Mitchell 1986

Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.
THE TWENTIETH TERZAGHI LECTURE
Presented at the American Society of Civil Engineers

1984 Annual Convention
October 4, 1984
JAMES K. MITCHELL
255
J. Geotech. Engrg. 1986.112:255-289.

INTRODUCTION TO THE TWENTIETH
TERZAGHI LECTURE
By Robert D. Darragh
James K. Mitchell admirably fulfills the selection criterion for Terzaghi

lecturers: continuing contributions to the technical and professional stat-
ure of geotechnical engineering. He has had a productive and creative
career spanning over three decades, most of which were spent at the
Univ. of California, Berkeley. Jim is best known for his research, profes-
sional practice and publications related to soil behavior, which is the
subject of this Terzaghi lecture.
An easterner by birth (New Hampshire) and education (Bachelor of
Civil Engineering from Rensselaer Polytechnical Institute in 1951 and
Master of Science and Doctor of Science degrees from Massachusetts
Institute of Technology in 1953 and 1956), Jim started his westward mi-
gration in 1955 by working for the US Army Corps of Engineers' Water-
ways Experiment Station in Vicksburg, MS. Following service as an of-
ficer in the Corps of Engineers, Mitchell was recruited by H. Bolton Seed
in 1958 to join the faculty at Univ. of California, Berkeley. He has been
a Professor of Civil Engineering and Research Engineer since 1968. His
other academic duties have included Chairmanship of the Geotechnical,
Transportation, Construction and Surveying Division from 1974 to 1977
and Chairmanship of the Civil Engineering Department from 1979 to
1985. At Berkeley, Jim has developed and taught courses in soil behav-
ior, soil and site improvement and foundation engineering. His research
activities include studies in soil mechanics, soil improvement and rein-
forcement materials and in-situ measurement of soil properties. In ad-
dition to his long list of technical publications, which numbered 175 in
October 1984, Mitchell is the author of the graduate level textbook,
"Fundamentals of Soil Behavior."
The Geotechnical Engineering Division of ASCE has also benefited
greatly from Jim Mitchell's talents. He served six years on the Executive
Committee of the GT Division from 1966 to 1971 and was Chairman of
the Technical Committee on Placement and Improvement of Soils during
some of its most productive years, 1974 to 1978.
ASCE has recognized Mitchell's contributions by a long string of pres-
tigious awards including the Society's highest award in 1972, the Nor-
man Medal, for a paper on lunar soil research, co-authored with other
investigators on the Apollo space program. Jim was awarded the Soci-
ety's Walter L. Huber Prize in 1965 and was selected three times by the
GT Division for the Thomas A. Middlebrooks Award for papers pub-
lished in the Division Journal (as the Sole Author in 1962 and Senior
Author in 1970 and 1973). Other honors include the NASA Medal for
Exceptional Achievement (1973), the American Society for Engineering
Education Western Electric Fund Award (1979-1980) and election to the
US National Academy of Engineering in 1976.
It is worth noting that Jim Mitchell has known personally all nineteen
of his distinguished predecessors as Terzaghi lecturers, four of whom
were in attendance when this lecture was given: H. Bolton Seed, the
256
J. Geotech. Engrg. 1986.112:255-289.

1967 lecturer and Jim's long time colleague at Berkeley; T. William Lambe
who was Mitchell's adviser at MIT and whom Jim introduced as the 1970
lecturer; Lymon C. Reese, the 1976 lecturer who received his Ph.D. De-
gree at Berkeley; and Ronald F. Scott, last year's lecturer, fellow grad-
uate student with Jim at MIT and co-worker on the Apollo lunar soil
research program.
As further evidence of the relative youth of the geotechnical engi-

neering profession, all twenty Terzaghi lecturers were personally known
by Karl Terzaghi either as students of his or as professional colleagues.
None, however, had as unique an encounter with Terzaghi as did Jim
Mitchell early in his engineering career. In 1953, Jim attended his first
of many International Conferences on Soil Mechanics and Foundation
Engineering. During a post conference tour on a lake in Switzerland,
Mitchell tripped Terzaghi on the boat deck. To the good fortune of the
geotechnical engineering profession, it was not necessary to conduct man
overboard drills for either the father of soil mechanics or a most prom-
ising graduate student.
257
J. Geotech. Engrg. 1986.112:255-289.

PRACTICAL PROBLEMS FROM SURPRISING
SOIL BEHAVIOR
By James K. Mitchell, 1 F. ASCE
(The Twentieth Karl Terzaghi Lecture)

ABSTRACT: In spite of our present very advanced analytical capabilities, our

present ability to predict actual field behavior is, in many cases, woefully in-
adequate. Four cases are described to illustrate one of the reasons why actual
and predicted performances are sometimes so widely divergent—namely, fail-
ure to understand how a soil might respond over time to changed conditions.
The four cases are: (1) The aging of quick clay after sampling, in which the
remolded strength increases in samples maintained at constant water content;
(2) time effects in freshly densified or deposited sand, in which natural sand
deposits can lose strength if disturbed but regain strength over time periods of
weeks to months; (3) apparently sound lime-stabilized soil that swells and dis-
integrates starting a few years after construction; and (4) the failure of excess
pore pressures to dissipate as predicted during the consolidation of soft clays.
Study of each of these problems has led to an understanding of the responsible
phenomena and to technical conclusions useful in geotechnical practice. Col-
lectively, they illustrate that soils are not inert materials, but that they can, and
often do, change with time and changes in environmental conditions. Soil me-
chanics, in either conventional or advanced forms, cannot account for three of
the four cases described. All conceivable physicochemical and biological changes
must be considered in any case to explain the unusual behavior. It is essential
to expect the unexpected and to deal with soils as they are, not as we might
wish them to be.
INTRODUCTION
Many of the complexities and challenges of geotechnical engineering

are a result of nature's nearly limitless range of earth materials, the states
in which they may be encountered, and the myriad consequences that
may accompany changes in the ambient conditions. In the 21 years since
Karl Terzaghi's death and the inauguration of this lecture series in his
honor, there have been exceptional advances in our ability to investigate
soil deposits in the field, to measure their properties both in the field
and in the laboratory, and to solve increasingly complex problems by
analytical and numerical means. Many will agree that our ability to ana-
lyze and compute has n o w outstripped our ability to recognize, to mea-
sure, and to understand the relevant properties and parameters for many
problems. Unfortunately, this means that in spite of our very advanced
analytical capabilities, our ability to predict actual field behavior is, in
many cases, woefully inadequate.
One could list a number of reasons w h y actual a n d predicted perfor-
mance are sometimes so widely divergent. It is m y objective in this lec-
ture to illustrate, by means of actual field cases, one of them: namely,
failure to understand h o w a soil might respond over time to changed
'Prof, of Civ. Engrg., Univ. of California, Berkeley, CA 94720.
Note.—Discussion open until August 1, 1986. 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
October 4, 1984. This paper is part of the Journal of Geotechnical Engineering,
Vol. 112, No. 3, March, 1986. ©ASCE, ISSN 0733-9410/86/00u3-u259/$01.00. Pa-
per No. 20428.
259
J. Geotech. Engrg. 1986.112:255-289.

conditions. These cases include: (1) Remolded strength increases in sam-
ples of quick clay maintained at constant water content; (2) sand de-
posits that are weakened as a result of densification; (3) apparently sound
lime-stabilized soil that swells and disintegrates a few years after con-
struction; and (4) the failure of excess pore pressures to dissipate during
the consolidation of soft clays. In each case the actual behavior could be
considered surprising in that it was neither expected nor understood

initially. Each teaches technical lessons useful for analysis of similar
problems in the future. Collectively, they demonstrate that soils cannot
be treated as inert engineering materials with fixed properties. Instead,
careful consideration must be given to all details of their geologic history
and present environment and to the potential consequences of chemical,
biological, and physical changes brought about by our geotechnical ac-
tivities.
AGING OF QUICK CLAYS
The Problem.—The characteristics of quick clays are very well known

and clearly illustrated by Figs. 1 and 2. The great loss in strength that
accompanies disturbance and remolding (Fig. 1) has been the cause of
numerous dramatic and often catastrophic landslides (Fig. 2). It was, in
fact, the unusual behavior of these extremely sensitive clays, perhaps
more than anything else, that motivated many important pioneering
studies of soil structure and physico-chemical phenomena in fine-grained
soils—e.g. studies of clay structure by Casagrande (1932), undisturbed
clays and sensitivity by Terzaghi (1941), thixotropy by Skempton and
Northey (1952), the clay-water electrolyte system by Rosenqvist (1953,
1959), salt leaching and sensitivity by Bjerrum (1954), and Lambe's de-
FIG. 1.—Undisturbed and Remolded Quick Clay (Photograph Courtesy of Haley

and Aldrich, Inc.)
260
J. Geotech. Engrg. 1986.112:255-289.

