Professional Documents
Culture Documents
H Azir Babak 2005
H Azir Babak 2005
by
Kenan Hazırbaba
2005
The Dissertation Committee for Kenan Hazırbaba Certifies that this is the
approved version of the following dissertation:
Committee:
Kenneth H. Stokoe, II
Stephen G. Wright
Clark R. Wilson
Kevin J. Folliard
Pore Pressure Generation Characteristics of Sands and Silty
Sands: A Strain Approach
by
Dissertation
Presented to the Faculty of the Graduate School of
in Partial Fulfillment
of the Requirements
Doctor of Philosophy
I would like first to thank my advisor, Professor Ellen M. Rathje, for her
continuous support, valuable guidance, time, insight, energy, and unbounded
enthusiasm. It has been a great pleasure for me to work with her. Professors
Kenneth H. Stoke II, Stephen G. Wright, Clark R. Wilson, and Kevin J. Folliard,
are due many thanks for serving on my dissertation committee and providing
valuable comments.
I also would like to thank the other members of the geotechnical faculty:
Professors Roy E. Olson, Robert B. Gilbert, Jorge G. Zornberg, Charles Woodruff
Jr., and former faculty Alan F. Rauch, for their contribution to my education at
The University of Texas at Austin.
Many thanks to all my fellow graduate students at UT for their friendship,
help, and support. The complete list of these great people would be too long to
acknowledge here.
Last, but certainly not least, I would like to thank my family: My lovely
wife Yıldız for her patience and understanding, and for being always by my side
during this long process, my mom and dad for their immeasurable long distance
support and love, my brothers Yusuf and Kerim for making me smile every time I
talk to them, and my uncle Ergin for his continuous encouragement. You all mean
the world to me.
iv
Pore Pressure Generation Characteristics of Sands and Silty
Sands: A Strain Approach
Publication No._____________
v
Dobry et al. (1982). Multiple series of strain-controlled cyclic direct simple shear
and cyclic triaxial tests were used to directly measure the excess pore water
pressure generation of sands and silty sands at different strain levels. The soil
specimens were tested under three different categories: a) at a constant relative
density, b) at a constant sand skeleton void ratio, and c) at a constant overall void
ratio. The results from each of these groups were examined. In addition,
laboratory measured pore water pressures of clean sands were compared to in situ
measured values.
The findings from this study were used to develop insight into the
behavior of silty sands under undrained cyclic loading conditions. In general,
beneficial effects of the fines were observed in the form of a decrease in excess
pore water pressure and an increase in the threshold strain. However, pore water
pressure appears to increase when enough fines are present to create a sand
skeleton void ratio greater than the maximum void ratio of the clean sand. The
comparison between laboratory and in situ measurements indicated that larger
pore water pressure was generated in situ.
vi
Table of Contents
List of Tables.......................................................................................................... xi
vii
Chapter 4: CYCLIC STRAIN-CONTROLLED TESTING PROGRAM............ 67
4.1 Introduction ......................................................................................... 67
4.2 GCTS cyclic direct simple shear system............................................. 67
4.3 Saturation and consolidation ............................................................... 73
4.4 Testing procedure ................................................................................ 75
4.5 Soils tested........................................................................................... 78
4.6 Strain-controlled testing program........................................................ 83
4.7 Specimen preparation.......................................................................... 91
4.7.1 Dry pluviation ............................................................................ 91
4.7.2 Water sedimentation................................................................... 91
4.7.3 Moist undercompaction.............................................................. 92
4.5.4 Slurry deposition ........................................................................ 99
4.5.5 Remarks on specimen preparation ........................................... 105
4.8 Verification of the cyclic direct simple shear system ....................... 105
4.9 Summary ........................................................................................... 112
viii
6.3 Selection of the controlling index parameters................................... 152
6.4 Tests at constant relative density....................................................... 156
6.5 Tests at constant sand skeleton void ratio ......................................... 166
6.6 Tests at constant overall void ratio.................................................... 176
6.7 Pore water pressure generation rate of sands and silty sands............ 185
6.8 Discussion ......................................................................................... 189
6.9 Summary ........................................................................................... 193
ix
References ........................................................................................................... 222
x
List of Tables
Table 4-7 CDSS Tests on sands containing fines at constant Dr=50% ................ 87
Table 4-8 CDSS Tests on sands containing fines at constant es=0.76 .................. 88
Table 4-9 CDSS Tests on sands containing fines at constant e=0.59 ................... 88
undercompaction .............................................................................. 95
Table 4-13. Results of the attempts made to form specimens with 35% and
Table 4-14 Trials of the slurry method with the GCTS CDSS set-up................. 104
Table 4-15 Stress-controlled CDSS tests on Monterey #0/30 sand .................... 106
xi
List of Figures
Figure 2-1 Cross section of particulate group showing packing changes that
Figure 2-3 Rate of pore water pressure generation (De Alba et al. 1976) ............ 10
Figure 2-4 Pore water pressure generation versus shear strain for various
Figure 2-5 Cyclic stress approach for the evaluation of liquefaction potential
Figure 2-7 Effect of seismic history on cyclic strength of sand (Seed 1979) ....... 18
Figure 2-9 Measured pore water pressure in saturated sands after ten loading
Figure 2-10 Relationship between stress ratio causing liquefaction and (N1)60
values for silty sand for event of magnitude M=7.5 (Seed et al.
1985)................................................................................................. 23
xii
Figure 2-12 Cyclic resistance curves for sand and silty sand at constant overall
Figure 2-13 Cyclic resistance versus silt percent (Erten and Maher 1995a)......... 26
Figure 2-14 Effect of silt content on liquefaction of layered silty sand (Amini
and Qi 2000)..................................................................................... 26
Figure 2-16 Phase diagram for the calculation of sand skeleton void ratio .......... 28
Figure 2-18 Phase diagram for the calculation of limiting fines content .............. 30
Figure 2-19 Cyclic resistance of silty sand at various constant sand skeleton
Figure 2-20 Variation in cyclic resistance with silt content (Polito and Martin
2001)................................................................................................. 33
Figure 2-21 Cyclic resistance of silty sand at constant relative density (Singh
1994)................................................................................................. 34
Figure 2-22 Variation in cyclic resistance with silt content for Yatesville sand
Figure 2-23 Cyclic resistance versus relative density for Yatesville sand
xiii
Figure 2-25 Pore water pressure generation at N=10 and in sand and silty sand
Figure 3-1 Cyclic shear stresses beneath level ground during seismic loading
Figure 3-2 Cross section of a typical triaxial test apparatus (Ishihara 1996) ........ 44
Figure 3-4 Results from a stress-controlled cyclic triaxial test (Ishihara 1996) ... 46
Figure 3-6 Cyclic strength of sands in terms of cyclic stress ratio versus
Figure 3-7 Cross section of a typical hollow cylinder torsional shear apparatus
Figure 3-8 Stresses acting on elements of the wall of a hollow torsional shear
Figure 3-9 Results from a typical torsional shear test (Ishihara 1996) ................. 52
Figure 3-10 Cyclic loading in direct simple shear test (Seed and Peacock
1971)................................................................................................. 54
Figure 3-11 Typical results from a stress-controlled cyclic direct simple shear
Figure 3-12 Schematic of the SGI-type simple shear device (Kjellman 1951) .... 57
Figure 3-13 NGI-type simple shear apparatus (Bjerrum and Landva 1966)......... 58
xiv
Figure 3-15 Schematic of the BAW simple shear apparatus (Franke et al.
