C O L L A P S E OF COMPACTED CLAYEY SAND

By Evert C. Lawton, 1 Richard J. Fragaszy, 2 and James H. Hardcastle,3

Members, ASCE

ABSTRACT: Compacted soils wetted under load can both swell and collapse
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

(subside) depending on their condition and the magnitude of the vertical overbur-
den stress. One-dimensional compression tests were conducted to clarify the in-
fluences of compaction method, compaction water content, relative compaction,
vertical stress level, and load-wetting sequence on post-compaction wetting-in-
duced volume changes in a moderately plastic clayey sand. Compaction method
and load-wet sequence had only a minor effect on wetting-induced collapse. The
double-odometer test was judged to be sufficiently accurate for use in evaluating
wetting-induced collapse. Both swelling and collapse were reduced or eliminated
by compacting the soil at water contents on the wet side of the line of optimums
for impact compaction. Collapse, but not swelling, could also be reduced by com-
pacting the soil to high levels of relative compaction. By plotting isograms of
volume changes in the compaction water content-relative compaction space, com-
binations resulting in no wetting-induced volume change were identified for var-
ious vertical stress levels corresponding to the values existing in large fills and
embankments.

INTRODUCTION

Increased use of very thick fills and high embankments constructed of

clayey soils has resulted in the creation of earth structures which swell when
wetted at low applied stresses and compress when wetted at high stresses.
Volume decreases accompanying increases in water content at essentially
unchanging total stresses in loose, partly saturated natural soil deposits have
been termed collapse, hydroconsolidation and hydrocompaction. That col-
lapse can occur also in engineered fills has been demonstrated in both the
field and the laboratory (Booth 1977; Cox 1978; Lawton 1986). The follow-
ing observations and conclusions regarding wetting-induced collapse in soils
have been reported in the literature:

1. Four conditions are necessary for collapse to occur in a soil (Barden et al.
1973; Mitchell 1976):
a. An open, partially unstable, partially saturated fabric.
b. A high enough total stress so that the structure is metastable.
c. A sufficiently large soil suction or the presence of a bonding or ce-
menting agent that stabilizes the soil in the partly-saturated condition.
d. The addition of water to the soil which reduces the soil suction or soft-
ens or destroys the bonding agent, thereby causing shear failures at the inter-
aggregate or intergranular contacts.
2. For any given set of conditions, the amount of collapse generally decreases
'Asst. Prof., Dept. of Civ. and Arch. Engrg., Univ. of Miami, P.O. Box 248294,
Coral Gables, FL 33124.
2
Assoc. Prof., Dept. of Civ. and Envir. Engrg., Washington State Univ., Pullman,
WA 99164.
3
Prof., Dept. of Civ. Engrg., Univ. of Idaho, Moscow, ID 83843.
Note. Discussion open until February 1, 1990. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on July
2, 1987. This paper is part of the Journal of Geotechnical Engineering, Vol. 115,
No. 9, September, 1989. ©ASCE, ISSN 0733-9410/89/0009-1252/$1.00 + $.15
per page. Paper No. 23860.

1252

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 with increasing pre-collapse moisture content, increasing pre-collapse dry den-
sity, and decreasing overburden pressure (Holtz 1948; Booth 1977; Cox 1978).
3. For any soil there are combinations of initial dry density, molding water
content, and overburden pressure at which no volume change will occur when
the soil is inundated (Booth 1977; Cox 1978).
4. For a given soil there appears to be a critical molding water content at
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

soaking above which no collapse will occur. For some soils the critical water
content is above standard Proctor optimum water content (Hilf 1956; Barden et
al. 1969).
5. There is a critical degree of saturation for a given soil above which neg-
ligible collapse will occur regardless of the magnitude of the pre-wetting over-
burden pressure (Mishu 1963; Booth 1977).