FIG. 2.—Landslide in Quick Clay (Photographs Courtesy of J. M. Duncan)
velopment of a comprehensive hypothesis for the structure of soil (1953),

among others.
It has been known for some time that significant changes in the prop-
erties of the quick clays of Canada and Scandinavia often develop with
time after sampling. These changes are manifested by increases in re-
molded strength and liquid limit and decreases in the liquidity index
with time without changes in moisture content. Fig. 3, from Lessard
(1978), illustrates the changes that occurred in a remolded quick clay
from Outardes-2, Quebec over a one year period. Note that these changes
in strength and consistency developed in spite of the clay's remaining
at constant water content.
Such changes, known to occur generally in quick clays after sampling,
clearly are of significant geotechnical importance, because laboratory tests
on aged samples can give misleading results relative to the clay prop-
erties in situ. The remolded strength of an aged sample may be several
times greater than that of the natural material, thus giving false impres-
sions concerning the sensitivity and the potential for progressive failures
and flow slides.
The Explanation.—Various aspects of aging have been described—e.g.,
Bjerrum and Rosenqvist (1956), Soderblom (1969, 1974), Mourn et al.
(1971), and Torrance (1976). These studies established that during aging
both the pore water salinity and the percentage of divalent cations in-
261
J. Geotech. Engrg. 1986.112:255-289.

i i ' i i i
L
1.2 p. -
£
*«, - • -
LIQUIDIT - • •• *
p
0.6
* * ;
i i i i i
,
WATER CONTENT-
I PLASTIC LIMITS
1 1 i 1 i i
.
- •••*,' • •*• -
• • .* *
RANGE AT • • -
- t =0
. • •
¥ ,
I , i i i 1
FIG. 3.—Changes in the Remolded Strength and Consistency of a Canadian Quick

Clay as a Function of Time (Lessard, 1978)
crease. All of the changes in physical properties are quite consistent with
these compositional changes (Mitchell, 1976). What has been lacking un-
til very recently has been a suitable explanation for why these compo-
sitional changes occur. The recent dissertation research of Lessard (1981)
at the Univ. of California, Berkeley provides such an explanation.
The quick clay which formed the basis for Lessard's study was from
LaBaie near the Saguenay River in Quebec. It was deposited in the
LaFlamme Sea, an extension of the Champlain Sea, about 10,000 years
ago following the last glaciation. The sampling site is about 20 km from
Saint-Jean Vianney, where a slide involving 20 km2 of land area and an
estimated volume of 200 million m3 occurred 400 to 500 years ago. A
flowslide occurred in the same clay in 1971 with a loss of 31 lives and
destruction of 40 homes (Tavenas et al., 1971).
The LaBaie clay is a banded, silty clay that is medium to stiff in its
undisturbed state. Samples were obtained using 76 mm diameter Shelby
tubes. The undisturbed strength, as determined by fall cone tests (Hansbo,
1957) varies from 50 to 70 kPa (1,000 to 1,400 psf). The remolded strength
was too low to be reliably measured by the fall cone test; it is estimated
to be less than 0.07 kPa (15 psf). The sensitivity of freshly sampled clay
is greater than 500.
Some geotechnical characteristics of the clay and their variation with
depth are given in Fig. 4. The values shown were determined within
one month after sampling using material that had been carefully ex-
truded and coated with several layers of wax immediately after sam-
262
J. Geotech. Engrg. 1986.112:255-289.

WATER CONTENT-% LIQUIDITY INDEX PARTICLE SIZES - % SHEAR STRENGTH-kPo
„Q 20 40 60 3 4 5 6 7 0 20 40 60 80 100 0 20 40 60 80 OO
FIG. 4.—Geotechnical Characteristics of the Quick Clay from LaBaie as a Func-

tion of Depth (after Lessard, 1981)
TABLE 1.—Mineralogical Composition of LaBaie Quick Clay

Mineral Percent by Weight
(1) (2)
Primary Minerals
Plagiodase 45
K-feldspar 12
Quartz 13
Amphibole 10
Calrite 8
Clay Minerals
Illite 10
Chlorite <1
Kaolinite <1
pling. The low values of liquid limit and plasticity index and the very
high values of liquidity index are characteristic of quick clays. Mineral-
ogical analyses yielded the results in Table 1. The amount of clay min-
erals is quite small, which means that much of the clay size fraction is
composed of rock flour.
The results of pore water chemistry analyses are given in Fig. 5. So-
dium and bicarbonate are the dominant ionic species. The concentra-
tions of adsorbed calcium and magnesium are low, which is character-
istic of quick clays. The chloride concentration is some 1,800 times less
than that in sea water, which indicates the clay has undergone extensive
leaching. The relatively high bicarbonate concentrations and the high
pH are characteristic of reducing environments and the consequences of
sulfate reduction in clays. Blackish bands through the light-to-dark blue-
gray clay are probably related to the presence of metastable iron sulfide.
The experimental program for the study of aging effects in this clay
involved sample storage under a variety of conditions and determination
263
J. Geotech. Engrg. 1986.112:255-289.

ORGANIC CONTENT-% CARBONATE-% pH
0 0.1 0.6 0 4 8 9.0 9.4 9.8
(o) ORGANIC CONTENT, CARBONATE CONTENT, AND pH
DISSOLVED IONS - m e q / l i l e r ADSORBED CATIONS-meq/lOOgm

0.5 0 4 6 12 0 4 61 G
(b) IONIC CONCENTRATIONS IN PORE WATER
FIG. 5.—Compositional Data on the Quick Clay from LaBaie as a Function of Depth:
(a) Organic Content, CaC03 Content, and pH (after Lessard, 1981); (b) Ionic Con-
centrations in the Pore Water
of geotechnical and compositional properties at intervals extending over

a period of one year. Detailed descriptions of the procedures and com-
plete test results are presented by Lessard (1981) and Lessard and Mitch-
ell (1985).
All samples had been coated with several layers of paraffin at the time
they were taken prior to storage in a wet room at 20° C and 100% relative
humidity. A few samples were kept in Shelby tubes with thick, wax e'nd
plugs. After two months some Samples were removed and specially stdred
in tight plastic containers containing air, nitrogen, or oxygen atmo-
spheres. Additional samples were maintained at 4° C.
The variation with time of the remolded strength, liquidity index, and
several compositional characteristics is shown in Fig. 6 for waxed and
unwaxed samples stored under both temperature conditions. It may be
seen that unwaxed samples at both 20° C and 4° C underwent significant
changes. Waxed samples stored at 4° C showed little effect of aging. Also,
a Shelby tube sample with 2 to 3 cm thick, wax end plugs stored at 20° C
did not exhibit significant aging effects.
The results confirm that the aging process leads to increases in pore
water salinity and concentration of divalent cations. In addition, the de-
264
J. Geotech. Engrg. 1986.112:255-289.

0 5 10 !5
TIME AFTER SAMPLING-MONTHS TIME AFTER SAMPLING - MONTHS
FIG. 6.—Effect of Time and Storage Temperature on Properties of LaBaie Quick

Clay
1 1 1
SAMPLE STORAGE
CONDITIONS
WAXED: -
o 20°C, AIR
• 4°C,AIR o O
» 20 °C, AIR, STERILIZED 0 O
UNWAXEO: o 9 *
i.o|_ O 20°C, AIR, 0 2 , N 2 ?,<•> -
£ • 4°C,AIR °*>3 "o
• 0.5 o
W>]+h?+]%
o °°
S CATIONS
*9i of
0
a e
&9 Q3Q93
0.05 1 1 1
0.1 0.5 1.0 5 10
CONCENTRATION-meq/1
FIG. 7.—Dependence of Remolded Strength on Cation Concentrations in LaBaie

Clay
crease in pH with time (Fig. 6) is significant. Collectively, these changes

are responsible for the increase in remolded strength and decrease in
liquidity index, because each serves to depress the double layer sur-
rounding the clay particles, which leads to a decrease in interparticle
repulsive forces. Fig. 7 shows that remolded strength correlates well with
both the concentration of divalent cations and the total cation concen-
265
J. Geotech. Engrg. 1986.112:255-289.

tration. The method of storage does not effect these correlations; rather
it influences the time required for concentration changes to occur.
The reasons these compositional changes develop can be understood
by considering chemical and biological transformations that occur during
quick-clay formation, then reverse during aging. Collectively, the pro-
cesses form a part of the sulfur cycle.
During deposition, organic matter from marine organisms is deposited

simultaneously with the illite, feldspar, and quartz particles that con-
stitute the bulk of a quick clay. Iron oxide minerals are also present in
small quantities. As the depth of burial increases with continued de-
position, so does the distance to an oxygen supply from the sea water
above.
Oxidation of the organic matter diminishes the oxygen content of the
pore water, and an anaerobic environment develops which favors the
reduction of other species. This environment causes reduction of ferric
oxides to soluble ferrous iron by organic matter decomposition. Simul-
taneously, sulfates in the pore water are reduced to hydrogen sulfide by
the organic matter with the aid of sulfate-reducing bacteria. Then follows
the formation of iron sulfide materials:
Fe2+ + H2S -» black amorphous FeS ^ > FeS2 (pyrite)