1979)................................................................................................. 60
Figure 3-16 Schematic of the bi-directional simple shear device (Ishihara and
Figure 3-17 Cross section of the UCB-2D simple shear apparatus (Boulanger
Figure 3-18 Constant volume cyclic direct simple shear test on dry sand
(Finn 1985)....................................................................................... 64
Figure 4-2 Location of sensors in GCTS cyclic direct simple shear apparatus .... 70
Figure 4-3 Set-up of roller base LVDT and relative shear deformation LVDT ... 72
Figure 4-4 Shear strain measurements from the relative LVDT and base
LVDT ............................................................................................... 73
Figure 4-7 Effect of staged testing on the generation of excess pore water
pressure............................................................................................. 78
Figure 4-10 Grain size distribution of Monterey #0/30 sand and Sil-Co-Sil 52
xv
Figure 4-12 Volume change during saturation...................................................... 96
Figure 4-14 Cyclic resistance versus relative density for Monterey sand
Figure 5-2 Excess pore water pressure generation as a function of cyclic shear
Figure 5-4 Excess pore water pressure generation rate per cycle for Monterey
Figure 5-6 Excess pore water pressure generation as a function of cyclic shear
xvi
Figure 5-7 Excess pore water pressure generation as a function of number of
Figure 5-8 Excess pore water pressure generation as a function of cyclic shear
Figure 5-15 Effect of relative density on initial excess pore water pressure
generation rate during the second cycle for Monterey #0/30 sand 132
generation rate during the 10th cycle for Monterey #0/30 sand...... 133
xvii
Figure 5-18 Excess pore water pressure measurements from cyclic triaxial
Figure 5-19 Pore water pressure generation curves obtained from cyclic
Figure 5-20 Pore water pressure generation from cyclic triaxial tests and
Figure 5-21 Grain size distribution and index properties of Monterey #0/30
Figure 5-22 State of stress for specimens under cyclic direct simple shear and
1975)............................................................................................... 140
Figure 5-23 Comparison of cyclic direct simple shear and cyclic triaxial
Figure 5-24 Pore water pressure generation curves from cyclic direct simple
Figure 5-25 Comparison of cyclic direct simple shear and cyclic triaxial
Figure 6-1 Theoretical variation of the minimum void ratio with fines content
Figure 6-2 Effect of fines content and fine particle diameters on minimum
Figure 6-3 Maximum and minimum void ratios versus fines content for
xviii
Figure 6-4 Sand skeleton void ratio versus overall void ratio for various fines
contents........................................................................................... 155
Figure 6-5 Pore water pressure histories at constant Dr=50% and γ=0.03% for
Figure 6-6 Pore water pressure histories at constant Dr=50% and γ=0.1% for
Figure 6-7 Pore water pressure histories at constant Dr=50% and γ=0.3% for
Figure 6-8 Pore water pressure generation curves at constant Dr=50% and
Figure 6-9 Pore water pressure generation curves at constant Dr=50% and
Figure 6-10 Excess pore water pressure ratio versus fines content at Dr=50%
Figure 6-11 Excess pore water pressure ratio versus fines content at Dr=50%
Figure 6-12 Excess pore water pressure ratio versus fines content at Dr=50%
Figure 6-13 Pore water pressure histories at constant es=0.76 and γ=0.03% for
various fines contents ..................................................................... 167
Figure 6-14 Pore water pressure histories at constant es=0.76 and γ=0.1% for
xix
Figure 6-15 Pore water pressure histories at constant es=0.76 and γ=0.3% for
various fines contents ..................................................................... 169
Figure 6-16 Pore water pressure generation curves at constant es=0.76 and
Figure 6-17 Pore water pressure generation curves at constant es=0.76 and
Figure 6-18 Excess pore water pressure ratio versus fines content at es=0.76
Figure 6-19 Excess pore water pressure ratio versus fines content at es=0.76
Figure 6-20 Excess pore water pressure ratio versus fines content at es=0.76
Figure 6-21 Pore water pressure histories at constant e=0.59 and γ=0.03% for
Figure 6-22 Pore water pressure histories at constant e=0.59 and γ=0.1% for
Figure 6-24 Pore water pressure generation curves at constant e=0.59 and N=1
Figure 6-25 Pore water pressure generation curves at constant e=0.59 and
xx
Figure 6-26 Excess pore water pressure ratio versus fines content at e=0.59
Figure 6-27 Excess pore water pressure ratio versus fines content at e=0.59
Figure 6-28 Excess pore water pressure ratio versus fines content at e=0.59
Figure 6-29 Rate of pore water pressure generation in sand and silty sand
Figure 6-30 Rate of pore water pressure generation in sand and silty sand
Figure 6-31 Rate of pore water pressure generation in sand and silty sand
Figure 6-32 Cyclic resistance values at Dr=30% (after Polito and Martin
2001)............................................................................................... 190
Figure 6-33 General evaluation of the effect of fines content on pore water
Figure 7-1 Schematic cross-section of the in situ testing set up (Rathje et al.
2005)............................................................................................... 197
Figure 7-2 In situ pore water pressure generation curves (Rathje et al. 2005) ... 198
Figure 7-3 Laboratory measurements of pore water pressure at σv'=100 kPa .... 200
Figure 7-4 Laboratory measurements from this study and Dobry et al. (1982).. 201
Figure 7-5 Pore water pressure generation curves for Monterey #0/30 and
xxi
Figure 7-6 Laboratory measurements at σv'=100 kPa along with in situ pore
water pressure measurements......................................................... 203
Figure 7-8 Effect of confining pressure on pore water pressure generation ....... 206
xxii
Chapter 1: INTRODUCTION
1
It is well established that the mechanism for the occurrence of liquefaction
under seismic loading conditions is the generation of excess pore water pressure,
and that generation of excess pore water pressure is controlled mainly by the level
of induced shear strains (Dobry et al. 1982). Additionally, pore water pressure
generation has a significant effect on the shear strength, stability, and settlement
characteristics of a soil deposit, even if it does not cause the soil to fully liquefy.
Thus, a more fundamental approach to study the effect of fines content on
liquefaction would be to examine the pore water pressure generation mechanism
directly through strain-controlled cyclic tests.
This research presents an effort to find a unified picture on the effect of
fines content on excess pore water pressure generation. Strain-controlled cyclic
direct simple shear tests are used to directly measure the excess pore water
pressure generation of sands and silty sands. An attempt is made to characterize
the measured pore water pressures. In addition, laboratory measured pore water
pressures are compared to values measured in situ.
2
undercompaction technique (Ladd 1978) was used to reconstitute the clean sand
specimens at three different relative densities (Dr = 32%, 50%, 93%). The tests
were performed at various shear strains ranging from 0.005% to 0.3%. Each
specimen was subjected to cycles of a constant shear strain level. Excess pore
water pressure generation was measured directly, and evaluated in terms of: (1)
number of loading cycles, (2) induced shear strains, and (3) pore water pressure
generation rate. The evaluation was made for each relative density group and all
testing series (Dr = 32%, 50%, 93%) were compared to investigate the effect of
relative density. During the first part of the research, a series of strain-controlled
cyclic triaxial tests was also carried out on clean sand specimens. The results from
this series were compared with those from the cyclic direct simple shear tests in
an effort to investigate the effect of test type on the generation of excess pore
water pressure.
The second part of the research was aimed at evaluating the effect of fines
content on the generation of excess pore water pressure. This part of the research
consisted of strain-controlled, undrained, cyclic direct simple shear tests on silty
sand specimens. Mixtures of Monterey #0/30 sand and Sil-Co-Sil 52 non-plastic
silt were used to reconstitute silty specimens with various fines contents up to
20%. The specimens were prepared at three different index parameters: (1) a
constant relative density of 50%, (2) a constant sand skeleton void ratio of 0.76,
and (3) a constant overall void ratio of 0.59. An evaluation of the fines effect on
the excess pore water pressure generation was made for each of these index
parameters. The change in excess pore water pressure with respect to number of
loading cycles and shear strains was examined and the trends in pore water
pressure with increasing fines content were evaluated in each case (e.g., constant
relative density, constant sand skeleton void ratio, constant overall void ratio).
3
The third part of the research was comprised of strain-controlled,
undrained, cyclic direct simple shear tests on a clean sand (aggregate sand)
obtained from a reconsituted soil deposit located in Austin. Rathje et al. (2005)
reported in situ pore water pressure measurements from a series of field
liquefaction tests on this soil deposit. The purpose of this part of the research was
to measure the excess pore water pressure generation in the same sand under
laboratory conditions, and make a comparison between in situ and laboratory
measured values. The specimens were prepared by the water sedimentation
technique to both mimic the natural soil deposition and match the specimen
preparation procedure of Rathje et al. (2005). A series of tests was performed at
shear strain levels ranging from 0.005% to 0.96% under two different initial
vertical effective stresses. The measured pore water pressures were evaluated in
light of the in situ measurements.
4
Chapter 2 presents a review of relevant background information. This
includes a definition of liquefaction, a discussion of the pore water pressure
generation mechanism, a description of the commonly used liquefaction
evaluation techniques, and a critical review of previous investigations on the
effect of fines on cyclic resistance and pore water pressure generation.