The present paper describes the results of a laboratory research program

conducted to clarify some aspects of the wetting-induced collapse behavior
of compacted clay fills. The following aspects of the collapse phenomenon
are specifically addressed in this paper:

1. The significance of the laboratory test procedures with regard to the con-
clusions reached in the evaluation of collapse potential. Important test procedure
considerations include the method of compaction used to prepare test specimens
and the load-wetting sequence used for the collapse test.
2. The influence of the compaction parameters such as relative compaction
and compaction water content on the amount of wetting-induced collapse exhib-
ited by clayey soils compacted to usual unit weights and subjected to stress levels
typically encountered in deep fills and high embankments. The objective is to
identify the critical compaction parameters tiiat can be controlled by the fill de-
signer in order to minimize potential wetting-induced collapse.

The research program consisted of a large number of one-dimensional

loading-wetting (collapse) tests performed on a expansive soil taken from a
deep fill in southern California. Conclusions reached from the test program
are presented following a brief description of the soil, test apparatus, and
test procedures. Complete details of the apparatus and test procedures can
be found in Lawton (1986).

SOIL AND METHODS

A slightly expansive soil classified as SC, clayey sand, in the Unified Soil
Classification System and A-6 (2) in the AASHTO System was used in the
investigation. The soil had a maximum grain size of 2.0 mm, a median size
of 0.15 mm, and contained 15 percent by weight particles smaller than 0.002
mm. The liquid limit and plasticity index were 34 and 15%, respectively.
The test program consisted mainly of double-odometer collapse tests con-
ducted on nominally identical test specimens compacted directly into odome-
ter rings using either impact, kneading, or static methods. Moisture-density
relationships developed from standard Proctor and modified Proctor impact
test procedures (ASTM D 698 and D 1557) for the soil are shown as solid
lines in Fig. 1. The majority of the tests were conducted on specimens com-
pacted using special scaled-down impact hammers designed to deliver the
same energy per unit volume of compacted soil as the conventional standard
and modified Proctor hammers. As shown by the heavy dashed curves of

1253

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 2.10

*™™-™- P r o c t o r I m p a c t
— —- S c a l e d - D o w n I m p a c t
Upper Curves: Mod. Proctor Energy 100
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

2.00- .Lower Curves: S t ' d Proctor Energy

1.90

1.80

>-
CXI

a
1,70

1.60-
10 15 20
MOLDING WATER CONTENT (%)

FIG. 1. Moisture-Density Relationships

Fig. 1, compaction with the special hammers produced lower maximum dry
densities than with the conventional compaction hammers. This result is ex-
pected because more friction per unit volume of soil is developed during
compaction into the odometer rings as compared to compaction into con-
ventional compaction molds (due to the higher ratio of inside surface area
to volume of soil for the odometer rings). However, the moisture-density
relationships developed for the special hammers are essentially identical in
form and have approximately the same line of optimums as for the conven-
tional hammers. In this paper all references to relative compaction will refer
to maximum dry density produced by the modified Proctor procedure de-
scribed in test method ASTM D 1557. The term "line of optimums" (illus-
trated in Fig. 1) will likewise refer to the line connecting coordinates of
optimum water content and maximum dry density for moisture density re-
lationships produced by the standard tests (ASTM D 698 and D 1557). For
the soil tested, the line of optimums corresponded to a constant degree of
saturation, S, of approximately 80%.
In a double odometer collapse test, one of a pair of identically compacted
specimens is loaded in the as-compacted condition to successively increasing
1254

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 +10
+8 A Soaked
+6 O As-Compacted

+4
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

+2
0
I—
CO -2
-4
I -6
-8
Relative Compaction = 80%
-10 Molding Water Content = 10%
-12 Impact Compaction
-M
' ' i i i i i i I i t I i i 111
10 20 50 100 200 500 1000 2000
OVERBURDEN PRESSURE (KPA)

