no, for e,/l +e,


;

= [(a+b)n,-b]

logIop+

* *

(3)

STRUCTURAL

BEHAVIOUR OF RESIDUAL SOILS OF THE CONTINUALLY

= A(no-B)

WET HIGHLANDS

PO+4
log,,--PC

OF PAPUA NEW GUINEA

211
(4)

where A and B are constants such that A = (a + b) and B = b/(u + b).


For the present soils, A= l-3 and B=0.55.
It should be emphasized that the preceding approximations are not suitable for close prediction of soil properties. The scatter of results obtained by the various investigators named
on Fig. 4 suggests that individual values may vary by f 30% or more from these correlations.
However it is thought that equations (l), (3) and (4) are a reasonable basis for analysing the
structural characteristics of the soil. For example, a comparison of equations (3) and (4)
shows that the compression of the soil is more linearly related to porosity than to void ratio.
This linearity is discussed further in a later section of this Paper which considers the consequences of high moisture contents.
A simple interpretation of equation (4) suggests that B represents an apparent minimum
porosity for the particular group of soils to which the equation is applied and the parameter A
is the flexibility of the soil skeleton. This flexibility (which is the inverse of the stiffness of the
soil skeleton) is defined in terms of porosity by equation (4).
Creep

Some measurements of the rate of creep of oedometer specimens (the oedometer rings being
lubricated with silicone grease and Ap/p = 1) have yielded the following conclusions :
(a) at loads greater than the critical pressure the creep rate of undisturbed

samples
(C, -0*006) was roughly ten times the creep rate at loads less than the critical pressure
(C, N 0*0002-0~0005),
(b) at loads greater than the critical pressure the creep rate was reasonably independent of
applied pressure increment whereas, below the critical pressure, the creep rate increased
roughly linearly with applied pressure increment,
(c) disturbance during sampling greatly increased measured creep rates at applied pressures less than the critical pressure.
The general effect of drying on these creep characteristics is discussed later in the Paper, with
respect to interpreting the general structural behaviour of the soil.
EFFECT OF DRYING

Without exception these soils become non-plastic when air-dried or oven-dried and do not
regain their plasticity when wetted again. Frost (1967) has emphasized the need for appropriate preparation prior to testing these soils. He has also observed that as a result of airdrying, the soaked, remoulded CBR of one soil from the region was increased from 2 to 60%.
To check the effect of drying on permeability, samples were dried and then remoulded and
compacted to their original dry density. The permeability measured after saturation of the
compacted sample was then compared with that of the same soil remoulded from the natural
state without drying. The results of these tests are shown on Table 4, where it can be seen that
as a result of dehydration and aggregation of the clay particles on drying, the permeability is
increased more than one hundredfold.
This aggregation of soil particles on drying is illustrated on Table 5 which shows that results
of particle size analysis of natural and dried samples of Soils 1 and 3. In these tests the dispersing agent was sodium hexametaphosphate.

212

K. B. WALLACE

Table 4.

Effect of drying on permeability

Soil

Soil remoulded

from natural state and


saturated

Dry density,
lb/cu. ft
No. 1
No. 3

Permeability,

Dry density,
Ib/cu. ft

5,7x 10-T
2.4 x lo-

:;:

Table 5.

cm/s

Soil dried, remoulded

and saturated

Permeability,

;:;;

::

cm/s

;;I:

Effect of drying on particle size

Soil

Tested after drying (105C)

Tested from natural state


Sand, %

No. 1

No. 3

3:

Silt, %

::

Clay, %

47

Saud, %

:z

Silt, %

Clay, %

45

20

18

It is often suggested that this dramatic improvement


of remoulded soil properties by drying
could be used to considerable
advantage
in construction.
However, the continually
wet
climate which is responsible
for formation
of the unusual clay minerals usually persists
throughout
construction
and limits the practicality of soil improvement
by drying to the most
compact sites such as airports or dam sites.
DISCUSSION