Chapter 3 summarizes the general categories of laboratory tests used for
liquefaction studies. Three commonly used test types, cyclic triaxial, cyclic
torsional, and cyclic direct simple shear, are discussed with an emphasis on the
advantages and disadvantages of each test.
Chapter 4 presents the details of the testing program. This includes a
description of the testing set-up used for this research, the index properties of the
soils tested, the testing procedures, and the different series of tests.
Chapter 5 presents and discusses the results of the tests performed on
clean sands. The pore water pressure generation in specimens prepared at three
different relative densities is evaluated. Also, results from cyclic direct simple
shear tests are compared with those from cyclic triaxial tests.
Chapter 6 presents the findings of the study on the effect of fines on
excess pore water pressure generation. This chapter provides a discussion of how
changing the fines content influences the excess pore water pressure generation
characteristics.
Chapter 7 examines excess pore water pressure generation in situ and
under laboratory conditions. A discussion of the in situ and laboratory measured
pore water pressures is presented.
Chapter 8 presents a summary of this research and the concluding remarks
of the research. Suggestions for future work in this area are also included in this
chapter.
5
Chapter 2: REVIEW OF LIQUEFACTION STUDIES
2.1 INTRODUCTION
In this chapter, first a brief discussion of the liquefaction phenomenon and
the relationship between liquefaction and pore water pressure generation is
presented. Next, the current approaches for the evaluation of liquefaction are
reviewed. The factors affecting cyclic resistance of soils are also discussed during
this review. Finally, previous investigations on the effect of fines on cyclic
resistance are presented.
6
flotation to the ground surface of buried concrete tanks (Dobry et al. 1982).
Therefore, liquefaction has become a prime subject of concern to both
geotechnical and structural engineers.
The generation of pore water pressure within a saturated soil beneath level
ground during an earthquake is generally assumed to be due to the cyclic shear
stresses and shear strains generated by upward propagation of shear waves,
although other forms of wave motions also exist in the soil deposit (Seed 1979).
Induced shear stresses and shear strains result in densification of the soil if it is in
a drained condition. Figure 2-1 shows schematically how cyclic densification
occurs during each loading cycle. If the soil is in an undrained condition, as cyclic
loading occurs the load from the overlying material is transferred from the soil
skeleton to the incompressible pore fluid, generating excess pore water pressure.
These excess pore water pressures accumulate and lead to a reduction in the
effective stresses in the soil. The mechanism of pore water pressure generation
during cyclic loading is illustrated in Figure 2-2. The tendency to densify and
decrease in volume (points A to B) is counteracted by a decrease in effective
stress (points A to C) for the undrained condition. The shear strength of
liquefiable soils is proportional to the effective stress acting on the interface
between particles. Hence a decrease in the effective stress due to excess pore
water pressure generation results in decreased shear strength and softening.
7
Figure 2-1 Cross section of particulate group showing packing changes that occur
during cyclic loading (Youd 1977)
8
Figure 2-2 Schematic illustration of pore water pressure generation during cyclic
loading (Seed 1979)
9
Various researchers (Lee and Albaisa 1974, De Alba et. al 1976, and Seed
et al. 1976) have shown that there is a strong relationship between the pore water
pressure ratio (u*= ru = ratio of the measured excess pore water pressure to the
initial effective stress, ∆u/σo') and the number of loading cycles. If the pore water
pressure ratio of a clean sand is plotted against the ratio of the cycle of loading to
the total number of cycles required to cause initial liquefaction, over a wide range
of densities the results will plot within a narrow band. Such a representation is
shown in Figure 2-3. De Alba et al. (1976) obtained the data in this figure from
large scale simple shear tests performed on Monterey #0 sand. The sand
specimens were prepared at relative densities varying between 54% and 90%.
The narrow band in Figure 2-3 for a large range in densities suggests that density
may affect the number of cycles to liquefaction (Nl), but it only has a minor effect
on the pore water pressure generation curve at any value of Ne/Nl, where Ne
denotes the number of cycles of stress applied.
Figure 2-3 Rate of pore water pressure generation (De Alba et al. 1976)
10
Dobry (1985) investigated pore water pressure generation as a function of
shear strain and number of loading cycles for sands representing a wide range of
relative densities (Figure 2-4). Dobry (1985) obtained these data from strain-
controlled, undrained, triaxial tests on seven different sands. Various specimen
preparation techniques were used to reconstitute the sand specimens at relative
densities ranging from 20% to 80%. The measured excess pore water pressures
indicated a relatively narrow band, as that shown in Figure 2-4. These data
indicate a threshold shear strain (~ 0.01%) below which excess pore water
pressure does not develop. At larger strains, significant pore water pressure
develops and at shear strains greater than 0.3% and 10 cycles of loading, initial
liquefaction is almost reached (u* > 0.8). The data in Figure 2-4 indicate that
shear strain and number of loading cycles are the main factors influencing pore
water pressure generation.
11
Figure 2-4 Pore water pressure generation versus shear strain for various sands
and densities (Dobry 1985)
12
2.3 EVALUATION OF LIQUEFACTION
There are a number of approaches available for the evaluation of soil
liquefaction and pore water pressure generation. Simplified procedures (e.g., Seed
and Idriss 1971, Dobry et al. 1982, Law et al. 1990, Kayen and Mitchell 1997) are
commonly used in engineering practice. There are two approaches for the
simplified procedures: (1) the cyclic stress approach and (2) the cyclic strain
approach. More complicated constitutive models (e.g., Finn et al. 1977, Prevost
1981, Zienkiewicz et al. 1990) have been developed to predict pore water pressure
generation, stiffness degradation, and settlements during seismic events.
13
Figure 2-5 Cyclic stress approach for the evaluation of liquefaction potential
(Seed et al. 1975a)
14
correlated with in situ tests such as the Standard Penetration Test (SPT). When
performing laboratory tests, the tests provide data regarding the shear stress level
that causes liquefaction in the number of expected earthquake loading cycles. This
is sometimes called the cyclic strength. When using in situ correlations, the shear
stress level required to cause liquefaction in a given number of cycles is
empirically related to in situ test parameters such as the SPT blow count (N1,60;
Seed et al. 1985), the Cone Penetration Test (CPT) tip resistance (qc1n; Robertson
and Wride 1998) or soil shear wave velocity (Vs1; Andrus and Stokoe 2000).
Because the empirical correlations are simple and incorporate a limited number of
parameters, the simplified procedure is widely used in practice.
After evaluating the shear stress required to cause liquefaction, these
values are also plotted versus depth (Figure 2-5). Liquefaction is predicted at the
site if the equivalent shear stresses induced by the earthquake are larger than the
shear stresses required to cause liquefaction.
When considering the results of stress-controlled laboratory tests, the
liquefaction potential is assumed to be primarily a function of the relative density
and the initial effective stresses acting on the sand. However, additional studies
(Mulilis et al. 1977, Seed 1979, Seed and Peacock 1971) revealed that several
factors beyond relative density and initial effective stress significantly influence
the results of stress-controlled tests. Some of these factors are the method of
sample preparation, prior seismic straining, and the lateral earth pressure
coefficient (Ko) and overconsolidation ratio (OCR).
Mulilis et al. (1977) performed cyclic triaxial tests on Monterey No. 0
sand specimens reconstituted at the same relative density by different methods. It
was discovered that the cyclic strength of the reconstituted specimens was greatly
affected by the sample preparation method. Figure 2-6 summarizes the results of
the effort by Mulilis et al. (1977) in terms of the cyclic stress ratio (CSR = σd/2σo′
15
Figure 2-6 The effect of specimen preparation method on cyclic strength (Mulilis
et al. 1977)
16
where σd=deviator stress and σo′=initial effective confining pressure) versus the
number of cycles to cause a peak cyclic pore water pressure ratio (u*, u*=∆u/ σo′
where ∆u=excess pore water pressure) of 100% or ± 2.5% axial strain. The
laboratory results show that the sample preparation technique can increase by a
factor of two the CSR required to cause liquefaction in a given number of cycles.
The most resistant sample preparation technique is moist tamping (Curves 1 and
2), while the least resistant is dry pluviation (Curve 7). These results indicate the
effect of soil fabric.
The effect of previous strain history was investigated by Seed et al.
(1975b). A freshly constituted specimen was compared with a specimen that had
been subjected to previous straining but did not experience initial liquefaction.