FIG. 2. Volumetric Strain versus Overburden Pressure Curves for As-Com-

pacted and Soaked Double Odometer Tests

vertical (overburden) stresses and allowed to attain equilibrium at each level

of vertical stress. Before the second specimen is loaded, it is inundated with
water under a small seating load. After the wetted specimen has stopped
swelling and therefore has come to an equilibrium water content, it is also
loaded sequentially to the same vertical stresses used for the unsoaked spec-
imen, and permitted to achieve equilibrium at each level of vertical stress.
The difference between the equilibrium void ratios of the two specimens at
each value of vertical stress is used to calculate the volumetric strain dif-
ference. The value of volumetric strain difference at any value of vertical
stress corresponds to a prediction of the volumetric strain that would occur
if the soil were loading to that stress in the as-compacted state and then
soaked with water. At stress levels where the wetted specimen has a larger
void ratio than the as-compacted specimen, the strain difference is positive
and the as-compacted soil is considered to have swelled. Negative strain
differences indicate that the wetted specimen has a smaller volume than the
as-compacted specimen. At stress levels where the strain differences are neg-
ative, the soil is considered to be susceptible to wetting-induced collapse.
Typical double-odometer test results are shown in Fig. 2. An important char-
acteristic of collapsible compacted soils is illustrated in Fig. 2; that is, the
magnitude of potential wetting-induced collapse for a soil compacted at a
given molding water content and a given dry density is a maximum at some
value of overburden pressure and decreases at overburden pressures higher
than this value. The maximum potential collapse can be seen in Fig. 2 to
occur at the overburden stress where the vertical distance between the two
curves is a maximum, and the potential collapse decreases at higher stresses

1255

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 +16 —
+11-
+12-
CD Double Oedometer
+10-
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

O Soaked-After-Loading
+8-
+6-
+4-
C/0 +2 -
0 —
-2"
-4-
-6-
-8 "
-10 -
JjJ- _J_ I i i 111
10 20 50 100 200 500 1000 2000
OVERBURDEN PRESSURE (KPA)
FIG. 3. Comparison of Volumetric Strain between Double Odometer and Soaked-
After-Loading Tests—Static Compaction (Relative Compaction = 80%, Molding
Water Content = 13%)

as the two curves approach each other. This characteristic can be explained
in terms of the prestress induced in the soil during the compaction process
and will be discussed in a later section of the paper.
The double odometer tests for this test program were performed using
Bishop-type loading units and 75 mm diameter by 20 mm thick fixed rings.
In order to facilitate the direct determination of the influence of dry density
and molding water content on the characteristics of collapse for this soil, the
double odometer tests were performed on pairs of identically compacted
specimens prepared at selected water contents (7, 10, 13, 16, 19, 22%), and
the applied energy was varied as necessary to produce values of dry density
corresponding to 80, 85, 90, and 95% relative compaction.

Accuracy of Double Odometer Collapse Test

The results of a typical double-odometer collapse test on pairs of speci-
mens statically compacted to a dry density corresponding to a relative com-
paction of 80% are shown in terms of volumetric strain difference in Fig.
3. Also illustrated in Fig. 3 is the collapse behavior of specimens tested
using a procedure that more closely simulates the load-wetting history of
compacted fills in the field. In this latter soaked-after-loading procedure,
identically prepared specimens are loaded in increments in the as-compacted
condition to various stress levels, allowing equilibrium to be attained under
each stress increment. Once stress equilibrium is obtained at the desired stress
level for a specimen, the specimen is inundated (soaked) and allowed to
achieve moisture equilibrium. Collapse (or swell) is calculated as only that
1256

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 +16
+14
+12 Q Double Oedometer

+10 O Soaked-After-Loading
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

LU
3
LU
+8
^~*
E-9
LU
LL.
+6
LU
^ +4
«r =a
I—
00 o;
2:
—1
+?
I > o;
1—
CO
(J
cc
\—
III 0 -2
VOLU

I—
LU
-4
•sz
=3
_l
0
-6
> -8
-10
J I ' I1 '1 I J_ J 1 I I III I
10 20 50 100 200 500 1000 2000
OVERBURDEN PRESSURE (KPA)
FIG. 4. Comparison of Collapse between Double Odometer and Soaked-After-
Loading Tests—Kneading Compaction (Relative Compaction = 80%, Molding Water
Content = 13%)