Variability of residual soil properties


As is typical of many residual soils, the properties of soils sampled from a given depth in a
given test pit varied considerably.
The results of six measurements
of the moisture content
of parts of a two-inch lump of Soil 3 ranged from 108% (n=0*76) up to 146% (n=O*80).
Several samples of Soil 2 had undrained shear strengths of about twice the typical average value
for this soil. These point to point variations in soil properties were often as large as variations with depth at a given site or variations between sites.
A similar lack of homogeneity has been observed for decomposed granite (Lumb, 1962), and
for the New Zealand volcanic ash clays (Gradwell and Birrell, 1954). This inhomogeneity
is
due to some or all of the following factors which are basic to the occurrence of residual soils,
particularly
when the soil is formed under well-drained conditions:
(a) variations in grain size and mineralogy of the parent material
(b) irregular chemical weathering
and leaching of weathering
products due to nonuniform seepage of ground water through the variable and fissured, weathered parent
material
(c) the dependence of the undisturbed
properties on the nature of cemented bonds, which
are only a very small part of the total volume of the soil particles and therefore may be
more variable than the mineralogy or the soil texture
(d) the occurrence of fissures which are often close to vertical: these fissures could be the
remnants of cracks or joints in the parent material or they could have been caused by
slumping or lateral unloading
of the soil as the landscape is dissected by streams; such
fissures will often determine the stability of road cuttings

STRUCTURAL

(e)

BEHAVIOUR

OF RESIDUAL

SOILS OF THE CONTINUALLY

WET HIGHLANDS

OF PAPUA

NEW

disturbance
of the cemented soil structure during sampling and during
laboratory specimens which will add to the variability of test results

GUINEA

trimming

213

of

This variability
of residual soil properties suggests that site investigations
for extensive
engineering
works should concentrate
on more intensive evaluation
at fewer sampling
sites. The location of these sites would be chosen according to changes in the basic soilforming factors: parent material, drainage, slope and, where applicable, climate.
Because of
the irrelevance of conventional
classification
systems with respect to residual soils (Wallace,
1970) and the important
influence of macroscopic structure, it is particularly
important that
site investigation
reports should recognize residual soils as such and contain objective descriptions of their undisturbed
strength (firm, soft, stiff etc.) and their macroscopic
structure
(friable, fissured, fragmented etc.).
High moisture contents
The most remarkable characteristic
of these soils is their unusually high field moisture contents, which of course represent proportionally
high void ratios and higher than normal
porosities.
To many soils engineers who intuitively associate high field moisture content with
low shear strength, high sensitivity and high compressibility,
these soils appear to be unusually
strong.
Much of the stiffness and strength of these soils may be attributed to the cemented
structure which is described in the next section; however, at loads high enough to break down
this cemented structure, the strength and stiffness can still be surprisingly high.
Is it reasonable to expect proporThis raises a simple question of practical importance.
tionally high compressibility
and low drained shear strength at high field moisture contents?
High field moisture contents (that is moisture contents in excess of 40%) may be due to the
presence of a highly active clay fraction or, as is the case for the present soils which contain
clay minerals of known low activity, high moisture content may be due to an unusual fabric of
high porosity.
The following remarks are relevant to the latter type of soil.
Analysis of compressibility
correlations in a preceding section has shown that for the present
soils, and for a wide variety of other porous soils, the relative compressibility
is linearly dependent on porosity rather than on void ratio and moisture content.
The drained shear strength
of these porous soils is a complicated function of the cross sectional area of the soil skeleton and
The cross sectional area will be linearly related to
the degree of interlocking
of soil particles.
porosity while the previous relationship between compressibility
and porosity suggests that the
interlocking
component
of shear strength could also be more linearly related to porosity.
Alternatively,
it could be argued that the basic mechanism of large volume changes during
compression
is one of local shear, and therefore compressibility
and drained shear strength
could be expected to have a similar primary dependence on soil structure.
The highly non-linear
relationship
between saturation
moisture content and porosity for
porous soils is reproduced in Fig. 5. This figure indicates that, compared with variations of
porosity with moisture content in the range of moisture contents between 20 and 70x, an
increase of moisture content from 100 to 150% represents a relatively small increase in porosity. Therefore, as compressibility
and probably also shear strength are linearly related to
porosity, it can be concluded that it is not reasonable to expect that soils with high moisture
contents will have proportionally
low shear strength and compressibility.
This conclusion
indicates that porosity or dry density is a more reasonable basis for description and grouping of
these extremely porous soils for engineering purposes.