The cyclic stress ratio required to cause 100% pore water pressure ratio was found
to be approximately 1.5 times higher for the pre-strained specimen, as shown in
Figure 2-7. The gain in cyclic strength for the pre-strained specimen occurred
even though the previous strain history caused no significant densification.
The stress ratio required to cause liquefaction is significantly influenced
by the lateral earth pressure coefficient and overconsolidation. Figure 2-8 presents
the results obtained from an experimental study by Seed and Peacock (1971). In
this study, saturated sand specimens were tested in a Roscoe-type simple shear
device with rigid walls. Different degrees of overconsolidation were induced in
the specimens, with overconsolidation ratios ranging from 1 to 8. The data
indicated an improvement in the cyclic strength with increasing overconsolidation
ratio and corresponding increase in Ko. Similar effects of overconsolidation were
presented by Finn et al. (1971), Seed and Peacock (1971) and Ishihara (1996).
The practical use of the cyclic stress approach based on laboratory testing,
which is based on the assumption that the liquefaction potential is essentially a
17
Figure 2-7 Effect of seismic history on cyclic strength of sand (Seed 1979)
18
function of the relative density and initial effective stress, becomes more difficult
after discovering the influences of all these factors on the laboratory-determined
cyclic strength. To minimize the effects of all these factors, the cyclic strain
approach may be used. This approach is affected less by these factors and is more
fundamental than the cyclic stress approach because it is the induced strains and
rearrangement of soil particles that causes pore water pressure generation.
19
Figure 2-9 Measured pore water pressure in saturated sands after ten loading
cycles in strain-controlled cyclic triaxial tests (Dobry et al. 1982)
20
cyclic shear strain and the expected number of strain cycles. The cyclic shear
strain, γ c , is calculated by :
τ av a σ vo
γc = = 0.65 max rd ...............................................................(2-1)
G(γ c ) g G(γ c )
21
Seed et al. (1985) investigated sites that did and did not experience
liquefaction during earthquakes. The fines content at these sites ranged from
about 10 to 80%, and the investigation involved field measurements of SPT blow
count and estimates of the earthquake induced cyclic stress ratio, CSR (τ/σv′).
Seed et al. (1985) found that for the same stress-corrected penetration resistance,
(N1)60, the liquefaction resistance increases with increasing fines content (Figure
2-10). However, it is unclear whether the shift in the curves with fines content in
Figure 2-10 is due to increased liquefaction resistance or the fines content
reducing the blow count for the same liquefaction resistance.
Tronsco and Verdugo (1985) reported that the liquefaction potential for
mine tailings dams increases with increasing fines content. According to Ishihara
(1985) tailings soils are liquefiable because of their non-plastic nature. Ground
failure of fine-grained soils was evidenced recently during the 1999 Kocaeli
earthquake (Sancio et al. 2002) and 1999 Chi Chi earthquake (Stewart 2001).
Other investigations regarding the effect of fines content on liquefaction
resistance have focused on laboratory testing. Laboratory evaluations usually
compare the cyclic resistance of clean sand to that of sand with various fines
contents at: (1) a constant overall void ratio, (2) a constant sand skeleton void
ratio, or (3) a constant relative density.
22
cyclic resistance with increasing fines content, for specimens prepared to a
constant overall void ratio (Figure 2-12). In his study, Koester (1994) performed
Figure 2-10 Relationship between stress ratio causing liquefaction and (N1)60
values for silty sand for event of magnitude M=7.5 (Seed et al. 1985)
23
0.25
0.15
0.1
0.05
0
0 10 20 30 40 50 60
Silt content (%)
Figure 2-11 Variation of cyclic resistance with increasing silt content at a constant
overall void ratio (Chang et al. 1982)
Figure 2-12 Cyclic resistance curves for sand and silty sand at constant overall
void ratio of 0.558 (Koester 1994)
24
stress-controlled, cyclic triaxial tests on specimens prepared from a uniform clean
sand and mixtures of the sand with low-plasticity silt (PI = 4). A constant overall
void ratio of 0.558, corresponding to 50% relative density of the clean sand, was
taken as the basis for the comparison of cyclic resistances of the clean sand and
the silty sand specimens. Figure 2-12 indicates that the largest cyclic resistance
was obtained from the clean sand specimens. The cyclic resistance decreased with
increasing fines content up to 20%, and then increased with further increases in
fines content. Koester (1994) also investigated the effect of plasticity of fines for a
plasticity index range of PI = 0 (non-plastic) to PI = 40 and concluded that the
plasticity of fines has much less important effect than the fines content
(percentage of fines) on cyclic resistance of soils at a given overall void ratio.
In an effort to investigate the liquefaction potential of silty sands, Erten
and Maher (1995a) performed stress-controlled, cyclic triaxial tests on Ottawa
sand (C-190) and sand containing various amounts of Sil-Co-Sil 125 non-plastic
silt. The overall void ratio of the sand and silty sand specimens ranged between
0.52 and 0.58, indicating a medium dense state of the clean sand. Figure 2-13
displays the results from Erten and Maher (1995a). The cyclic resistance was
found to slightly decrease by addition of 10% fines. 20% fines content resulted in
significant decrease in cyclic resistance, and 30% fines content indicated
approximately the same cyclic resistance as that of 20% fines content.
Amini and Qi (2000) reported an increase in cyclic resistance with
increasing fines content at a constant overall void ratio (Figure 2-14). Their study
consisted of stress-controlled, cyclic triaxial tests on specimens with low
plasticity silt up to 50% fines content.
Polito and Martin (2001) used Monterey #0/30 sand and non-plastic silt,
and measured the cyclic resistance from stress-controlled, cyclic triaxial tests for
specimens with fines contents ranging from 0% (clean sand) to 100% (pure silt).
25
Figure 2-13 Cyclic resistance versus silt percent (Erten and Maher 1995a)
Figure 2-14 Effect of silt content on liquefaction of layered silty sand (Amini and
Qi 2000)
26
The soil specimens were prepared at a constant overall void ratio of 0.68. This
void ratio corresponds to 61% relative density of the clean sand. Polito and Martin
(2001) found that the cyclic resistance decreases dramatically, to less than one-
third of the clean sand value, at 35% fines content. This resistance remained
relatively constant between 35% and 75% fines content and increased to almost
the clean sand value at 100% fines (Figure 2-15).
Figure 2-15 Cyclic resistance of Monterey sand versus silt content at a constant
overall void ratio (Polito and Martin 2001)
27
1+ e
es = − 1 ………………………...………………………………(2-2)
1 − FC
where e = overall void ratio of soil; and FC = percentage of fines in decmial. To
better understand the derivation of this equation, a phase diagram is used (Figure
2-16). In this phase diagram, the volume of the voids is denoted by Vv and the
volumes of fine and sand particles are denoted by Vf and Vs, respectively. If the
total of volumes of solids is assumed to equal 1 (Vf +Vs=1), the overall void ratio
is equal to the volume of voids. Another assumption is that the specific gravity of
fines is equal to that of the sand. In this case, Vf is equal to FC and Vs equals (1-
FC). The sand skeleton void ratio is then calculated as if the fines are part of the
voids volume, and Equation 2-2 is obtained.
The concept of sand skeleton void ratio suggests that the fines fill the
voids formed between the sand grains and thus the behavior of a sand with a
moderate amount of fines should be governed by the sand skeleton void ratio
instead of the overall void ratio (Figure 2-17a). However, when the sand skeleton
Figure 2-16 Phase diagram for the calculation of sand skeleton void ratio
28
FC = Fines content
LFC = Limiting fines content
void ratio exceeds the maximum void ratio of the clean sand, there are sufficient
fines to prevent grain-to-grain contact of the sand particles (Figure 2-17b). In this
case, the fines constitute the dominant structure and carry the shear forces while
the coarse grains may act as reinforcing elements (Thevayanagam et al. 2000).
The transition point above which there are enough fines that the sand
grains are no longer in contact is called the limiting fines content. The limiting
fines content (LFC) of a sand may be calculated using the following equation:
Wfines G sf .e s
LFC = = ……………………...…......(2-3)
Wsand + Wfines G sf .e s + G ss (1 + e f )
where Wfines= fines solid weight; Wsand= sand solid weight; Gsf = specific gravity
of the fines; Gss = specific gravity of the sand; ef = void ratio of the fines; and es =
maximum index void ratio of the sand. Polito (1999) defined the limiting fines
content as the ratio of fines solid weight to sand solid weight and used the
following equation:
Wfines G sf .e s
LFC = = ………………………………….…………(2-4)
Wsand G ss (1 + e f )
29
content. Derivations of Equations 2-3 and 2-4 are shown on a phase diagram in
Figure 2-18.