The amount of FeS and FeS2 that can be produced is limited by the rate
of diffusion of sulfate from the overlying sea water and/or by the amount
and reactivity of detrital iron. These transformations can occur in a pe-
riod of only several years.
Simultaneously with the sulfate and ferric oxide reduction, carbon
dioxide is produced by bacterial oxidation of organic matter. The result
is an increase in alkalinity (pH increase) and decrease in the amount of
dissolved Ca2+ and Mg 2+ , as the latter precipitates as Mg-calcite.
If the potentially-quick deposit is uplifted above sea level, sulfate be-
comes scarce, oxidation of organic matter is slow because of the depleted
0 2 content, and sulfides remain stable. Fresh-water leaching decreases
the salt content, which in combination with the low Ca2+ and Mg2+ con-
centrations that result from the sulfate reduction processes, provides the
necessary conditions for the existence of a quick clay—i.e., low salt con-
tent, high percentage of monovalent cations in the adsorbed layers on
the clay particles, and high pH.
When a quick clay is sampled or exposed, some contact with the air
and oxygen is inevitable. In the presence of this oxygen some of the
remaining organic matter is oxidized to form carbonic acid which, in
turn, dissolves calcium carbonate, thus increasing the concentrations of
calcium and bicarbonate in the pore water. Even extremely low partial
pressures of 0 2 are sufficient to initiate oxidation phases of the sulfur
cycle. The oxidation of pyrite forms sulfuric acid and ferric hydroxide.
The reaction can be rapid at high pH. Slow transformation of Fe (OH)3
to yellow geothite (FeO-OH) may give a brownish color to the clay.
The sulfuric acid reacts with the Mg-calcite to increase the concentra-
tions of Mg2+ and Ca2+ in the pore water and in the adsorbed complex
on the clay particles. Sodium and potassium are displaced from the dou-
ble layer to the pore water. The salinity increase and the increase in
266
J. Geotech. Engrg. 1986.112:255-289.

divalent cations are responsible for increases in the remolded strength
and the liquid limit and decreases in the sensitivity and liquidity indices.
A complete description of these reactions, including phase diagrams and
reaction kinetics, is presented by Lessard (1981).
Although the formation of quick clays and the changes in their geo-
technical properties during aging can be explained by the processes de-
scribed above, amorphous materials, especially iron oxides, may also play
a part. When precipitated at particle contacts, these materials can act as
a cementing agent. They would be expected to increase the strength and
brittleness of the undisturbed clay both in situ and during sample aging.
The generation of additional iron oxides and hydroxides during the ox-
idation processes associated with quick clay aging could also lead to higher
undisturbed strengths in good quality samples than in situ. Amorphous
materials should not have a major effect on the strength of freshly re-
molded samples, however, because their contribution would be lost dur-
ing the remolding process.
Implications for Practice.—The aging of quick clays is an excellent ex-
ample of how even seemingly small changes in environmental condi-
tions can result in significant changes in properties. These changes can
occur over time periods typical of those associated with the field and
laboratory testing phases of many projects—e.g., several weeks to a few
months. If extreme care is not exercised during sample storage, labo-
ratory tests may give misleading results. Simple pH measurements at
the time of sampling, and then again at the time of testing, can provide
a rapid and easy means for assessing whether aging processes have oc-
curred. To minimize aging effects it is recommended that:
1. Exposure of samples to air be minimized.
2. Thick wax caps be used with rust-free Shelby tube samples.
3. Samples be stored at low temperatures to slow down the reaction
rates.
Recognition and clarification of the aging phenomenon leads also to
the possibility that the processes described above could be responsible
for some of the time-dependent strength gain in remolded marine clays
that has in the past been attributed to thixotropic hardening (Skempton
and Northey, 1952; Mitchell, 1960).
Finally, the possibility for quick clay "stabilization" by air injection
suggests itself. If the in situ clay could be partly exposed to oxygen,
perhaps by diffusion, then the associated reduction in sensitivity could
lessen the potential for catastrophic flow slides. This is because a higher
remolded strength would lessen the severity of progressive failure fol-
lowing a local disturbance.
AGING OF SANDS
Whereas the properties of clays have long been known to depend on

physico-chemical factors of various types, it perhaps comes as more of
a surprise to find that clean sands may also exhibit phenomena akin to
sensitivity and time-dependent strength gains after disturbance or after
initial deposition.
267
J. Geotech. Engrg. 1986.112:255-289.

FIG. 8.—Cross Section of Jebba Dam and Foundation Zones Densified by Vibro-
compaction and Blasting
UPSTREAM V.
COFFERDAM j ^ i * V-..„_ LEGEND:
FIG. 9.—Plan View of Jebba Dam
The Problem.—The recently completed Jebba Hydroelectric Devel-

opment on the Niger River, Nigeria, includes a 42 m high-zoned rockfill
dam founded on up to 70 m of river alluvium. Elevation and plan views
are shown in Figs. 8 and 9, respectively. The riverbed alluvium com-
prises clean, fine-to-coarse sand with an overall mean grain size of about
1.0 mm and an average uniformity coefficient of 2.94.
As the relative densities of the sands varied appreciably in both the
vertical and horizontal directions, the loose-to-medium dense zones had
to be densified to maintain settlements within acceptable limits and to
withstand the design earthquake forces. Vibrocompaction was used for
densification of the upper 25 m. Below this level blasting was used to
depths of 40 m. No precedent could be found for the use of blasting for
densification at such great depth. The overall foundation treatment pro-
gram is described by Solymar et al. (1984), and a comprehensive de-
scription and analysis of the blasting program is given by Solymar (1984).
268
J. Geotech. Engrg. 1986.112:255-289.

STATIC CONE RESISTANCE (MPa)
0 10 20 30 40 50 0 10 20 30 40
BEFOHE BLASTING 2 DAYS AFTER

BLASTING
FIG. 10.—CPT Records before and Shortly after a Blasting Densification Test
It came as a great surprise during an initial test program to learn that

the cone penetration resistance shortly after blasting was less than be-
fore, even though the evidence indicated that there had been positive
densification. Typical cone penetration test (CPT) records before and
shortly after a blast test are shown in Fig. 10. The regular curved lines
indicate values of cone resistance corresponding to different values of
relative density that were developed specifically for the sand at this site
using the method of Villet and Mitchell (1981). Results such as these
made it difficult to convince the client that a continuation of the test
program was in the best interests of the project, let alone continuation
of blasting through the proposed production phase!
However, subsequent CPT tests suggested that substantial increases
in penetration resistance had developed over a period of several months.
This is shown in Fig. 11. From results such as these, it was concluded
that the penetration resistance of the densified zones probably would
increase sufficiently over time to satisfy the design criteria. Accordingly,
the proposed blasting program was carried out.
Ground surface settlements totaling from 0.3 to 1.1 m were measured
immediately following the two of three blast coverages used in any area
in the production program, yet in most instances there was a significant
initial decrease in penetration resistance. Subsequent surface settlements
were negligible. Invariably significant increases in penetration resistance
were measured over several months following the blasting in zones that
were initially loose. An example is shown in Fig. 12. Additional exam-
269
J. Geotech. Engrg. 1986.112:255-289.

STATIC CONE RESISTANCE (HPa}
0 10 20 30 40 50 0 10 20 30 40 0 10 20 30 40 SO
BEFORE BLASTING 2 DAYS AFTER 9 9 DAYS AFTER

BLASTING BLASTING
FIG. 11,—CPT Records illustrating Penetration Resistance Increase with Time af-
ter Blasting Densification of Clean Sand—Test Program
pies are presented by Solymar (1984) and Mitchell and Solymar (1984).
The latter show additional data that indicate time-dependent increases
in penetration resistance following sand densification by dynamic com-
paction and vibrocampaction as well.
The surprising soil behavior, then, is that in-situ densification of sat-
urated clean sands can cause a reduction in penetration resistance, but
that with time the penetration resistance can increase to values greater
than before the densification. The problem is to explain this behavior,
to recognize conditions where it can occur, and to allow for it in an
appropriate manner in the geotechnical engineering of projects.
Explanation.—Calculations supported by water level measurements in
observation wells established that excess pore water pressures generated
by the sand disturbance could not have been the cause, as these pres-
sures were fully dissipated in a matter of hours. The observed behavior
can be explained, however, in terms of a very simple hypothesis. Sands
when initially deposited or immediately after in situ densification re-
spond as young, unbonded materials. Within weeks or months inter-
particle bonding and structure develops, leading to a "sensitivity." Sub-
sequent disturbance of the sand causes a structure breakdown and loss
of strength.
A special test done at the Jebba project excellently illustrates this sen-
sitivity and loss of strength. The rig for facilitating placement of charges
270
J. Geotech. Engrg. 1986.112:255-289.