part of the specimen deformation accompanying the wetting process. Fig. 3

indicates that while the double odometer test overestimated the swelling strains
accompanying wetting, the collapse deformations predicted from the double
odometer tests for statically compacted samples are similar to those mea-
sured in the soaked-after-loading tests. Similar results in Fig. 4 for speci-
mens compacted using kneading compaction show good correlation between
the results from the double odometer and soaked-after-loading tests, but not
as good as for the statically compacted specimens. The better correlation for
the statically compacted specimens is attributable to the smaller chance for
random variation in the static compaction procedure. In consideration of these
and similar results reported by Booth (1977) and Justo, Delgado, and Ruiz
(1984), it was decided that the testing economy offered by the double odom-
eter procedure for collapse more than offsets any increase in accuracy pro-
vided by the soaked-after-loading approach. To obtain the same information
derived from one double odometer test conducted on two samples in this
study, seven soaked-after-loading tests conducted on seven samples would
have been required. All subsequent results described in this paper are there-
fore based on results of double-odometer collapse tests.

Influence of Compaction Method

Beginning with the work of Seed and Chan (1959), it has become gen-
erally accepted that the method of compaction can have a significant influ-
1257

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 +16
+M
+12 Q STATIC

5 + ioi- V IMPACT
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

O KNEADING
y +8-
LlJ

3 +6
t +4

I 0 &=-=r-^-
-2
-H
-6 -R.C.=80%, w=10%
-8 -R.C.=85%, w=16

-10
J I I I I I.I I _L J„ ' • i i i I

10 20 50 100 200 500 1000 2000

OVERBURDEN PRESSURE (KPA)
FIG. 5. Effect of Compaction Method on Volumetric Strain Difference versus
Overburden Pressure Curve for (Relative Compaction = 80%, Molding Water Con-
tent = 10%) and (Relative Compaction = 85%, Molding Water Content = 16%)

ence on the properties of compacted clays. In order to examine the influence

of the compaction method on the collapse behavior of the sandy clay used
in this study, double-odometer tests were performed on otherwise identical
specimens compacted with static and kneading methods as well as with im-
pact compaction. Comparison of the collapse behavior for specimens com-
pacted to a relative compaction of 80% at a molding water content of 10%
and for specimens compacted to a relative compaction of 85% at a molding
water content of 16 percent are shown in Fig. 5. In addition to illustrating
a widely recognized characteristic of the collapse behavior of compacted
soils, namely, that soils compacted at the lower water contents exhibit more
collapse than soils compacted at higher water contents, Fig. 5 indicates that
the method of compaction has relatively minor influence on the collapse
behavior. This result is consistent with the relationships among compaction
methods, molding water content, and induced shear strains suggested by Seed
and Chan (1959). In their model, Seed and Chan concluded that the structure
of a soil compacted on the dry side of the optimum water content is essen-
tially unaffected by compaction method because the shear strains induced by
any of the commonly used methods are small. As will be demonstrated in
later sections of this paper, compacted soils that exhibit wetting-induced col-
lapse are restricted to those compacted on the dry side of the line of opti-
mums for impact compaction. Thus it appears that any convenient compac-
tion method can be used to produce test specimens for evaluation of wetting-

1258

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 caused collapse potential of compacted soils. Of course, the compaction method
does have significant influences on many other engineering properties of
compacted clays, especially when a wide range of compaction water contents
are considered. Examination of the influences of compaction parameters in
the remaining sections of this paper are made for specimens compacted using
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

the scaled-down impact hammer.

INFLUENCES OF COMPACTION PARAMETERS ON COLLAPSE

Compaction Water Content

The importance of compaction water content on both the amount of col-
lapse and the vertical overburden stress level at which the maximum amount
of collapse occurs is clearly illustrated in Figs. 6 and 7. (The term critical
overburden stress will be used hereinafter to refer to the overburden stress
at which maximum collapse occurs for a given molding water content and
a given relative compaction.) The drier the soil was at compaction, the greater
was both the maximum collapse or swelling at a particular vertical over-
burden stress and the critical overburden stress. Compacting at water con-
tents near or wet of the line of optimums reduced collapse as well as swell-
ing. The wetting induced collapse was reduced because the wetter, more
compressible as-compacted soil had already experienced a substantial amount

+16 - A w - 7%

+14 - 0 w = 10%
a w = u%
+12
^ w = 16%
LU
+10
+8 Note: Tests a t w => 19% and 22% predicted
negligible volumetric s t r a i n a t a l l
+6 overburden pressures.