Sand Vs = 1
Definition 2:
LFC=Wf / Ws = Gsf.es / [Gss.(1+ef)]
Figure 2-18 Phase diagram for the calculation of limiting fines content
30
Table 2-1 Limiting fines content of various soil combinations
LFC LFC
Soils Gss Gsf es ef (Eq. 2-3) (Eq. 2-4)
(%) (%)
31
Polito and Martin (2001) reported a similar increasing trend of the cyclic
resistance with increasing fines content below the limiting fines content for
Yatesville sand specimens prepared to a constant sand skeleton void ratio (Figure
2-20a). However, the same study showed a constant cyclic resistance below the
limiting fines content for Monterey #0/30 sand specimens prepared to a constant
sand skeleton void ratio, as shown in Figure 2-20b.
Figure 2-19 Cyclic resistance of silty sand at various constant sand skeleton void
ratios (Kuerbis et al. 1988)
32
Monterey #0/30 sand
Sand skeleton es = 0.75
Figure 2-20 Variation in cyclic resistance with silt content (Polito and Martin
2001)
33
2.4.3 Constant relative density
Finally, it is possible to consider a constant relative density and evaluate
the cyclic resistance for various fines contents. Singh (1994) investigated the
effect of non-plastic fines on the cyclic resistance from specimens reconstituted at
a constant relative density of 50%. The results of stress-controlled, cyclic triaxial
tests showed a decrease in cyclic resistance up to 20% fines content (Figure 2-21).
Singh (1994) found that the cyclic resistances from the specimens containing 20%
and 30% were approximately the same, and that the cyclic resistance of 100% silt
specimens was smaller than that of clean sand, yet larger than those of 20% and
30% silt specimens.
Figure 2-21 Cyclic resistance of silty sand at constant relative density (Singh
1994)
34
In their study, Polito and Martin (2001) suggested that for a constant
relative density of 30%, the cyclic strength is constant below the limiting fines
content. Above the limiting fines content, the cyclic strength is smaller, but it
remains approximately constant with further increases in fines (Figure 2-22).
There appears to be a problem with the evaluation of cyclic resistance in
terms of relative density presented by Polito and Martin (2001). According to
Figure 2-22, it appears that they tested specimens with high silt content at 30%
relative density. One should be able to confirm the existence of such specimens
from Figure 2-23, where they display all of the results (i.e., all fines contents and
relative densities) obtained in their study. Unfortunately, Figure 2-23 does not
support the existence of such specimens (30% relative density, FC > 60%). Using
the presented data in Figure 2-23, a possible way to develop such a figure as
Figure 2-22 is to determine the trend of the data points, as shown in Figure 2-24
by solid lines, and use that data to infer the results at 30% relative density.
However, it is not clear if this is how they developed Figure 2-22.
35
Figure 2-22 Variation in cyclic resistance with silt content for Yatesville sand
specimens prepared to constant relative density (Dr=30%) (Polito
and Martin 2001)
Figure 2-23 Cyclic resistance versus relative density for Yatesville sand (Polito
and Martin 2001)
36
Figure 2-24 Possible procedure used to generate Figure 2-22
37
a) Pore water pressure generation of sand and silty sand up to 30% fines content
b) Pore water pressure generation of sand and silty sand up to 60% fines content
Figure 2-25 Pore water pressure generation at N=10 and in sand and silty sand
(Erten and Maher 1995b)
38
strain levels of 0.01% and 0.4%. In contrast, addition of 60% fines showed an
increase in the excess pore water pressure generation. This trend suggests that the
excess pore water pressure generation decreases with increasing fines content up
to 10% and then starts to increase with further increases in fines contents between
shear strains of 0.01% and 0.4%. At strain levels larger than 0.4%, the excess pore
water pressure increases with increasing fines content.
The previously published studies on the effect of fines content on cyclic
strength, unfortunately, do not present a unified picture. In addition there is little
to no information presented in these studies with regard to the effect of fines on
pore water pressure generation. The goal of this study is to shed light on the
conflicting results reported in the literature and to advance the current findings by
evaluating how fines affect pore water pressure generation in strain-controlled
tests. Strain-controlled tests provide a better mechanism to study pore water
pressure generation because specimens are subjected to the same strain level and
shear strain is most closely related to inducing excess pore water pressure.
2.5 SUMMARY
During earthquakes the shaking of saturated granular soils results in
generation of excess pore water pressure. The developed excess pore water
pressure may cause a loss of strength (liquefaction) or stiffness that leads to the
settlement of buildings, landslides, the failure of earth dams, or other hazards.
There are two common approaches for evaluating the liquefaction
potential. These are: (1) the cyclic stress approach, and (2) the cyclic strain
approach. In the cyclic stress approach, the average cyclic stress induced by the
earthquake is compared to the cyclic strength of the soil determined from a
laboratory testing or field testing program. Liquefaction is expected where
earthquake-induced shear stresses are greater than the cyclic resistance of the soil.
39
The cyclic strain approach is based on the unique relationship between excess
pore water pressure generation and induced shear strains. This relationship is
determined from laboratory specimens, thus for an expected shear strain level the
amount of the pore water pressure generation can be predicted.
The literature review for the effect of fines on the pore water pressure
generation and cyclic resistance reveals that there is no consensus among the
researchers as to how fines affect the cyclic resistance and that there is little
information about the generation of excess pore water pressure in liquefiable
soils. Previously published laboratory investigations were summarized under
three distinct categories: (1) studies at constant overall void ratio, (2) studies at
constant sand skeleton void ratio, and (3) studies at constant relative densities.
The studies in each of these categories reported either a decreasing trend or an
increasing trend in the cyclic resistance with addition of fines. The results of
reviewed studies are summarized in Table 2-2.
40
Chapter 3: CYCLIC LABORATORY TESTS FOR
LIQUEFACTION STUDIES
3.1 INTRODUCTION
This chapter describes the different laboratory tests available to study
liquefaction behavior. In the field, prior to the dynamic earthquake loading, the
soil is assumed to be in the at rest (Ko) condition, as represented in Figure 3-1a.
The upward propagating shear waves produce an irregular, yet cyclic, history of
dynamic shear stresses on the horizontal and vertical planes (Figure 3-1). The
duration of the cyclic loading is usually assumed to be short enough that the pore
water cannot dissipate and thus the soil responds undrained during dynamic
loading. In laboratory testing, it is important to duplicate the in situ loading
conditions as accurately as possible. There are three major types of laboratory
tests used to study liquefaction. These are: (1) triaxial tests, (2) torsional shear
tests, and (3) direct simple shear tests. The extent to which these tests are capable
of simulating the stress state induced by a seismic event depends on the
nonuniformities of stresses and strains induced in the sample, the rotation of the
principal stress axes, and duplication of the plane strain condition. Also, each
testing method imposes a slightly different stress condition on the specimen,
which leads to different testing results. The following sections present a
discussion of each method and address the advantages and disadvantages of each.
41
3.2 CYCLIC TRIAXIAL TESTING
Cyclic triaxial testing for liquefaction evaluation was first performed by
Seed and Lee (1966). This study represented the first attempt to monitor the cyclic
liquefaction behavior of soil. In this type of testing, a cylindrical soil specimen
formed in a latex membrane is contained in a cell. The sample is initially
consolidated under an effective confining pressure σo′ and then subjected to a
Figure 3-1 Cyclic shear stresses beneath level ground during seismic loading
(Seed 1979)
42
cyclic axial stress of σd under undrained conditions until initial liquefaction occurs
or a specified axial strain level is reached. The stresses on a plane of 45o through
the sample are analogous to those produced on a horizontal plane in situ during an
earthquake. A typical triaxial configuration and simulation of the stresses during a
cyclic triaxial test are shown in Figures 3-2 and 3-3, respectively. An increase of
σd in the axial stress induces a shear stress of σd/2 on the 45o plane. When the
direction of the axial stress is reversed, the direction of the shear stress on the 45o
plane is also reversed. Hence, the 45o plane is subjected to a cyclic shear stress of
σd/2 in the opposite direction. The additional normal stress acting on the 45o plane
due to the cyclic axial loading is disregarded because it is purely compressive and
mostly transmitted to the pore water without changing the existing effective
confining stress σo′.