K. B. WALLACE

214

Fig. 6 (below). Idealized residual soil structure

I.0

0.8

0.6
x
c
8
8
a.
0.4

LCernented

Bonds

0.2

20

50

70
150
100
Moisture Content I%1

Fig. 5 (left). Variation of porosity with moisture


content of a saturated soil (G,=27)

One consequence
of the high moisture content in construction
on the present soils is that
they require comparatively
small amounts of additional water to bring them to the liquid limit.
A typical highly plastic sedimentary clay (LL = 60, PL = 30) at a natural moisture content of
40% would require the addition of a volume of water equal to about 25% of its bulk volume to
bring the soil to the liquid limit.
A sandy clay, similar to the present soils, at a natural
moisture content of 12OA (LL= 135), would reach the same consistency after the addition of
less than 10% (by total volume) of water.
This illustrates how the surface of trafficked earthfill will be reduced to a quagmire by relatively light rainfall.
Cemented soil structure
The fabric or macrostructure
of the volcanic soils (as viewed at X20 magnification)
consists
of a coarse open skeleton of rock-forming
minerals (hornblende,
plagioclase and quartz)
surrounded by a viscous gel of highly hydrated clay minerals and sesquioxides.
The fabric of
the limestone soil can be considered to be similar except that the skeleton is comprised of
aggregated clay mineral particles rather than the original rock-forming
minerals.
Other
residual soils will have a skeleton that consists partly of rock-forming
mineral? and partly of
aggregated weathering
products.
This idealized structure of fine-grained
residual soils is
sketched in Fig. 6. The proposed structure is different from that of many sedimentary sandy
and silty clays, which may be considered to consist of coarse particles dispersed throughout
a
clay matrix (Trollope and Chan, 1959).
The results of compression
and shear tests described earlier show that the particles of the
primary soil skeleton are cemented together at their contacts to form a continuous
threedimensional
structural framework.
The nature of this bond has not been definitely established but the absence of carbonates, the low organic content, the presence of bonds in the completely residual limestone soil and successful dispersal of the clay fraction after free iron
removal suggest that the bonds are associated with iron and aluminium hydroxides precipitated
during rock weathering.
Measured free iron content of the soil gave no indication of strength
or stiffness but such a correlation would not be expected, as the amount of iron oxides required

STRUCTURAL

BEHAVIOUR

OF RESIDUAL

SOILS

OF THE

CONTINUALLY

WET

HIGHLANDS

OF PAPUA

NEW

GUINEA

215

to form the bonds could be much smaller than the measured total free iron contents of between
one and ten per cent. Higher free iron contents would also be associated with more advanced
weathering of the soil skeleton.
Sesquioxide cementing agents have also been indirectly detected in some African clays
(Terzaghi, 1958 and Newill, 1961) and also in cemented Canadian clays (Quigley, 1968 and
Sangrey, 1972b). It is interesting to note that the rock-forming minerals present in the cemented Canadian clays are the same as those found in andesitic ash soils. Therefore it is expected
that at low stresses, when the soil skeleton strength controls drained stress-strain characteristics, there would be similarities between the behaviour of the two types of soil.
Relative roles of cemented bonds and viscous gel

When the present soils are dried the clay minerals aggregate so that the viscous gel is permanently destroyed. It was thought that if oedometer specimens could be prepared from soil
which had been dried and then resaturated without disturbance, these specimens would
possess cemented bonds without the viscous gel and a comparison with the behaviour of normal undisturbed samples would reveal something of the relative roles of the bonds and gel.
Consolidation tests on six undisturbed dried, resaturated oedometer specimens of Soil 2 produced the following results. (In the following discussion low pressures and high pressures
refer to applied pressures which are less than or greater than the critical pressure, respectively.)
(a) Removal of the gel only slightly lowered the critical pressure from 3.5 to 2.8 ton/sq. ft.
This reduction could be explained by the greater disturbance of the dried-resaturated
specimens.
(b) Compressibility at high pressures was not affected by removal of the gel.
(c) Consolidation of the dried-resaturated soil was almost instantaneous.
This is due to
the large increase in permeability noted earlier in this Paper.
(d) Removal of the gel reduced the creep rate at high pressures to about 20% of that of the
natural soil. Creep rates for the dried-resaturated soil at low pressures were similar to
those at high pressures.
General structural behaviour