STATIC CONE RESISTANCE (MPa)
0 10 20 30 40 50 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
PRE -BORING):
*'
1 \\ o
i-
EPTH FOR VIBROCOMP

It
UJ
n
<
< o>
CC
o
(IJPT O a
o U)
X
SEC OND
K
V
1
-+J \
i
tr 9
1
*>50> 60%
™*
BEFORE BLASTING 2 DAYS AFTER 3 0 DAYS A F T E R 57 DAYS AFTER 157 DAYS AFTER
2ND B L A S T 3 " 0 BLAST 3R° BLAST 3"° BLAST
COVERAGE COVERAGE COVERAGE COVERAGE
FIG. 12.—CPT Records Illustrating Penetration Increase with Time after Blasting
Densification of Clean Sand—Production Blasting
STATIC CONE RESISTANCE TOTAL

(MPa) FORCE (kN)
FT m 1.64 FT^ u ^BLASTHOLF.
10 20 30
+ +* 1.64 FT
(
c !\ IT
CPT A
+ .+-
1
CPT fl-**
1
> r CPTC^
X
H CPT D-*.
0.
UJ 4 0 -
TEST LAYOUT
\
{
i
{
*
J-
FIG. 13.—Results of Special Test to Investigate the "Sensitivity" of Sand Deposit

to Disturbance
271
J. Geotech. Engrg. 1986.112:255-289.

for the deep-blasting program utilized a long, 150-mm OD casing which
was advanced to the desired depths by a vibratory pile hammer. In Fig.
13 the curve labeled "before" in the right hand plot shows the down-
ward force required to push the casing into the undisturbed sand. The
curve labeled "after" shows the force required to repenetrate the pipe
at the same location. The resistance to penetration of the casing was
greatly reduced at depths greater than 7 m as a result of the disturbance.

The reduction in penetration resistance was much less above this depth,
because the hydraulic fill had been in place only a few weeks. The CPT
test results at point A, also shown in Fig. 13, show that the zone of
disturbance extended away from the point where the pipe was driven.
A reduction in penetration resistance was also measured at point B, which
is 1.5 m away horizontally from the blast tube insertion point.
That time-dependent strength gains can follow disturbance and den-
sification is shown by the data in Figs. 11 and 12. Test results that il-
lustrate time-dependent strength gain in freshly deposited sand are pre-
sented in Fig. 14. A lO-m-thick, hydraulic sand fill was placed along the
right bank of the river. The results of a number of CPT tests done 50 to
80 days after fill deposition are compared with those from tests done 4
to 10 days after placement. The penetration resistance approximately
doubled between the measurement periods.
A substantial amount of additional evidence, both laboratory and field,
is summarized by Mitchell and Solymar (1984) to illustrate time-depen-
dent increases in stiffness, strength, penetration resistance, and resis-
tance to cyclic loading in clean sands following deposition, placement,
or densification. Interestingly, observations similar to those described
here have been reported in the Russian literature for about 30 years
1 STATIC CONE RESISTANCE (MPa)
0 408012O0 4O8O120 0 4080120 0 4080120 0 40 80120 0 40 80120
f:
© 4-10 DAYS AFTER PLACEMENT (B) 50-80 DAYS AFTER PLACEMENT

RIGHT BANK SAND PAD DEPTH OF DREDGED SAND - 10m
FIG. 14.—-Penetration Resistance Profiles for a Hydraulic Fill at Two Times after
Placement
272
J. Geotech. Engrg. 1986.112:255-289.

DRV BULK DENSITY ( lb/ft3)
90 95 100 105 110 115
20 | 1 1 1 1 1 Hi
DRY BULK DENSITY ( gm/cm3)
FIG. 15.—Penetration Resistance of Naturally Deposited and Artificially Sedimen-

tal Sand (Durante and Voronkevich, 1955)
(Durante and Voronkevich, 1955). One of their results, showing a lower

penetration resistance for a freshly redeposited alluvial sand compacted
to the same density as the natural, undisturbed deposit, is shown in
Fig. 15.
Dudler et al. (1968) observed that disturbance without densification
can result in a strength loss. They also report a drop in penetration re-
sistance in some zones as a result of vibrocompaction at the Aswan High
Dam. Both changes in sand fabric and rupture of cementing bonds be-
tween particles could be responsible for these effects. Dudler and Iulin
(1981) report an increase in the penetration resistance of a sand fill at
the Riga Hydroelectric Station by a factor of 2.4 over a period of 80 months.
The precise mechanisms responsible for the sand structure and time
effects discussed above are not known. Possibilities are analyzed by
Mitchell and Solymar (1984). The most probable cause of the observed
time-dependent strength gain phenomenon seems to involve the for-
mation of silicic acid gel films on particle surfaces and the precipitation
of silica or other material from solution or suspension as a cement at
particle contacts. The solubilities of quartz and amorphous silica and their
solubility rates are high enough to be consistent with the observed rates
of strength increase.
Practical Application.—Structure sensitivity of sand deposits and time-
dependent strength gain in freshly deposited or disturbed sands may
affect the stress-deformation and strength properties enough to warrant
careful consideration in geotechnical engineering practice. Among the
important considerations are the following:
1. The strength and modulus values of reconstituted samples may be

significantly lower than those of the sand in situ.
2. In situ densification of sands may lead initially to reduced pene-
tration resistance.
273
J. Geotech. Engrg. 1986.112:255-289.

3. Substantial increases in penetration resistance with time may follow
placement of hydraulic fills or in situ deep densification of natural sand
deposits. Quality control measurements of densification or assessment
of liquefaction potential in freshly placed or densified sands made at
early times could be conservative.
4. The apparent failure of in-situ deep densification to improve the
properties of some sands may in some cases have been caused by the
densification being inadequate to compensate for strength loss due to
structure breakdown.
5. Test programs for the evaluation of deep-densification effectiveness
should be designed so as to enable assessment of time effects.
DELAYED FAILURE OF LIME-STABILIZED PAVEMENT BASES
The Problem.—Lime treatment of expansive silty clay soils was used

for subbase construction for an approximately three-mile (5-km) section
of major arterial street in Las Vegas, Nevada. The use of lime treatment
for base stabilization was relatively new in this area. Insofar as is known
it had been used previously for only two similar projects—another street
section and a school parking lot—and performance in the year or two
following their construction had been good.
Laboratory compaction and "R-value" tests had shown that a 4% by
weight treatment of the expansive soil with granular quicklime resulted
in significant improvement in the soil strength and stability. The stabi-
lometer test R-value is designed to evaluate the resistance of compacted
soils to deformations after a period of soaking, and it provides one of
0 UNTREATED ' '

© 4 % CaO .
® © _
©
• m
m -
G • G
O
_
0
o -
°8 °
o
o
o
o
o
-
o
8 o -
i 7 °i S ?° ,
10 20 30 40
MOISTURE CONTENT-%
FIG. 16.—Effect of Quicklime Treatment on the Stability (R-Value) of Expansive

Silty Clay
274
J. Geotech. Engrg. 1986.112:255-289.

7/// i / / ; ' > ' > s '/ 'i i / i i ;//
V 7 4 " ASPHALT CONCRETE;//
5" OR 8" AGGREGATE BASE

12" LIME TREATED SUBBASE

(4% CaO + EXPANSIVE SILTY CLAY)
FIG. 17.—Pavement Section Containing Lime-Treated Subbase
the parameters needed for flexible pavement design using the State of
California method (California Department of Transportation, 1978, 1981).
R-values as a function of compaction moisture content for both treated
and untreated soil are shown in Fig. 16.
The results of these tests and traffic information were used to establish
the pavement-thickness design indicated in Fig. 17. The street width was
90 ft, curb-to-curb, and consisted of four, 12-ft-wide travel lanes, two
parking lanes, and a center divider/left lane median strip. Construction
was done during the spring of 1975. Construction specifications were
typical for this type of work. The only significant difficulty during con-
struction was that the subgrade was wet and soft in some localized areas,
making it difficult or impossible in some cases to achieve the specified
compaction. It was concluded from quality control tests and field in-
spections, however, that overall the lime-treated material was placed and
compacted in accordance with the engineers' recommendations.
FIG. 18.—Pavement Surface Failures
275
J. Geotech. Engrg. 1986.112:255-289.

FIG. 19.—Appearance of Intact, Unfailed Pavement Sections
FIG. 20.—Failure of Parking Lot Pavement Containing a Lime-Treated Base
The completed construction appeared of good quality and initial per-

formance was excellent. However, beginning in the fall of 1977 signs of
distress began to appear in the form of surface heaving and cracking.
By the spring of 1978 the distress in some areas had become major, with
heaves amounting to several inches in places. Some examples are shown
in Fig. 18. These zones were concentrated mainly in the western half of
the project. Some portions of the road did not exhibit distress, and in
these areas the pavement is excellent in appearance and condition, as
may be seen in Fig. 19.
Several investigations have been made to determine the nature and
causes of this unanticipated failure which developed suddenly more than
two years after the initial construction. Note also that similar types of
failure developed subsequently in the two prior lime stabilization proj-
ects. The school parking lot as it appeared in the spring of 1982 is shown
in Fig. 20.
Explanation.—Post-mortem investigations of the failed pavement have
shown rather conclusively that:
1. The failure was not traffic-induced, and the structural design of the
pavement section was adequate.
2. Construction was done generally in accordance with the specifi-
cations.
3. The heave developed in the lime-treated soil. In the failed zones
the soil has a much lower density and higher water content than either
the untreated soil or the intact, treated soil. Some comparisons are given
in Table 2.
276
J. Geotech. Engrg. 1986.112:255-289.