Figure 3-4 shows the typical response of a loose to medium dense sand
specimen from an undrained cyclic triaxial test. During the early loading cycles
(cycles 1 to 7), the specimen exhibits very little deformation, although the pore
water pressure increases progressively. Then, the specimen suddenly develops
large deformations after the seventh cycle. At this point, liquefaction occurs.
Typically, liquefaction in cyclic triaxial tests is defined as five percent double
amplitude axial strain (Ishihara 1996, Boulanger et al. 1991). The double
amplitude strain is defined as the total of the maximum compressive axial strain
(εc) and maximum extensive axial strain (εe) during a loading cycle (Figure 3-5).
It should be noted that a sinusoidal pattern is observed in the pore water pressure
generation because of increases in the mean stress induced by the cyclic deviator
stress.
Cyclic triaxial tests have been successfully used to determine the
liquefaction resistance of granular soils. As a result of many studies, it has been
shown that the liquefaction resistance of reconstituted sand specimens is primarily
43
Figure 3-2 Cross section of a typical triaxial test apparatus (Ishihara 1996)
44
Figure 3-3 Cyclic stresses in a cyclic triaxial test (Ishihara 1996)
45
Figure 3-4 Results from a stress-controlled cyclic triaxial test (Ishihara 1996)
5
4
3 Compression
2
Axial strain,ε (%)
1
εc
0
-1
εe
-2
-3 Extension
-4
-5
0 2 4 6 8 10
Time, T (sec)
46
influenced by initial confining stress, intensity of shaking, the number of loading
cycles, and void ratio (Ishihara 1996). The effects of initial confining stress and
cyclic shear stress are usually combined in terms of the cyclic stress ratio,
CSR=τ/σo′=σd/2σo′. Hence, the cyclic strength of a sand can be characterized by
varying the density of samples and measuring the number of cycles to liquefaction
for various cyclic stress ratios. Figure 3-6 presents such measurements from
cyclic triaxial tests performed on undisturbed and reconstituted sand specimens.
The undisturbed specimens were at 54% and 78% relative densities, whereas the
freshly deposited specimen was reconstituted at 56% relative density. According
to Figure 3-6, the specimen with 78% relative density exhibited larger cyclic
strength than the specimen with 54% relative density, and the comparison of 54%
and 56% relative density specimens shows that the cyclic strength of the
undisturbed specimen is approximately twice that of the freshly deposited
specimen.
Figure 3-6 Cyclic strength of sands in terms of cyclic stress ratio versus number
of loading cycles (Yoshimi et al. 1989)
47
Although the cyclic triaxial test has been used in many liquefaction
investigations, the stress conditions in this type of test are not an ideal duplication
of the in situ condition. In the in situ condition, if the ground surface is horizontal,
the major and minor principal stresses act on vertical and horizontal planes
(Figure 3-1). Earthquake-induced shear stresses cause a slight rotation of these
principal planes. In a typical cyclic triaxial test, the orientation of the principal
planes instantaneously rotates 90o during each cycle of loading, unlike the smooth
and slight rotation that occurs in the field (Seed and Peacock 1971). Another
limitation of cyclic triaxial testing is the friction between the end platens and soil
sample, which restricts the lateral deformation of the sample at its ends and also
creates nonuniform stress and strain conditions at the ends. The effects of end
restraint can be minimized by using specimens with a height to diameter ratio of
1.5 to 2.5 (Bishop and Henkel 1962, Tatsuoka et al. 1986). The tendency of the
sample to neck during the axial extension part of the stress cycle as failure
approaches has been reported among the major drawbacks of cyclic triaxial
testing (Seed and Peacock 1971, Casagrande 1975). Necking can cause very
dense samples to develop very large deformations, and thus, undergo premature
failure. Nevertheless, cyclic triaxial tests conducted with care may yield results
accurate enough for a basic understanding of the cyclic behavior of soils.
48
Figure 3-7 Cross section of a typical hollow cylinder torsional shear apparatus
(Ishihara 1996)
49
Figure 3-8 Stresses acting on elements of the wall of a hollow torsional shear
specimen (Hight et al. 1983)
50
Because no shear stresses are introduced on the wall of the specimen, the
radial stress (σr) is a principal stress (σr=σ2). To maintain mutual orthogonality of
principal stress axes, the direction of both σ1 and σ3 must rotate in a direction
perpendicular to the radial direction by the same angle, α (Hight et al. 1983).
During testing, in addition to the axial stress, a rotational shear is applied slowly
(as slow as 30 minutes/cycle) through the top cap of the specimen. As the soil
specimen softens, the lateral pressure coefficient (Ko) increases, and the specimen
tends to shorten and deform under the load. To keep the height constant the
confining pressure on both sides of the specimen needs to be adjusted. Typically,
a vertical LVDT used in a closed-loop feedback system regulates the confining
pressure. Typical results from cyclic torsional shear tests are presented in Figure
3-9. It should be noted that the pore water pressure develops progressively with
application of torsional shear stresses, yet in a non-fluctuating manner. As
discussed previously, in cyclic triaxial specimen the pore water pressure
fluctuates with application of the cyclic stresses (Figure 3-4).
Cyclic torsional shear testing, unlike cyclic triaxial testing, allows for
continuous and smooth rotation of the principal planes, which makes it more
representative of in situ conditions. In addition to the ability of replicating in situ
stresses, the torsional shear device has the advantage of allowing for large shear
strains, which becomes important when studying the residual strength of
liquefiable soils because this condition requires large strains. Also, testing at
complex load paths is possible by controlling the pressures on both the outer and
inner surface, as well as the vertical load (Towhata and Ishihara 1985).
Besides its advantages, cyclic torsional shear testing has some potential
disadvantages. The torsionally applied load produces non-uniform shear strains
along the radial direction of the specimen. In solid cylindrical specimens, the
51
Figure 3-9 Results from a typical torsional shear test (Ishihara 1996)
52
strain is zero in the center and reaches a maximum at the edge. This non-uniform
strain distribution is not an ideal condition and strongly affects liquefaction
testing results. To minimize this effect, hollow cylindrical specimens are
preferred. However, use of hollow specimen leads to an increase in the surface
area to volume ratio, which is likely to introduce significant membrane
penetration effects (O’Reilly and Brown 1991).
53
Figure 3-10 Cyclic loading in direct simple shear test (Seed and Peacock 1971)
54
Figure 3-11 Typical results from a stress-controlled cyclic direct simple shear test
(Seed and Idriss 1982)
55
The first known direct simple shear device was developed by the Royal
Swedish Geotechnical Institute (SGI) in 1936 (Kjellman 1951). In this
configuration, a short cylindrical soil specimen was surrounded by a rubber hose
which was confined within a stack of aluminum rings (Figure 3-12). Later in
1961, the Norwegian Geotechnical Institute (NGI) modified this SGI-type
apparatus by employing a wire-reinforced rubber membrane instead of the stack
of aluminum rings for the confinement of the soil specimen. This modified
configuration is called the NGI-type direct simple shear device (Bjerrum and
Landva 1966). The schematic representation of the NGI-type device can be seen
in Figure 3-13. Roscoe (1953) developed the so-called Roscoe-type direct simple
shear device at Cambridge University. This device tests a cubical soil specimen
which is laterally confined between rigid walls (Figure 3-14). While both the
Roscoe and NGI-type devices are similar in that they attempt to reproduce an in
situ simple shear condition, due to mounting difficulties the Roscoe-type device is
impractical for testing undisturbed specimens (Boulanger et al. 1991).
Franke et al. (1979) introduced the Bundersantal fur Wasserbau (BAW)
direct simple shear apparatus which is shown in Figure 3-15. This device was
developed to test saturated sand specimens under undrained conditions. The soil
specimen is confined in a plain rubber membrane and the lateral confinement is
provided by the cell pressure which is regulated through a closed-loop feedback
system to maintain a constant volume. Ishihara and Yamazaki (1980) described
the development of a bi-directional direct simple shear device which is depicted in
Figure 3-16. A short cylindrical specimen is enclosed laterally by a rubber
membrane that is confined in a stack of Teflon coated annular plates. Horizontal
loads are applied to the specimen at the top in two mutually perpendicular
directions. More recently, Boulanger et al. (1993) developed the University of
California at Berkeley bi-directional cyclic direct simple shear device (UCB-2D)
56
which is schematically illustrated in Figure 3-17. This apparatus was designed to
overcome many of the shortcomings of the previous simple shear devices. The
UCB-2D system accepts 4-inch diameter specimen confined in a rubber
membrane. The lateral confinement is achieved by application of cell pressure
which also allows for testing fully saturated specimens.