+4 S ^ v .
+2 ^^it^.
0 ~~^ V ~~^ V > H
TC^-W jy" ^
-2 - \ ^ V y ^ V ^
-4 No^v sr
-6
o ^NH^T
-8 \ ^ ® ^ /
-10
i 1 1 1 1 1 11 1 1 1 1 1 l 1 1 1 P~~-v>,
10 20 50 100 200 1000 2000
OVERBURDEN PRESSURE (KPA)

FIG. 6. Effect of Molding Water Content on Volumetric Strain Difference versus

Overburden Pressure Curve at Relative Compaction = 80% (Impact Compaction)

1259

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 +16
A w = 7%
+14 Q w = 10%
+12
3 e w » 13%
V w = 16%
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

UJ +10 h
I +8 -
t +6

+
£ 2
& o
o
5 -2

-6
Note: Tests at w = 19% predicted negligible
-8 volumetric strain at ail overburden
pressures.
•10
J ' ' ' i) iI _L _L I » i i i I.
10 20 50 100 200 1000 2000
OVERBURDEN PRESSURE (KPA)
FIG. 7. Effect of Molding Water Content on Volumetric Strain Difference versus
Overburden Pressure Curve at Relative Compaction = 85% (Impact Compaction)

of load-induced compression before being permitted to take up additional

water. An explanation for the fact that both the maximum amount of collapse
and the critical overburden stress decreased with increasing compaction water
content is not immediately apparent. As is discussed in a later section of the
paper, this result is believed to be related to the inverse relationship between
compaction water content and the magnitude of the compactive prestresses
induced in specimens compacted to the same dry density with a particular
compaction method.

Relative Compaction
The influence on collapse behavior of the second major compaction pa-
rameter, relative compaction, is shown in Fig. 8. The curves show that for
the soil tested in this study, increasing the relative compaction not only sub-
stantially reduced the wetting-induced collapse potential at any overburden
pressure in the collapse region, it also increased the critical overburden pres-
sure at which the collapse was a maximum. However, unlike the technique
of using increased compaction water content to reduce collapse, the use of
increased relative compaction resulted in significantly increased wetting-caused
swelling of the soil.
As is frequently the case with compaction parameters, selection of values
to optimize a particular engineering property or behavior can result in less
than desirable characteristics for other behavior. In order to illustrate how
1260

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 +16
+14
s +12
+10
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

UJ

!_>
REN

+8
HI
u_
1—1
+b
Q
+4
z
ss +2
1—
CO 0
<. >
cc -2
h-
LU

=D
-4
-J
CD -6
>
-8
-10
I ' I t-JULuJ U 1 I L_L_L-UJ
10 20 50 100 200 500 1000 2000
OVERBURDEN PRESSURE (KPA)
FIG. 8. Effect of Relative Compaction on Volumetric Strain Difference versus
Overburden Pressure Curve at Molding Water Content = 10% (Impact Compac-
tion)

the optimal values of compaction water content and relative compaction could
be selected to minimize both collapse and swelling, isograms of wetting-
caused volume changes can be plotted as in Fig. 9. The isograms of the
figure are for the net volumetric strain differences accompanying wetting of
the compacted soil at an overburden stress level of 400 kPa.
Two important characteristics of the collapse behavior of the expansive
clayey sand are illustrated in Fig. 9 and similar plots made for other over-
burden pressures by Lawton (1986). First, in all of these plots the isograms
corresponding to the smallest measurable amount of wetting-induced col-
lapse are confined to a region representing compaction water contents dry
of the line of optimums. In other words, no isogram representing wetting-
induced collapse crosses the line of optimums. Because the line of optimums
also represents approximately a line of constant degree of saturation, it can
be stated (for this soil) that no post-construction wetting-induced collapse
will occur if the soil is compacted to a density and water content for which
the degree of saturation is greater than or equal to about 80 percent.
The second important characteristic illustrated by the isograms of Fig. 9
is that there is a region in the compaction water content-relative compaction
space where neither collapse nor swelling will occur as the soil is wetted
under load. The location of this no volume change region depends on the
1261