Before discussing the observed soil behaviour in terms of the soil structure the earlier conclusions concerning the structural behaviour of the undisturbed natural soil are summarized
as follows.
(e) At low pressures the soil compressibility is low and is similar to the rebound.
(f) The volumetric strain at high pressures in the oedometer test was directly proportional
to _4(n,,-B), where n,, is the initial porosity, B is an apparent minimum porosity and A is
the flexibility of the soil skeleton.
(g) The creep rate at low pressures is approximately linearly proportional to the applied
pressure increment.
(h) The creep rate at high pressure is roughly ten times greater than at low pressures and
is reasonably independent of applied pressure increment.
(i) The undrained shear strength is comparatively high and is not affected by soaking.
(j) The soil is relatively insensitive.
(k) Drained shear strength is comparatively high.
From the preceding statements it can be interpreted that the bonds are independent of the
viscous gel and that at low pressure the bonds are intact and the structural behaviour is controlled by the stiff, elastic soil skeleton. Sangrey and Townsend (1969) have shown that the

216

K. B. WALLACE

volumetric strain of intact Canadian clays is directly proportional


to applied pressure.
The
lower compressibility
of the present soils together with normal experimental
variability precludes any more definite conclusions than that the volumetric strain of the present soils, when
intact, increases with applied pressure and is to a large extent recoverable.
The strength of the cemented bonds in residual soils is quite variable, as is shown by the
variation
in critical pressures and undrained
shear strengths.
This does not allow close
definition of yield criteria such as those given by Sangrey (1972a) for finer grained, more
homogeneous,
cemented sedimentary soils.
General field observations
of residual soils indicate that even if rough criteria could be established in the laboratory, these criteria would be
modified by discontinuities
at the design site.
The dependence of the rate of creep of the intact soil on the applied pressure, together with
the similarity of the creep of dried-resaturated
soil above and below the critical pressure,
suggests that at low pressures the creep mechanism is one of readjustment
of stresses in the
primary skeleton and fracture of the weaker bonds.
Similar progressive microfracturing
has
been studied in detail in investigations
into the creep of brittle rock at low temperature
and
pressure (Scholz, 1970).
The effect of the viscous gel in producing a much greater and apparently pressure-independent creep at high pressures could be attributed to a micropore mechanism (Barden, 1968) and
indicates that when the bonds are fractured the gel supports a portion of the load at contacts
between the coarser particles.
The low sensitivity and high drained shear strength indicate
that, even in a highly remoulded state, there is considerable interlocking of the coarse particles
and that the proportion
of the total load carried by the gel is small.
This indicates that the
viscous gel can drain freely through the porous skeleton and prevent any significant pore pressures on shear planes.
The high value of apparent minimum porosity indicated by oedometer
tests suggests that the porosity of the soil will be high, even in a highly compressed or sheared
state.
However, the Author has observed several specimens in which the soil skeleton has
tended to collapse to much lower porosities under repeated shear stresses.
From this discussion, it is concluded that the structural behaviour of the soil is consistent
with the proposed model of the soil structure and that, at low applied stresses and, to a lesser
extent, at high applied stresses it is the porous soil skeleton which is most important in determining structural behaviour.