TABLE 2.—Characteristics of Untreated and Lime-Treated Soil Samples Taken in
March 1S82
Water Dry unit Liquid Plasticity
Type of content weight limit index
soil Sample Depth (ft) (%) (Ib/cu ft) (%) (%) PH
(1) (2) (3) (4) (5) (6) (7) (8)
Untreated la 2-3 15.5 114.6 36.3 25.6 8.5

soila 2a 1-1-1/2 21.4 104.7 41.1 26.0 8.2
3a 1/2-1 18.1 104.2 64.8 . 43.5 8.4
4a 4-5 14.5 94.3 39.9 21.1 8.1
5a 1-2 10.0 92.8 31.9 13.4 8.7
Treated soil lb 2-2-1/2 36.5 77.5 59.6 20.8 9.5
from failed 2b 1-2 46.6 71.4 61.6 16.8 9.5
zonesb 3b 1-2 55-85 63.3 64.6 6.9 10.5
Treated soil lc 1-1-1/2 16.9 99.5 NP 9.8
from intact, 2c 1-1-1/2 89.0 54.5 15.9 10.4
unfailed 3c 1-1-1/2 25.4 93.7 47.6 12.5 9.7
zonesb 4c 1-2 21.2 95.6 57.8 30.2 9.1
5c 1-1/2-2-1/2 15.1 98.5 33.6 12.9 8.8
"Samples taken adjacent to pavement.
'Samples taken from beneath pavement surface.
Was the unusual behavior of the lime-treated soil due to unique char-
acteristics of the soil, the lime, the compaction water, the construction
procedures, or the curing? Interest centered on the soils, because no
significant abnormalities could be found in the other factors.
Fine-grained soils in the desert southwest of the United States con-
trast with soils in other areas in that they often contain significant amounts
of sulfate and carbonates. Sodium sulfate, Na 2 S0 4 , and gypsum, CaS0 4
• 2H 2 0, are the common sulfate forms, and calcium carbonate, CaC0 3 ,
and dolomite, MgC0 3 , are the usual carbonate forms. The dominant
clay minerals in these soils are expansive smectites.
The question to be addressed, then, was: can there be a series of events
such that a lime-treated soil of this type will be strong and intact just
after compaction but will, under certain conditions, deteriorate at some
later time? Tests were done to investigate the soil composition and
chemistry and the reaction products, and how they were related to the
observed behavior.
The results of these tests provided the following information:
1. The untreated soil contained significant amounts of soluble sodium

sulfate (up to 1.5 percent by weight).
2. The untreated soil contained large amounts of gypsum and calcite
and some dolomite [Ca-Mg(C03)2].
3. Lime was present at the end of construction in both zones that sub-
sequently failed and in zones that have remained intact.
4. The lime-treated soils in the failed zones had swelled to water con-
tents higher than, and densities lower than, those for the untreated soil
also exposed to water.
5. Samples taken from intact and unfailed zones can swell if exposed
to water.
277
J. Geotech. Engrg. 1986.112:255-289.

6. Expansive clay minerals were present in the untreated soil.
7. Calcium silicate hydrates (CSH), the cementing material which re-
sults from the reaction between lime and clay minerals in successful lime
stabilization, could not be found in the samples from the failed and un-
failed zones. The pH must be 12.4 or greater for these reactions to occur.
Since the pH of the treated zones was only 8 to 10.5, it is unlikely that
future cementation reactions will occur.

8. Significant amounts of ettringite, Ca6[Al(OH6)]2 (S0 4 ) 3 • 26H 2 0, and
thaumasite, Ca6[Si(OH)6]2 (S0 4 ) 2 (C0 3 ) 2 • 24H 2 0, were indicated by X-rdy
diffraction in both failed and unfailed zones along sections of the street
where the heaving failures predominated. These materials are known to
be very expansive, and their formation is responsible for the deteriora-
tion of concrete by sulfate attack.
9. Negligible amounts of ettringite and thaumasite were found in treated
samples from the eastern half of the project where failures had not oc-
curred.
These findings taken together lead to the following hypothesis for the
series of events leading to the unexpected Swelling and pavement fail-
ure.
When quicklime and soil are initially mixed in the presence of water,
the quicklime hydrates according to
CaO + H z O -> Ca(OH)2 + heat
Some calcium goes into solution and exchanges with sodium on the clay,
and along with light cementation by carbonate and gypsum, suppresses
the swelling tendencies of the expansive clay minerals when the treated
soil is exposed to water at short times after compaction. The result is a
compacted soil that exhibits negligible swell and a high R-value after
soaking. Some of the strength increase as compared to the untreated soil
may also be attributed to the reduction in water content caused by the
hydration of CaO.
In the presence of sodium sulfate, however, the available lime is de-
pleted according to
Ca(OH)2 + Na 2 S0 4 -» CaS0 4 + 2NaOH
If the pH rises to 12.4, then silica (Si02) and alumina (A1203) will dissolve
from the clay, or they may be present in amorphous form initially. These
compounds can then combine with the calcium, carbonate, and sulfate
present in the soil to form ettringite arid thaumasite.
This will further deplete the available lime, leading to a drop in pH,
which will, in turn, prevent the formation of cement (CSH), since ad-
ditional Si0 2 can not be dissolved from the clay. The expansive char-
acteristics of the clay minerals can then reassert themselves. Of greater
importance, however, is the fact that ettringite and thaumasite are very
expansive materials (Mehta and Hu, 1978).
If then the treated material is given access to water, a large amount
of swell may occur. This swell may exceed that exhibited by the un-
treated soil. In the case of the Las Vegas street, the water probably gained
access through joints between the asphalt concrete pavement and the
278
J. Geotech. Engrg. 1986.112:255-289.

portland-cement concrete gutter and curb, and between the asphalt con-
crete and the raised portland-cement concrete median.
Since these analyses were completed, a pavement failure at a parking
lot in Wichita, KS has occurred that followed essentially the same be-
havior. Ettringite was found in the swollen soil.
Lessons for Future Projects.—These experiences with delayed failure
of lime-treated expansive soils teach several important lessons.

1. If sulfates are present in a soil, then the use of lime or portland
cement as admixture stabilizers should be approached with great cau-
tion, and, in many cases, may have to be avoided altogether.
Prior to field experience in the lime treatment of sulfate-bearing soils
has been very limited. There are no cautions in any of the literature that
is most widely consulted for information and advice on lime treatment.
However, from some experiences in southern California, it has been
suggested that lime should not be used if there is more than 5,000 ppm
soluble sulfate in the soil. Also, the results of a study by Sherwood (1962)
unfortunately never received the attention that they should have. His
work, done in the United Kingdom, showed that water immersion sub-
sequent to compaction of lime and cemented-treated sulfate-bearing soils
could lead to expansion and cracking. He also suggested the possibility
of the formation of ettringite as a result of reaction between the sulfate
and the clay.
2. Satisfactory test results shortly after lime or cement treatment do
not necessarily guarantee satisfactory performance after longer time pe-
riods.
3. Local chemical and mineralogical factors can have important influ-
ences on soil properties and behavior.
4. It is important to be alert for unique soil compositions and to try
to anticipate what might possibly happen as a result of interactions be-
tween materials in situations where there is no prior experience.
ANOMALOUS PORE PRESSURES AFTER LOADING AND DURING

CONSOLIDATION OF SOFT CLAYS
Each of the three cases described so far concerned changes in prop-

erties and behavior that were unanticipated. In the final problem, changes
that are expected do not occur.
The Problem.—Terzaghi's development of a theory for clay consoli-
dation sixty years ago is generally considered to mark the beginning of
modern soil mechanics. Success in predicting settlement magnitudes and
rates for different field cases has ranged from excellent to poor. Many
reasons have been advanced to explain those cases where agreement has
been poor. More sophisticated theories and constitutive models have been
developed to improve predictions.
In recent reviews, Crooks et al. (1984) and Becker et al. (1984) sum-
marized approximately 50 field cases of the consolidation of soft clay.
They found that in a significant number, the observed behavior was not
as predicted. The anomalous observations included:
1. Differences in predicted and observed initial pore pressure re-
sponse.
279
J. Geotech. Engrg. 1986.112:255-289.

E
(_ o
ro
_J
30 m
8
6 Brown Gray Organic Ctay, Upper 0.5m Dry Crust

4- Postglacial Green-Black Clay
E with Black Sulfide Blotches
. 2 Specks and Organic Colloids, Many Shells
c
.2 OH
a Postglacial Gray Ctay, Very Uniform in Color
> -2 with Occasional Subangular Gravels in the Size of 4mm
UJ - 4
Glacial Varved Clay
-6- Thin Layer of Medium Gray Sand
_£. k A
-8
FIG. 21.—Soil Profile and Test Fill at Vasby, Sweden (Mesri and Choi, 1985)
2. Continued porewater pressure build-up after completion of load-

ing. This phenomenon was observed in 11 of 31 cases for which data
were available.
3. Differences between field consolidation rates and those predicted
based on laboratory test results.
4. Changes in pore pressure dissipation rates during and following
construction.
5. Apparent lack of strength gain with consolidation following load
application.
Much of the anomalous behavior can be explained qualitatively through
considering such factors as the location and effectiveness of drainage
layers, differences between assumed and actual stress paths, the non-
linearity of the pressure versus void ratio relationship, variation of per-
meability with changes in void ratio, and propper evaluation of the stress
history and current stress state.
However, an aspect of the consolidation behavior in some cases that
has not yet been fully explained is the failure of pore pressures to dis-
sipate at a rate commensurate with the settlement and corresponding
reductions in void ratio and water content. Possibly, the best known
case of this type is the test fill at Vasby, Sweden. Interestingly, this case
was a direct result of Terzaghi's proposal in 1946 for a long-term field
test at this site to investigate the long-term consolidation and secondary
compression of soft organic clays (Chang et al., 1973). A 2.5-m-thick gravel
fill was built over a 1-m-thick clay layer as shown in Fig. 21.
Piezometric data indicated that significant, excess pore water pres-
sures persisted in the middle of the clay layer for a long period after the
application of loading while settlements continued. In 1968, 21 years af-
ter construction, the magnitude of the pore pressure near the mid-plane
of the clay layer was still nearly equal to the applied stress increment at
that time, after adjustment for submergence during settlement. This ex-
ceptionally high pore pressure is shown in Fig. 22, where it is compared
with predictions by the classical Terzaghi theory using the data pre-
sented by Chang (1981).
280
J. Geotech. Engrg. 1986.112:255-289.