Figure 3-12 Schematic of the SGI-type simple shear device (Kjellman 1951)
57
Figure 3-13 NGI-type simple shear apparatus (Bjerrum and Landva 1966)
58
Figure 3-14 Roscoe-type simple shear apparatus (Roscoe 1953)
59
Figure 3-15 Schematic of the BAW simple shear apparatus (Franke et al. 1979)
60
Figure 3-16 Schematic of the bi-directional simple shear device (Ishihara and
Yamazaki 1980)
61
Figure 3-17 Cross section of the UCB-2D simple shear apparatus (Boulanger and
Seed 1993)
62
membrane which was confined by rigid walls. This configuration allowed for the
conditions of constant vertical load, constant height, and zero lateral strain; thus
constant volume during undrained testing. Although the Roscoe-type design
meets all of the boundary conditions required for undrained, cyclic direct simple
shear testing, specimen preparation in a cubical form is extremely difficult
(Peacok and Seed 1968, Finn et al. 1971).
In an effort to overcome specimen preparation difficulties Finn and Vaid
(1977) and Finn et al. (1978) used an NGI-type system with a wire reinforced
membrane and tested dry sand specimens to infer the cyclic resistance of the soil.
The specimen height was kept constant yet the vertical load was allowed to
fluctuate. The drop in the vertical stress was assumed to be equal to the pore water
pressure that would have been generated in a saturated specimen (Figure 3-18).
Franke et al. (1979) introduced a simple shear device in which the lateral
confinement was provided by chamber pressure through a plain rubber
membrane. The vertical load was independent of the cell pressure and kept
constant. Drainage was not allowed during cyclic testing, and a constant height
was achieved by adjusting the lateral confinement (cell pressure). Boulanger et al.
(1993) described a similar system in which they used saturated specimens with
constant height and thus constant volume. More recently, Kammerer (2002)
introduced a unique method for undrained, cyclic direct simple shear testing. The
specimens were confined by a wire-reinforced membrane. The vertical load was
kept constant, and the wire-reinforcement kept the cross-sectional area
approximately constant. Because drainage was not allowed, constant volume was
achieved. The advantage of this method is that Ko conditions are somewhat
maintained. The method introduced by Kammerer (2002) was adopted for this
research.
63
Figure 3-18 Constant volume cyclic direct simple shear test on dry sand
(Finn 1985)
Cyclic direct simple shear testing, like all other laboratory testing
procedures, has its own unique shortcomings. Significant research has focused on
the problems faced during simple shear testing. The most recognized drawback of
simple shear testing is the inability to apply proper complementary shear stresses
on the vertical sides of the specimen. The lack of complementary shear stresses
along the sides produces non-uniform stresses within the specimen, which causes
64
the specimen to rock or pinch depending on the setup. In the context of stress
uniformity, Amer et al. (1987) performed cyclic direct simple shear tests (with an
NGI-type device) on circular samples of dry sands using different specimen
diameter to height ratios (D/H). Amer et al. (1987) concluded that for D/H ratios
greater than 8, the dynamic properties measured in cyclic direct simple shear were
essentially independent of D/H, and thus, were not significantly affected by the
lack of complementary shear stresses. In another experimental study, Franke et al.
(1979) tested samples with diameter to height ratios between 3.75 and 7.5 in an
NGI-type cyclic direct simple shear device and found no difference in the
resulting liquefaction behavior. Numerical studies by a number of investigators
(Lucks et al., 1972; Shen et al., 1978; Saada et al., 1982; Budhu and Britto, 1978)
also concluded that in samples with D/H ratios greater than 4, the lack of
complementary shear stresses affected only a relatively small portion of the
sample.
The testing system used for this research is the GCTS cyclic direct simple
shear (CDSS) system, which is discussed in detail in Chapter 4. The GCTS CDSS
system is an NGI-type apparatus, and accepts 4-inch diameter specimens. The
device has been designed to impose a chamber pressure, thus conventional back-
pressure saturation procedures can be employed and tests on fully saturated
specimens are possible. The undrained testing procedure described by Kammerer
(2002) was used.
3.5 SUMMARY
There are three major types of laboratory liquefaction tests. These tests
are: (1) the cyclic triaxial test, (2) the torsional shear test, and (3) the cyclic direct
simple shear test. Of these three categories, cyclic triaxial tests have been widely
65
used in liquefaction investigations despite the fact that they are not providing
ideal duplication of in situ loading conditions. Cyclic triaxial tests are the easiest
tests to perform and are primarily useful for parametric studies. Torsional shear
tests and cyclic direct simple shear tests are more reliable in terms of the
representation of the in situ loading conditions. Thus, results from these tests are
valuable for understanding the behavior of liquefiable soils. The drawbacks of
the torsional shear test include the non-uniform shear strains along the radial
direction of the specimen for solid specimens and the increased surface area to
volume ratio for hollow specimens. The most common types of simple shear
devices are the Roscoe-type apparatus, which tests cubical specimens, and the
NGI-type apparatus, which tests short cylindrical specimens. Most simple shear
devices fall under one of these two categories. The main disadvantage of cyclic
direct simple shear test is the inability to apply proper complementary shear
stresses on the vertical sides of the specimen. This problem is minimized by
employing specimens with a diameter to height ratio of at least four. The testing
system used for this research is an NGI-type cyclic direct simple shear system.
The majority of this research focused on cyclic direct simple shear testing, but
some cyclic triaxial tests were also performed.
66
Chapter 4: CYCLIC STRAIN-CONTROLLED TESTING
PROGRAM
4.1 INTRODUCTION
This chapter presents the details of the testing program, which primarily
aimed to assess the pore water pressure generation characteristics of sands and
silty sands as related to shear strain and number of loading cycles. The following
section introduces the testing set-up used for this investigation. Next, saturation,
consolidation, and cyclic testing procedures are outlined. The soils tested are
discussed in terms of grain size distribution and index properties. The testing
program is presented, followed by a discussion of the specimen preparation.
Finally, verification of the testing system is presented.
67
Volume Change Device
Vacuum Chamber
Pressure Panel
Water De-airator
68
top cap, (5) an internal horizontal LVDT permitting measurement of horizontal
shear displacement, (6) an internal LVDT providing the measurement of the shear
displacement of the bottom platen relative to the top cap, (7) an external
horizontal LVDT, and (8) differential pressure transducers measuring cell
pressure and effective stress. All components of the instrumentation system are
summarized in Table 4-1 and presented in Figure 4-2.
Capacity or
Measurement Sensor Sensitivity
Range
Effective Pressure (Pore water Pressure Transducer ± 125 psi ± 0.1 psi
pressure)
69
Vertical Load Actuator
Internal Vertical LVDT’s
Vertical External LVDT
Figure 4-2 Locations of sensors in GCTS cyclic direct simple shear apparatus
The horizontal (shear) displacement measurements are of crucial
importance in cyclic direct simple shear testing, and particularly in strain-controlled
tests. As discussed in the previous chapter, the lack of complementary shear
stresses along the vertical sides of a simple shear specimen leads to a non-uniform
stress distribution, which causes the specimen to rock. The original design of the
GCTS cyclic direct simple shear device included an internal LVDT attached to the
roller base and thus allowed for the measurement of the horizontal displacement of
the base platen only. To compute shear strain with this measurement, it is assumed
that the top cap does not move. However, rocking of the specimen causes the top
cap to translate. To incorporate the effect of rocking, a unique design that
measures the shear displacement of the bottom platen relative to the top cap was
developed. This design is shown in Figure 4-3. The measurements from the
original LVDT (measuring the displacement of the base platen only) and the
relative displacement LVDT indicated a significant difference, particularly at small
strains. Figure 4-4 shows the shear strain measurements from a medium dense
specimen (Dr = 50%), which was subjected to a strain-controlled, undrained, cyclic
direct simple shear loading. The measurement based on only the deformation of the
base platen overestimates the shear strain applied to the specimen by as much as
three times. For this test, the relative deformation LVDT measured a shear strain of
0.01% whereas the base LVDT measured a shear strain of 0.03%. Because the
relative deformation LVDT takes into account the rocking effect and provides more
realistic shear strains, it was adopted as the primary shear strain LVDT for all of
the shear strain applications and measurements throughout this research.