CONCLUSION

The conclusions arrived at in the preceding discussion are based on a limited number of tests
on residual soils formed in a continually wet environment.
Compression and shear tests have
concentrated
on the sandier volcanic soils for which it was easier to prepare good undisturbed
samples but similar results have been obtained on fewer samples of volcanic silty clay and the
more friable silty clay on limestone.
All the soils tested were well-drained lateritic soils. It is
appropriate
therefore to conclude by commenting
on the breadth of application
of the conclusions reached in this Paper.
From the literature it appears that many residual soils have the cemented structure idealized
herein.
High shear strength and a link between shear strength and degree of weathering are
frequentlyreported
for residual soils. It is expected that point to point variability of soil properties will be high for most residual soils. The primary dependence of compressibility
on
the porosity and critical pressure is well established for a variety of porous sedimentary and

STRUCTURAL

BEHAVIOUR

OF RESIDUAL

SOILS OF THE CONTINUALLY

WET HIGHLANDS

OF PAPUA

NEW

GUINEA

217

residual soils. Low sensitivity whenremoulding at field moisture content is associated with the
coarse cemented soil structure but it is not known to what extent residual soils can be generally
assumed to have a coarse, porous skeleton, although it is considered that such a structure is
associated with formation under well-drained conditions.
One aspect of the behaviour of the present soils which is thought to be peculiar to a much
narrower range of residual soils is the stability of the cemented bonds on soaking or drying.
This is thought to be associated with soil formation under continually wet conditions and will
not apply to soils formed under seasonal climates. Sowers (1963) postulates a mechanism of
fracture of bonds through unequal expansion of low plasticity residual soils when wetted, while
the Author has observed that residual soils in the seasonally wet Eastern Highlands of New
Guinea are much more difficult to sample in an undisturbed state and disintegrate readily on
soaking.
Accumulated soils engineering experience with sedimentary soils has shown that correlations
of the properties of the soils with simple parameters are extremely useful for determining relavant description and classification systems and for preliminary estimates of soil properties,
but are subject to typical variations of & 30% around the predicted values. Residual soil
properties will be more variable but any simple working generalizations on structural behaviour
will be valuable aids in soils engineering. Such generalizations can only come from wider
discussion of the formation, nature and structural behaviour of residual soils from a variety of
sources. The Author believes that when the general characteristics of residual soils are better
known much of their behaviour will be interpretable in terms of our much wider knowledge of
the mechanics of sedimentary soils.
ACKNOWLEDGEMENTS

The work that has been described in this Paper was completed while the Author was a
member of the staff of the Papua New Guinea Institute of Technology.
The Author is grateful
to the Institute for its generous support of the project. The geotechnical tests were carried out
in the Institutes Soil Mechanics Laboratories with the assistance of Mr M. J. Roland, Mr J. P.
Rizzi and Mr J. Gaya.
The valuable support of the Southern Highlands District Commissioner, Mr Des Clancy, and
his staff, of the Catholic Mission, Mendi, and of civil engineering students Mr John Kavagu,
Mr Gavera Morea and Mr Thomas Tohiana during the field work is also acknowledged.
REFERENCES
Australian Mineral Development Laboratories (1971). Clay mineral analysis of eight soil samplesfrom the
New Guinea Highlands. Unpublished Report MP2033/71.
Adelaide.
Barden, L. (1968). Primary and secondary consolidation of clay and peat. Geotechnique 18, NO. 1, l-24.
Birrell, K. S. (1952). Some physical properties of New Zealand volcanic ash soils. Proc. 1st A.N.Z. Conf.
Soil Mech.,

Melbourne,

30-34.

Blake, D. H. and Liiffler, E. (1971).

and New Guinea.


Frost, R. J. (1967).
Conf. Soil Mech.,

Volcanic and glacial landforms

Geol. Sot. Amer.

Importance
Bankok,

on Mount Giluwe, Territory of Papua

Bull. 82, 1605-1614.

of correct pretesting preparation

of some tropical soils.

Proc. 1st S.E. Asian

43-53.

Gradwell, M. W. & Birrell, K. S. (1954). Physical properties of certain volcanic clays. New Zealand Jnl
Sci. Technol. B36, No. 2, 37-48.
Haantjens, H. A. &Rutherford,
G. K. (1964). Soil zonality and parent rock in a very wet tropical mountain
region.
Proc. 8th Znt. Cong. Soil Sci., Bucharest 493-500.
Kay, J. N. & Krizek, R. H. (1971). Analysis of uncertainty in settlement prediction.
Geotech. Engng 2,
119-129.