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 100
-2.00

LINE OF OPTIMUMS
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

-1,90

1,80

-1.70

1,60
10 12 14 16 18 20 22
MOLDING WATER CONTENT (%)
FIG. 9. Isograms of Volumetric Strain Difference as Function of Relative Com-
paction and Molding Water Content at Overburden Pressure = 400 kPa (Impact
Compaction)

overburden stress level at which post-compaction wetting takes place. This

means that there is a range in relative compactions (or compacted densities)
for each compaction water content and overburden pressure where no wet-
ting-induced volume changes will occur. This range of relative compactions
(or densities) becomes larger as the compaction water contents approach the
line of optimums. As observed above for wetting-induced volume decreases
(collapse), this range includes all reasonable values of relative compaction
when the compaction water content is on the wet side of the line of opti-
mums.
For a given overburden pressure, a horizontal line drawn through the cen-
ter of the region of no volume change, as shown in Fig. 9 as a short dashed
line, would represent a critical value of relative compaction. The critical
relative compaction represents a dry density for which no volume change
will occur in the soil upon wetting regardless of the molding water content.
The influence of the overburden pressure on the critical relative compaction
is illustrated in Fig. 10. Not surprisingly, the relative compaction must be
increased if collapse is to be prevented at high overburden stresses without
compaction water content control.

1262

J. Geotech. Engrg., 1989, 115(9): 1252-1267

Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

&•«

Q_
O

I
UJ
ce

•a:

200 500 1000 2000

OVERBURDEN PRESSURE (KPA)
FIG. 10. Critical Relative Compaction versus Overburden Pressure (Impact Com-
paction)

Isograms of wetting-induced volume changes can also be plotted using

compaction or prewetting degree of saturation instead of water content. When
plotted in this way, the line of optimums appears as a vertical line of constant
degree of saturation bounding the region of no collapse. Since a single value
of degree of saturation is adequate to describe the boundary of the region
of no collapse, it would appear to be more useful to characterize compaction
conditions in terms of critical degree of saturation rather than in terms of
the several different critical compaction water contents that represent the line
of optimums.

COMPACTIVE PRESTRESS AND CRITICAL OVERBURDEN PRESSURE

In an earlier section of the paper a critical overburden stress was defined

as the value of the vertical stress at which the maximum amount of collapse
occurs for a given molding water content and a given relative compaction.
Figs. 6 and 7 showed the inverse relationship between critical overburden
stress and compaction water content and Fig. 8 illustrated the dependence
1263

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 of the critical overburden stress on relative compaction. Results of tests con-
ducted in this study support the conclusion of Witsman and Lovell (1979)
that the critical overburden stress is also equal to a property of compacted
cohesive soils referred to as compactive prestress.
Compactive prestress is an apparent preconsolidation pressure exhibited
by both partly-saturated and saturated compacted clays in one-dimensional
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