0.1
TIME FACTOR
1.0 10
TIME -YEARS
FIG. 22.—Vasby Test Fill Pore Pressures Compared with Terzaghi Theory Predic-
tions
The curve labelled "Degree of Consolidation by Terzaghi Theory" in

Fig. 22 gives the average pore pressure versus time relationship for the
full layer that is predicted by classical theory. The curves labelled ut/u0
represent the theoretical excess pore pressure ratios versus time for points
where Z/H is 0.5 and 1.0, Z being the depth, and H the maximum length
of drainage path. As the clay layer at Vasby is drained at both, top and
bottom, H is half the total layer thickness. Measured values of excess
pore pressure ratio at a point where Z/H =0.91 are shown based both
on the original fill load applied in 1947 and on the total stress increment
1
:"»^m^
1.0
t-0.8
SAND AND STIFF CLAY
1976 1977 1976

101 II 1 121 1 2 | 3 | 4 1 5 16171 8 19 1101 | l12 | 1 2 | 3 | 4 | 5 | 6 | 7 1 B 19
H
-
• ^
, . 1 ,
1
r
• . ^ •
°\J •
xxaxmaxam coo e o n aaaxtn axm t
1 "*"' » « n o o „'
1 ED rfroocccmc
ajt^ooo
""•teKWOBaao. M ^ t a n <: " " W W t t k w oca,
i l oL2?
'bxxxpo c 1 ED
'•WW** 1
S | |
FIG. 23.—Settlement and Pore Pressure Observations for the Penang, Malaysia
Test Fill (from Adachi and Todo, 1979)
281
J. Geotech. Engrg. 1986.112:255-289.

that remained in 1968 after correction for submergence due to settle-
ment.
Other field cases have been reported that indicate settlement contin-
uing at nearly constant values of excess pore water pressure, sometimes
referred to as pore pressure stagnation, over significant periods of time—
e.g., Crawford and Burn (1976), Mesri and Choi (1979), Holtz and Broms
(1972), Choa et al. (1981). At the New Liskeard, Ontario highway em-
bankment, pore pressures continued to increase for up to four months
after all fill placement, and there was little pore pressure dissipation over
the next 2-1/2 years, although surface settlements of up to 2 feet were
measured (Stermac et al., 1967). Under a test fill in Penang, Malaysia,
a plot of settlement versus time suggested the rate of consolidation set-
tlement has become very low even though excess pore water pressures
of 50 to 60 percent of their initial value remained (Adachi and Todo,
1979). Fig. 23 illustrates this case and shows settlements continuing at
essentially constant pore water pressure.
The surprising aspect of soil behavior to be examined and explained,
if possible, is continuing consolidation and settlement without corre-
sponding dissipation of pore pressures. The associated changes in shear
strength are also of interest.
Hypothesis.—Of the four problems discussed in this lecture, this is
the only one that may be amenable to quantification using soil mechan-
ics theory. There is general agreement among the investigators who have
described anomalous pore pressure behavior during clay consolidation
that it arises because of structural breakdown during compression. Soft,
sensitive clays are characterized by void-ratio versus log-pressure curves
of the type shown in Fig. 24. A review of cases shows that invariably
the clays are slightly overconsolidated, with OCR values typically in the
range of 1 to 2. When such clays are subjected to stress, only a small
deformation is required before the state corresponding to u'p, the effec-
tive preconsolidation pressure, is reached. Any small increase in effec-
tive stress above a'v leads to a very great increase in compressibility as
a result of a tendency for the structure to collapse. This can be accom-
panied by a large decrease in permeability.
The increase in compressibility and decrease in permeability lead im-
mediately to an abrupt and large decrease in the coefficient of consoli-
dation. Thus the rate of consolidation becomes greatly slowed. The im-
portance of changes in compressibility and permeability on clay
consolidation was discussed by Leonards and Altschaeffl (1964).
More recently Mesri and Choi (1985) proposed a method to take these
effects into account. A multi-layer analysis that includes variation of con-
solidation parameters with depth and during the consolidation process,
is carried out. Their reanalysis of the Vasby case agreed with measured
and predicted settlements, as seen in Fig. 25.
Predicted pore pressure distributions at Vasby by Mesri and Choi (1985)
agree in form with measured values, as may be seen in Fig. 26. This
lends additional support to the importance of taking compressibility and
permeability variations into account. Nonetheless, the measured values
of pore pressure are still greater than the predicted values. Mesri and
Choi suggested that instrumentation problems and interpretation un-
certainties may account for the differences. Large strain and self-weight
282
J. Geotech. Engrg. 1986.112:255-289.

0
Observed
Predicted by
Mesri and Choi
(1985)
Vasby, Sweden
1947
i IIIIIII > '' I

10 100 1000 10,000
LOG VERTICAL EFFECTIVE S T R E S S - a ' T i m e , Days
FIG. 24.—Characteristics of the Com- FIG. 25.—Computed and Observed

pression Curve of Soft Clay Settlements, Vasby Test Fill (Mesri and
Choi, 1985)
Observed
Predicted by
Mesriond Chpi(l985)
10 20 30 40
Excess Pore Water Pressure, kPa
FIG. 26.—Observed Excess Pore Pressures and Pressures Predicted by the Method
of Mesri and Choi (1985)
effects (Schiffman et al., 1984) may also retard the rate of pore pressure
dissipation.
It is likely, however, that continued structural readjustment of the clay
during consolidation is a major factor. Structure breakdown when ef-
283
J. Geotech. Engrg. 1986.112:255-289.

T 1 r—
TIME OF
TIME llhouBand mln|
FIG. 27.—Observed and Predicted Pore Pressure versus Time after Consolidation
under Isotropic Stress for Undisturbed San Francisco Bay Mud (Bonaparte, 1981)
fective stresses pass the preconsolidation pressure, or "destructuration,"

can generate excess pore pressures which add to the partially dissipated
excess pore pressure induced by initial loading.
This effect is considered in the model for soft clay behavior developed
by Kavazanjian and Mitchell (1980) and improved by Bonaparte (1981).
It provides a means for evaluating immediate and time-dependent vol-
umetric and deviatoric deformations. It has also been demonstrated to
be useful for predicting pore pressure variations with time for a variety
of triaxial and plane strain loading conditions.
One of the cases studied by Bonaparte (1981) is relevant to the anom-
alously high pore pressure phenomenon considered here. Data have been
presented by Kaldveer (1964), Arulanandan et al. (1971) and Holzer et
al. (1973) to demonstrate that if consolidation is stopped while a clay is
undergoing secondary compression, then pore pressure will develop as
function of time. They attributed this pore pressure generation to the
inability of the clay structure to come to complete equilibrium before the
drainage was terminated. According to the theory, the magnitude and
generation rate of the pore pressure depend on the prior period of con-
solidation and the ratio of the coefficient of secondary compression to
the recompression index.
Fig. 27 shows measured pore pressures as a function of time after the
cessation of drainage for three samples of undisturbed San Francisco Bay
mud. The mud had been previously allowed to consolidate for different
times into secondary compression under an isotropic stress. In each case
no excess pore pressure was detected at the time the drainage valve was
closed. The predicted variation of pore pressure with time is given by
the curves. The rather close agreement is encouraging and suggests that
continued research using such an approach may lead to suitable pre-
dictive procedures.
Considerations for Engineering Practice.—Failure to anticipate and
understand the pore pressure generation and stagnation phenomena can
lead to several problems in engineering practice. Among them are:
284
J. Geotech. Engrg. 1986.112:255-289.