218

K. B. WALLACE

Koizumi, Y. & Ito, K. (1963). Compressibility of a certain volcanic clay. Soil and foundution 3, No. 2,
37-48.
Lambe, T. W. & Whitman, R. V. (1969). Soil mechanics 318-321. New York: Wiley.
Lohnes, R. A., Fish, R. 0. & Demeril, T. (1971). Geotechnical properties of selected Puerto Rican soils in
relation to climate and parent rock. Geol. Sot. Amer. Bull. 82, Sept, 2617-2624.
Lumb, P. (1962). The properties of decomposed granite.
Geotechnique 12, No. 3, 226-243.
Martin, F. J. & Doyne, H. C. (1927). Laterite and laterite soils in Sierra Leone. Jnl Agric. Sci. 17, 530-547.
Mehra, 0. P. & Jackson, M. L. (1960). Iron oxide removal from soils and clays by a dithionite-citrate
system buffered with sodium bicarbonate.
Proc. 7th Nat. Conf. Clays Clay Miner. 317-327. London:
Pergamon Press.
Miki, G. (1960). Geotechnical properties of volcanic ashes occurring in Japan.
Proc. Asian Reg. Conf Soil
Mech., Delhi 1-8.
Newill, D. (1961). A laboratory investigation of two red clays from Kenya.
Geotechnique 11, No. 4,
302-318.

Osterman, J. (1959). Notes on the shearing resistance of soft clays. Acta Polytechnica. Stand. No. 263.
Penta, F., Croce, A. & ESU, F. (1961). Engineering properties of volcanic soils. Proc. 5th Znt. Conf. Soil
Mech. 1, 285-291.
Quigley, R. M. (1968). Discussion on: Landslide on the Toulnustouc River, Quebec by R. J. Conlon.
Can. Geotech. Jnl5, No. 3, 175-177.
Ruxton, B. P. (1968). Measures of the degree of chemical weathering of rocks. Jnf Geoi. 76, 518-527.
Sangrey, D. A. (1972b). On the causes of natural cementation in sensitive soils. Can. Geotech. Jnl9, No.
117, 117-119.
Sangrey, D. A. (1972a). Naturally cemented sensitive soils. Geotechnique 22, No. 1, 139-152.
Sangrey, D. A. & Townsend, D. L. (1969). Characteristics of three sensitive Canadian chzys. Research
Department of Civil Engineering.
Report No. 63. Ontario, Canada: Queens University at Kingston.
Proc. 2nd Cong. Znt. Sot. Rock Mech.
Scholz, C. (1970). Role of microfracturing in rock deformation.
2-8.
Skempton, A. W. (1957). Discussion on: The Planning and design of the new Hong Kong airport.
Proc.
Znstn Civ. Eng. 7, 305-307.
Sowers, G. F. (1963). Engineering properties of residual soils derived from igneous and metamorphic rocks.
Proc. 2nd Pan Amer. Conf. Soil Mech. Brazil 1, 39-62.
Terzaghi, K. (1958). Design and performance of Sasamua Dam. Proc. Znstn Civ. Engrs 9, 369-394.
Trollope, D. H. & Chan, C. K. (1959). Soil structure and the step-strain phenomenon.
JnI Soil Mech.
Fdns Div. Am. Sot. Civ. Engrs 86, No. SM2, l-39.
Vargas, M. (1953). Some engineering properties of residual clay soils occurring in Southern Brazil. Proc.
3rd Znt. Conf. Soil Mech., Zurich 1, 67-71.
Wallace, K. B. (1970). Systematic description of soils for road construction in Papua and New Guinea.
42nd ANZAAS tong., Port Moresby. Paper No. 6, Engineering Section.
Wallace, K. B. (1971). Residual soils of the continually wet Highlands ofPapua New Guinea. Civil Engineering Department Bulletin No. 2. Lae, New Guinea: Papua New Guinea Institute of Technology.

Das könnte Ihnen auch gefallen