compression tests. The compactive prestress depends on the form and amount
of mechanical energy used to compact the soil. The magnitude of the com-
pactive prestress can be estimated using the graphical techniques used to find
the preconsolidation pressure of undisturbed samples of natural soils. That
the compactive prestress exhibited by the as-compacted soil is also the ver-
tical stress at which the potential for maximum wetting-induced collapse ex-
ists is apparent from Fig. 2. In this figure the compactive prestress of the
as-compacted specimen is roughly the value of the vertical stress where a
tangent to the nearly horizontal early part of the compression curve intersects
the tangent drawn for the steeply sloping straight-line late part of the curve.
Alternatively, the compactive prestress can be estimated as a value of ver-
tical stress slightly larger than the value at which the rate of curvature of
the compressive curve is a maximum (Casagrande technique). Using either
of these approaches, the compactive prestress of the soil of Fig. 2 is ap-
proximately 800 kPa. Since the slope of the compression curve for the wet-
ted specimen is invariably flatter than the slope of the steepest part of the
compression curve of the as-compacted specimen, it is apparent that upon
soaking the maximum collapse as predicted by the double-odometer tech-
nique for a relative compaction of 80% and a molding water content of 10%
occurs at the vertical stress identified above as the compaction prestress. It
is also clear from Fig. 2 that allowing the soil to swell as it takes up water
effectively eliminates the compactive prestress.
That the maximum collapse potential for a given compacted condition ex-
ists at an overburden stress equal to the compactive prestress can be ex-
plained in terms of the overburden stress, dry density, and degree of satu-
ration at the time water is added to the specimen. At overburden stresses
lower than the compactive prestress, very little compression takes place in
the as-compacted soil due to the overburden pressure and the dry density
and degree of saturation at all values of vertical stress are approximately the
same. Therefore, the potential for wetting-induced collapse increases with
increasing vertical stress. However, if the overburden stress exceeds the level
of compactive prestress, more compression occurs due to the application of
the overburden pressure, resulting in an increased dry density and a higher
degree of saturation. Both these characteristics reduce the potential for wet-
ting-induced collapse. At some value of overburden stress the soil is com-
pressed to a degree of saturation greater than critical saturation and there is
no effect when water is added.
If the relationship between compactive prestress and overburden stress for
maximum collapse is accepted, the relative values of the maximum collapse
for specimens compacted by kneading, impact, and static compaction shown
in Fig. 5 indicate that the prestressing effect of the compaction method in-
creases in the order kneading, impact and static. Although the differences
are small for the moderately plastic soil used in this study, this result may
support the view held by many engineers concerning the efficiency of static
compaction. That is, more energy is required to compact a cohesive soil to

1264

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 a given density using static (smooth wheel rollers) compaction than is re-
quired to compact the soil to the same desntiy using a kneading (sheepsfoot
roller) type compaction. However, once the given density is achieved, the
statically compacted material will be less compressible under load at vertical
stresses smaller than the compactive prestress. The statically compacted soil
with the larger compactive prestress will also exhibit a larger maximum wet-
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

ting-induced collapse and this maximum collapse will occur at a slightly

higher critical overburden stress than for the impact or kneading compacted
soil.
The curves of Figs. 6 and 7 showing the inverse relationships among the
maximum amount of collapse, the overburden stress at which it occurs, and
the compaction water content are also consistent with the relationships be-
tween compactive prestress'and compaction water content observed by Wits-
man and Lovell (1979). For soil specimens compacted to the same dry den-
sity using the same method, Witsman and Lovell found that the apparent
compactive prestress decreased with increasing compaction water content.
Because the slope of the compression curve of the as-compacted specimen
becomes more like that of the wetted specimen as the compaction water
content is increased toward optimum, it is apparent that the vertical differ-
ences between the two curves, shown in Fig. 2, and hence the maximum
amount of collapse also must decrease along with the compactive prestress.

SUMMARY AND CONCLUSIONS

Interactions between the compaction variables (water content, dry density

or relative compaction, and compaction method) and the post-compaction
stress and moisture regimes are complex and not yet completely understood.
This paper has presented the results of an experimental study designed to
clarify some of the relationships between compaction variables and the wet-
ting-induced collapse behavior of a moderately plastic clayey sand which
also exhibited wetting-induced swelling under different stress conditions. The
following conclusions concerning the collapse behavior of this soil were de-
veloped from the study:

1. Depending on the level of overburden stress at which it was wetted, the

moderately plastic soil of this study exhibited both volume increases (swelling)
and volume decreases (collapse) while sustaining essentially unchanged vertical
stresses.
2. At the stress levels where wetting induced volume changes were observed,
the amount of swelling or collapse varied inversely with the compaction water
content. At a given dry density, the overburden stress level at which the max-
imum amount of collapse took place also varied inversely with the compaction
water content.
3. Collapse as well as swelling could be eliminated for all reasonable levels
of overburden pressure which occur in fills of usual dimensions by compacting
the soil at a water content greater than or equal to the optimum impact com-
paction water content corresponding to the compacted dry density. Stated in dif-
ferent terms, collapse and swelling could be eliminated by compacting the soil
to a degree of saturation greater than or equal to the degree saturation corre-
sponding to the line of optimums for impact compaction.
4. Collapse and swelling could also be eliminated by compacting the soil at

1265

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 any water content to a relative compaction equal to some critical value. The
critical value of relative compaction increased with the magnitude of the vertical
overburden stress at which the soil was wetted.
5. The maximum amount of wetting-induced collapse occurred when the soil
was allowed to take in water under an overburden stress equal to the compactive
prestress.
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

6. The double-odometer test was sufficiently accurate to evaluate collapse po-

tential for this soil but did not give good results for evaluating swelling potential.
The method of compaction used to prepare test specimens does not appear to be
a critical parameter in collapse studies.