1. Inability to reconcile piezometer and settlement data.
2. Uncertainty about the reliability of piezometer data.
3. Incorrect predictions of pore pressure dissipation rates.
4. Inability to predict shear strength increases during consolidation.
The latter can be of particular concern when knowledge of strength is

needed for control of fill placement rates and stability evaluation. The
rates of shear strength increase at sites where anomalous pore pressure

conditions have been observed seem not yet to have been studied in
detail.
CONCUJSIQNS
Each of the situations described in the preceeding pages arose as a

surprise, and each presented problems of geotechnical importance. From
their study it has been possible to learn about the aging of quick clays,
about time effects in freshly deposited or densified sands, about delayed
deterioration of lime-treated soils as a result of sulfate reactions, and
about anomalous pore pressures in consolidating soft clays. The specific
technical conclusions are listed at the end of each section and will not
be restated here.
There are other conclusions, less specific in a technical sense, but per-
haps more important from the standpoint of future success in geotech-
nical engineering practice:
1. Soils are not inert materials. They can, and often do, change with
time and changes in conditions.
2. What soils are arid how they behave depends, as does the behavior
of people, on both genetic and environmental factors. Hence, a thor-
ough understanding and appreciation for the geologic history and for-
mational processes of a soil deposit are essential. Anticipation of future
performance must be based on consideration of all conceivable physical,
chemical, and biological changes. Regrettably, we don't know enough
about the latter, and additional, carefully designed and executed re-
search is needed.
3. None of the first three cases could have been anticipated or ex-
plained through application of either conventional or advanced soil me-
chanics theory or sophisticated mathematical analysis methods. Fur-
thermore, the use of soil mechanics theory and numerical methods for
analyses and prediction using properties obtained from tests on aged
samples in the case of quick clays, or immediately after placement or
densification in the case of some sands could give grossly misleading
results.
4. The foregoing statements are not intended to imply that soil me-
chanics does not have an important role in geotechnical engineering. For
example, quantification of the anomalous pore pressure phenomena de-
scribed in the preceding section should be possible using existing prin-
ciples of soil mechanics.
5. We need to learn better how to expect the unexpected, especially
when confronted with new problems in new environments or settings.
Terzaghi had a remarkable ability to change his mind and revise his
285
J. Geotech. Engrg. 1986.112:255-289.

opinions as n e w evidence became available. H e appreciated fully both
the usfulness and limitations of theory, and most of all, h e dealt with
soils and rocks as they are, not as h e wished t h e m to be. Many engineers
seem to disregard this important concept in present-day research inves-
tigations.
Finally, it is appropriate to return to the title of this lecture, "Practical

Problems from Surprising Soil Behavior." Was the behavior really sur-
prising, or in reality w a s it w e w h o were surprised? It seems likely, in
my view, that h a d Karl Terzaghi, with his life-long interest in the p h y s -
ical and chemical aspects of real soil behavior, encountered similar prob-
lems, h e would have relentlessly p u r s u e d investigations of their causes
until suitable explanations could be found. Given nature's nearly lim-
itless range of earth materials a n d mankind's continuing n e w uses for
them, we can ill afford not to do likewise in similar situations.
ACKNOWLEDGMENTS
The writer is indebted to m a n y teachers, colleagues, students, a n d

friends w h o have generously given time, knowledge, and encourage-
ment to his continuing education in geotechnical engineering. Without
the insights and understanding that have come from these individuals
over the years, m y analyses of the cases described herein would not
have been possible.
I thank J. M. Duncan for finding the right words for me to express
the concept of this lecture in its title. H. B. Seed, C. L. Monismith, G.
A. Leonards, G. Lessard, a n d W. C. B. Villet m a d e careful reviews of
the manuscript a n d offered m a n y helpful comments and suggestions.
E. C. Tse assisted in the preparation of the section on anomalous pore
pressure behavior.
Drafting of the figures was d o n e by G. Pelatowski, and preparation
of the manuscript was d o n e by Marcia Golner, Nancy Hoes, a n d Liz De
Britto. Their assistance is acknowledged with thanks and appreciation.
APPENDIX.—BIBLIOGRAPHY
Adachi, K., and Todo, H. (1979). "A Case Study on Settlement of Soft Clay in
Penang," Proc. 6th Asian Regional Conference on Soil Mechanics and Foundation En-
gineering, Singapore, Vol. 1, pp. 117-120.
Arulanandan, K., Shen, C. K., and Young, R. B. (1971). "Undrained Creep Be-
havior of a Coastal Organic Silty Clay," Geotechnique, Vol. 21, No. 4, pp. 359-
Becker, D. E., Crooks, J. H. A., Jeffries, M. G., and McKenzie, K. (1984). "Yield
Behavior and Consolidation, II: Strength Gain," Proc, ASCE Symposium on Sedi-
mentation Consolidation Models, Prediction and Validation, pp. 382-398.
Bjerrum, L. (1954). "Geotechnical Properties of Norwegian Marine Clays," Geo-

technique, Vol. 4, pp. 49-64.
Bjerrum, L., and Rosenqvist, I. Th. (1956). "Some Experiments with Artificially
Sedimented Clays," Geotechnique, Vol. 6, pp. 124-136.
286
J. Geotech. Engrg. 1986.112:255-289.

Bonaparte, R. (1981). "A Time-Dependent Constitutive Model for Cohesive Soils,"
Ph.D. Dissertation, Dept. of Civ. Engrg., Univ. of California, Berkeley, 263 pp.
California Department of Transportation (1978). Transportation Laboratory Manual
of Test, Vol. 2, Test 301, R-Value Test.
California Department of Transportation (1981). Highway Design Manual, Section

7-651, Asphalt Concrete Pavement.
Casagrande, A. (1932). "The Structure of Clay and its Engineering Significance,"

Contributions to Soil Mechanics, Boston Society of Civil Engineers, 1925-1940, pp. 72-
112.
Chang, Y. C. E. (1981). "Long Term Consolidation Beneath the Test Fills at Vasby,
Sweden," Swedish Geotechnical Institute, Report No. 13, Linkoping, 148 pp.
Chang, Y. C. E., Broms, B., and Peck, R. B. (1973). "Relationship Between the
Settlement of Soft Clays and Excess Pore Pressure Due to Imposed Loads," Proc.
8th International Conference on Soil Mechanics and Foundation Engineering, Moscow,
Vol. 1, pp. 93-96.
Choa, V., Karunaratne, G. P., Ramaswamy, S. D., Vijiratnam, A., and Lee, S.
L. (1981). "Drain Performance in Changi Marine Clay," Proc. 10th International
Conference on Soil Mechanics and Foundation Engineering, Stockholm, Vol. 3, pp.
623-626.
Crawford, C. B., and Burn, K. N. (1976). "Long-Term Settlements on Sensitive

Clay," Laurits Bjerrum Memorial Volume, Norwegian Geotechnical Institute, Oslo,
pp. 117-124.
Crooks, J. H. A., Becker, D. E., Jeffries, M. G., and McKarzie, K. (1984). "The
Significance of Effective Stress Paths and Yield Behavior on the Field Consoli-
dation of Soft Clays," Proc. ASCE Symposium on Prediction and Case Histories of
Consolidation Performance, San Francisco, CA, Oct., 1984.
Dudler, I. V., Durante, V. A., and Smernov, C. D. (1968). "Experience Gained

in Using the Penetrometer Probe for Soil Investigation in Conjunction with En-
ergy-Related Construction in the Soviet Union," INFORM-ENERGO, Moscow, 63
pp.
Dudler, I. V., and Iulln, A. N. (1981). "Sand Structural Characteristics Revealed

During Density Measurements Using the Penetrometer Probe," Hydrogeology and
Engineering Geology—Field Methods of Investigation, Novocherkassk, pp. 25-35.
Durante, V. A., and Voronkevich, S. D. (1955). "Studies of Sand Densities for

Construction Purposes," Annals of Moscow University, No. 2, pp. 131-137.
Hansbo, S. (1957). "A New Approach to the Determination of the Shear Strength
of Clay by the Fall Cone Test," Royal Swedish Inst. Proc. No. 14.
Holzer, T. L., Hoeg, K., and Arulanandan, K. (1973). "Excess Pore Pressures
During Undrained Creep," Canadian Geotechnical Journal, Vol. 10, No. 1, pp. 12-
24.
Holtz, R. D., and Broms, B. (1972). "Long-Term Loading Tests at Ska Edeby,
Sweden," Proc. ASCE Specialty Conference on Performance of Earch and Earth-Sup-
ported Structures, Purdue Univ., Vol. 1, pp. 435-484.
Kaldveer, P. (1964). Unpublished Research Report, Univ. of California, Berkeley.
287
J. Geotech. Engrg. 1986.112:255-289.