The applicability of these results and conclusions to other soils is not en-
tirely clear. It can be stated with reasonable certainty that other soils pos-
sessing similar qualities will likely behave in a similar manner. Because a
wide variety of compacted soils ranging from sands to clays have been de-
scribed in the literature as collapsing under certain conditions, the authors
believe the results and conclusions presented in this paper regarding collapse
are applicable in some degree to all compacted soils, although some minor
modifications are needed to apply some of them to nonexpansive soils. Those
results and conclusions discussed in terms of swelling, of course, apply only
to expansive soils. Further study regarding the types of compacted soils most
susceptible to collapse is definitely needed; research in this area is currently
being conducted by the second author.

APPENDIX I. REFERENCES

Barden, L., Madedor, A. O., and Sides, G. F. (1969). "Volume change character-
istics of unsaturated clay." J. Soil Mech. and Found. Div., ASCE, 95(1), 33-51.
Barden, L., McGown, A., and Collins, K. (1973). "The collapse mechanism in
partly saturated soil." Engrg. Geol., 7, 49-60.
Booth, A. R. (1977). "Collapse settlement in compacted soils." CSIR Research Re-
port 324, NITRR Bulletin 13, Pretoria, South Africa.
Cox, D. W. (1978). "Volume change of compacted clay fill." Proc. of Conference
on Clay Fills, Institution of Civil Engineers, London, Nov., 79-87.
Hilf, J. W. (1956). "An investigation of pore-water pressure in compacted cohesive
soil." Technical Memorandum 654, U.S. Department of Interior, Bureau of Re-
clamation, Denver, Colo.
Holtz, W. G. (1948). "The determination of limits for the control of placement mois-
ture in high rolled earth dams." Proc, ASTM, Philadelphia, Pa., 1240-1248.
Justo, J. L., Delgado, A., and Ruiz, J. (1984). "The influence of stress-path in the
collapse-swelling of soils at the laboratory." Proc. of the Fifth International Con-
ference on Expansive Soils, Adelaide, South Australia, May, 67-71.
Lawton, E. C. (1986). "Wetting-induced collapse in compacted soil," thesis pre-
sented to Washington State University, at Pullman, Wash., in partial fulfillment
of the requirements for the degree of Doctor of Philosophy.
Mishu, L. P. (1963). "Collapse in one-dimensional compression of compacted clay
on wetting," thesis presented to Purdue University, at West Lafayette, Ind., in
partial fulfillment of the requirements for the degree of Master of Science.
Mitchell, J. K. (1976). "Fabric, structure, and property relationships." Fundamentals
of Soils Behavior, John Wiley & Sons, New York, N.Y., 222-252.
Seed, H. B., and Chan, C. K. (1959). "Structure and strength characteristics of
compacted clays." J. Soil Mech. and Found. Div., ASCE, Oct., 87-128.
Witsman, G. R., and Lovell, C. W. (1979). "The effect of compaction prestress on
compacted shale compressibility." Report No. FHWA/IN/JHRP-79/16, Purdue
Univ., West Lafayette, Ind., Sept.

1266

J. Geotech. Engrg., 1989, 115(9): 1252-1267

 APPENDIX II. NOTATION

The following symbols are used in this paper:

R.C. =relative compaction based on maximum dry density determined
according to ASTM D 1557;
Downloaded from ascelibrary.org by Universiti Teknologi Mara on 09/01/24. Copyright ASCE. For personal use only; all rights reserved.

S = degree of saturation;
w = molding (compaction) water content; and
^voi = volumetric strain.

i,
ii
•k

1267

J. Geotech. Engrg., 1989, 115(9): 1252-1267