Kavazanjian, E., and Mitchell, J. K. (1980). "Time-Dependent Deformation of
Behavior of Clays," Journal of the Geotechnical Engineering Division, ASCE, Vol. 106,
No. GT6, pp. 611-630.
Lambe, T. W. (1953). "The Structure of Inorganic Soil," ASCE Separate No. 315,
October.
Leonards, G. A., and Altschaeffl, A. G. (1964). "Compressibility of Clay," Journal

of the Soil Mechanics and Foundations Division, ASCE, Vol. 90, No. SM5, pp. 133-
155.
Lessard, G. (1978). "Traitement Chimique des Argiles Sensibles d'Outardes-2,"

Memorie de M.Sc.A., Ecole Polytechnique de Montreal, 187 pp.
Lessard, G. (1981). "Biogeochemical Phenomena in Quick Clays and Their Effects
on Engineering Properties," Ph.D. Dissertation, Dept. of Civ. Engrg., Univ. of
California, Berkeley, CA, 335 pp.
Lessard, G., and Mitchell, J. K. (1985). "The Causes and Effects of Aging in
Quick Clays," Canadian Geotechnical Journal, Vol. 22, No. 3.
Mehta, P. K., and Hu, F. (1978). "Further Evidence for Expansion of Ettringite
by Water Adsorption, Journal of the American Ceramic Society—Discussions and Notes,
Mar.-Apr., 1978, pp. 179-180.
Mesri, G., and Choi, Y. K. (1979). "Excess Pore Water Pressures During Con-
solidation," Proc. 6th Asian Conference on Soil Mechanics and Foundation Engineering,
Singapore, Vol. 1, pp. 151-154.
Mesri, G., and Choi, Y. K. (1985). "Settlement Analysis of Embankments on Soft

Clays," Journal of Geotechnical Engineering Division, ASCE, Vol. I l l , No. 4, pp.
441-464.
Mitchell, J. K. (1960). "Fundamental Aspects of Thixotropy in Soils," Journal of

the Soil Mechanics an Founations Division, ASCE, Vol. 86, No. SM3, pp. 19-52.
Mitchell, J. K. (1976). Fundamentals of Soil Behavior, John Wiley & Sons, New York,
NY, 422 p.
Mitchell, J. K., and Solymar, Z. V. (1984). "Time-Dependent Strength Gain in

Freshly Deposited or Densified Sand," Journal of the Geotechnical Engineering Di-
vision, ASCE, Vol. 110, No. GT11, pp. 1559-1576.
Mourn, J., Loken, T., and Torrance, J. K. (1971). "A Geotechnical Investigation
of the Sensitivity of a Normally Consolidated Clay from Drammen, Norway, Geo-
techniaue, Vol. 21, No. 4, pp. 329-340.
Rosenqvist, I.Th. (1953). "Considerations on the Sensitivity of Norwegian Quick

Clays," Geotechnique, Vol. 3, No. 5, pp. 195-200.
Rosenqvist, I.Th. (1959). "Physico-Chemical Properties of Soils: Soil-Water Sys-

tems," Journal of the Soil Mechanics and Foundations Division, ASCE, Vol. 85, No.
SM2, Part 1, pp. 31-53.
Schiffman, R. L., Pane, V., and Gibson, R. E. (1984). "The Theory of One-Di-
mensional Consolidation of Saturated Clays, IV. An Overview of Nonlinear Fi-
nite Strain Sedimentation and Consolidation," Proc, ASCE Symposium on Sedi-
mentation Consolidation Models, Prediction and Validation, pp. 1-29.
288
J. Geotech. Engrg. 1986.112:255-289.

Sherwood, P. T. (1962). "Effects of Sulfates on Cement and Lime-Stabilized Soils,"
Highway Research Board Bulletin, 353 pp. 98-107.
Skempton, A. W., and Northey, R. D. (1952). "The Sensitivity of Clays," Geo-
technique, Vol. 3, No. 1, pp. 30-53.
Soderblom, R. (1969). "Salt in Swedish Clays and its Importance for Quick-Clay
Formation: Results from Some Field and Laboratory Studies," Swedish Geotech-
nical Institute, Proceedings, 22, 63 pp.
Soderblom, R. (1974). "Aspects on Some Problems of Geotechnical Chemistry,

Part III," Swedish Geotechnical Institute, 55, pp. 452-468.
Solymar, Z. V. (1984). "Compaction of Alluvial Sands by Deep Blasting," Ca-

nadian Geotechnical Journal, Vol. 21, No. 2, pp. 305-321.
Solymar, Z. V., Iloabachie, B. C , Gupta, R. C , and Williams, L. R. (1984). "Earth

Foundation Treatment at Jebba Dam Site," Journal of Geotechnical Engineering, ASCE,
Vol. 110, No. GT10, pp. 1415-1430.
Stermac, A., Lo, K. Y., and Barsvary, A. K. (1967). "The Performance of an Em-
bankment on a Deep Deposit of Varved Clay," Canadian Geotechnical Journal, Vol.
4, No. 1, pp. 45-61.
Tavenas, F., Chagron, J.-Y., and LaRochelle, P. (1971). "The St.-Jean Vianney
Landslide: Observations and Eyewitness Accounts," Canadian Geotechnical Jour-
nal, Vol. 8, pp. 463-478.
Terzaghi, K. (1941). "Undisturbed Clay Samples and Undisturbed Clays," Journal
of the Boston Society of Civil Engineers, Vol. 28, No. 3, pp. 211-231.
Torrance, J. K. (1976). "Pore Water Extraction and the Effect of Sample Storage
on the Pore Water Chemistry of Leda Clay," American Society for Testing and Ma-
terials, Special Technical Publication, 599, pp. 147-157.
Villet, W. C. B., and Mitchell, J. K. (1981). "Cone Resistance, Relative Density
and Friction Angle," in Cone Penetration Testing and Experience, G. M. Norris and
R. D. Holtz, Ed., ASCE, pp. 178-208.
289
J. Geotech. Engrg. 1986.112:255-289.

Mitchell 1986

Uploaded by

Copyright:

Available Formats

Mitchell 1986

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mitchell 1986

Uploaded by

Copyright:

Available Formats

Downloaded from ascelibrary.org by UNIVERSITE LAVAL on 07/05/14. Copyright ASCE. For personal use only; all rights reserved.

THE TWENTIETH TERZAGHI LECTURE

Presented at the American Society of Civil Engineers

J. Geotech. Engrg. 1986.112:255-289.

James K. Mitchell admirably fulfills the selection criterion for Terzaghi

J. Geotech. Engrg. 1986.112:255-289.

As further evidence of the relative youth of the geotechnical engi-

J. Geotech. Engrg. 1986.112:255-289.

By James K. Mitchell, 1 F. ASCE

(The Twentieth Karl Terzaghi Lecture)

ABSTRACT: In spite of our present very advanced analytical capabilities, our

Many of the complexities and challenges of geotechnical engineering

J. Geotech. Engrg. 1986.112:255-289.

considered surprising in that it was neither expected nor understood

AGING OF QUICK CLAYS

The Problem.—The characteristics of quick clays are very well known

FIG. 1.—Undisturbed and Remolded Quick Clay (Photograph Courtesy of Haley

J. Geotech. Engrg. 1986.112:255-289.

FIG. 2.—Landslide in Quick Clay (Photographs Courtesy of J. M. Duncan)

velopment of a comprehensive hypothesis for the structure of soil (1953),

J. Geotech. Engrg. 1986.112:255-289.

FIG. 3.—Changes in the Remolded Strength and Consistency of a Canadian Quick

J. Geotech. Engrg. 1986.112:255-289.

FIG. 4.—Geotechnical Characteristics of the Quick Clay from LaBaie as a Func-

TABLE 1.—Mineralogical Composition of LaBaie Quick Clay

J. Geotech. Engrg. 1986.112:255-289.

(o) ORGANIC CONTENT, CARBONATE CONTENT, AND pH

DISSOLVED IONS - m e q / l i l e r ADSORBED CATIONS-meq/lOOgm

(b) IONIC CONCENTRATIONS IN PORE WATER

of geotechnical and compositional properties at intervals extending over

J. Geotech. Engrg. 1986.112:255-289.

FIG. 6.—Effect of Time and Storage Temperature on Properties of LaBaie Quick

FIG. 7.—Dependence of Remolded Strength on Cation Concentrations in LaBaie

crease in pH with time (Fig. 6) is significant. Collectively, these changes

J. Geotech. Engrg. 1986.112:255-289.

During deposition, organic matter from marine organisms is deposited

Fe2+ + H2S -» black amorphous FeS ^ > FeS2 (pyrite)

J. Geotech. Engrg. 1986.112:255-289.

Whereas the properties of clays have long been known to depend on

J. Geotech. Engrg. 1986.112:255-289.

FIG. 9.—Plan View of Jebba Dam

The Problem.—The recently completed Jebba Hydroelectric Devel-

J. Geotech. Engrg. 1986.112:255-289.

BEFOHE BLASTING 2 DAYS AFTER

It came as a great surprise during an initial test program to learn that

J. Geotech. Engrg. 1986.112:255-289.

BEFORE BLASTING 2 DAYS AFTER 9 9 DAYS AFTER

J. Geotech. Engrg. 1986.112:255-289.

EPTH FOR VIBROCOMP

STATIC CONE RESISTANCE TOTAL

FIG. 13.—Results of Special Test to Investigate the "Sensitivity" of Sand Deposit

J. Geotech. Engrg. 1986.112:255-289.

greatly reduced at depths greater than 7 m as a result of the disturbance.

1 STATIC CONE RESISTANCE (MPa)

0 408012O0 4O8O120 0 4080120 0 4080120 0 40 80120 0 40 80120

© 4-10 DAYS AFTER PLACEMENT (B) 50-80 DAYS AFTER PLACEMENT

J. Geotech. Engrg. 1986.112:255-289.

DRY BULK DENSITY ( gm/cm3)

FIG. 15.—Penetration Resistance of Naturally Deposited and Artificially Sedimen-