71
Specimen
confined with a
wire-reinforced
membrane
L-shaped plate
connected to top platen
Figure 4-3 Set-up of roller base LVDT and relative shear deformation LVDT
72
0.1
0.08
0.06
0.04
Shear Strain, γ, %
0.02
-0.02
-0.04
-0.06
shear defo from base platen relative shear defo
-0.08
-0.1
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Number of Cycles, N
Figure 4-4 Shear strain measurements from the relative LVDT and base LVDT
73
Figure 4-5 Assembled simple shear cell
dioxide was circulated through the sample for approximately 15 minutes to displace
any air from the soil pores. Flushing the sample with carbon dioxide helps achieve
a higher degree of saturation, because the carbon dioxide is more soluble in water
than air. Next, de-aired water was allowed to flow through the specimen from the
bottom until no further bubbling was observed in the vacuum chamber.
Full saturation of soil specimens is essential to accurately record pore water
pressures and to obtain meaningful results during liquefaction testing (Chaney
1978). Therefore, saturation of specimens was achieved through application of
relatively low backpressures of approximately 100 kPa. All B-value measurements
were performed at a stress increment of 10 kPa and the net effective stresses acting
on the specimen during the saturation process were kept around 15 kPa. To
74
minimize the air diffusion through membrane, the cell pressure was kept relatively
low (~80 kPa). For loose specimens 2 to 3 hours of backpressure application
typically resulted in B-values greater than 0.96. Dense specimens and specimens
with fines content greater than 10% required hours of back pressure application (in
some cases 48 hours) to attain acceptable B-values. After a minimum B-value of
0.96 was obtained, the consolidation procedure was started.
The use of a wire-reinforced membrane enforces approximate Ko conditions
in the soil by minimizing lateral deformations during consolidation and testing.
One-dimensional consolidation of the specimen was achieved by linearly
increasing the vertical pressure to the desired value. The cell pressure was not
changed during this process. Volume change and axial deformation measurements
during consolidation were made using the volume change transducer and the
internal LVDTs, respectively. The volume change indicated by the axial LVDT
measurements was compared with that measured by the volume change transducer
to ensure that consolidation was one-dimensional and minimal bulging occurred
during consolidation. The sample was allowed to consolidate for thirty minutes.
For cyclic triaxial testing, plain latex membranes were used and a similar
saturation procedure was incorporated. For consolidation, the cell pressure was
increased to achieve the desired effective confining pressure. Isotropic
consolidation of the specimen was completed in thirty minutes.
75
The cyclic shear strain was applied in a sinusoidal manner. To duplicate the
field loading conditions during an earthquake, the vertical load was kept constant
while approximately zero lateral strain was achieved with the help of the wire-
reinforced membrane. The experiment was terminated after 50 cycles during small
strain tests and after liquefaction was achieved during large strain tests.
Liquefaction was defined as an excess pore water pressure ratio (u*) of about 90%
or higher. Typical results from a strain-controlled cyclic direct simple shear test are
displayed in Figure 4-6. A sand specimen is subjected to a constant amplitude shear
Figure 4-6 Results from a typical strain-controlled undrained cyclic direct simple
shear test
76
strain level (~0.3% in this case, Figure 4-6a). The pore water pressure generation
follows a smooth progressive pattern, until reaching a pore water pressure ratio, u*,
of about 90% (Figure 4-6b). The induced shear stress decreases with time and
becomes extremely low as liquefaction is approached (Figure 4-6c). Such a trend in
the shear stress causes the stress-strain cycles to form loops as those shown in
Figure 4-6d.
In the case of the strain-controlled, cyclic triaxial tests, after consolidating
the specimen to the desired effective confining stress, uniform axial strain cycles
were applied.
In an effort to increase the number of tests and thus amount of the data,
staged testing was evaluated. Hsu and Vucetic (2004) used staged testing when
studying the cyclic settlement of soil specimens. Figure 4-7 presents an evaluation
of the staging procedure for pore water pressure generation evaluation. Three
different clean sand specimens, all prepared at a relative density of 44%, were
tested using different staging conditions. Specimen 1 was first subjected to a shear
strain level of 0.03%. After 20 cycles the test was stopped (u*= 32% at this point)
and drainage was allowed. After drainage, the relative density of the specimen did
not change significantly (~ 46%). The drainage valves were then closed and a
shear strain level of 0.3% was applied. The pore water pressure generation of this
specimen at 0.3% shear strain is shown with Curve 1 in Figure 4-7. Similarly,
Specimen 2 was tested first at a shear strain of 0.01% (u*= 11% at this point),
allowed to drain, and then tested at a shear strain of 0.3%. The response of
Specimen 2 at 0.3% shear strain is displayed by Curve 2. Finally, Specimen 3 was
directly tested at 0.3% shear strain with no previous straining. Curve 3 shows the
response of this specimen. As can be seen, previous straining (staged testing) has a
significant effect on the generation of excess pore water pressure. Depending on the
level of previous straining, dramatic decreases in pore water pressure generation
may occur, as observed in Specimen 1. The effect is much smaller when the
77
1
, ∆U/σ'v 0.9
0.8
Excess pore pressure ratio
0.7 3
0.6
2
0.5
0.4 1
0.3
1 (staged: first γ=0.03% and then γ=0.3%)
0.2
2 (staged: first γ=0.01% and then γ=0.3%)
0.1 3 (non-staged: only γ=0.3%)
0
0 2 4 6 8 10 12 14 16 18 20
Number of Cycles, N
Figure 4-7 Effect of staged testing on the generation of excess pore water pressure
previous strain level is close to the threshold strain. Based on these observations,
staged testing was not performed and each specimen was subjected to a single shear
strain level throughout this research.
78
100
90 Uniformity coefficient, Cu=1.6
80 Coefficient of curvature, Cc=1.1
Percent passing (%)
70
60
50
40
30
20
10
0
10 1 0.1 0.01
Grain size (mm)
Dyvic et
This Kammerer Riemer Rad Mulilis
Measured Properties al.
Study (2002) (1997)* (1987)* (1975)*
(1982)*
Minimum Void Ratio, emin 0.57 0.541 0.53 0.56 0.56 0.57
Maximum Dry Unit Weight, γd,max 16.82 16.81 16.99 16.9 16.66 16.53
3
(kN/m )
Maximum Void Ratio, emax 0.85 0.885 0.86 0.85 1.059 0.85
Minimum Dry Unit Weight, γd,min 14.27 13.96 13.98 14.3 12.63 14.05
3
(kN/m )
*Monterey #0 sand
79
Monterey #0/30. The values measured in this study agree well with those reported
by others.
To form silty sand specimens for the investigation of the effect of fines
content on pore water pressure generation, various amounts of Sil-Co-Sil 52 non-
plastic silt were mixed with the Monterey #0/30 sand. Sil-Co-Sil 52 is produced by
U.S. Silica Company from crushed silica. The grain size distribution of this non-
plastic silt is presented in Figure 4-9. It has a maximum grain size of 0.074 mm and
a minimum grain size of less than 0.002 mm. The median grain size, D50, is 0.0135
mm. The silt is white in color and has a specific gravity of 2.65. It is classified as
non-plastic silt (ML). The grain size distributions and index properties of
Monterey #0/30 sand and Sil-Col-Sil 52 non-plastic silt mixtures for 5%, 10%,
15%, and 20% fines contents are shown in Figure 4-10 and Table 4-3, respectively.
A detailed discussion about the determination of the maximum and minimum void
ratios of these soil mixtures is presented in Chapter 6 (Section 6.2).
#200 sieve
100
90
80
Percent passing (%)
70
60
50
40
30
20
10
0
1 0.1 0.01 0.001
Grain Size (mm)
80
100
90
80
Percent passing (%)
70
60 FC=15%
50
FC=10%
40
FC=20%
30
20
FC=5%
10
0
10 1 0.1 0.01 0.001
Grain size (mm)
Figure 4-10 Grain size distribution of Monterey #0/30 sand and Sil-Co-Sil 52 non-
plastic silt mixtures
Measured Properties
0 5 10 15 20
81
Another soil used in this research was the aggregate sand. Aggregate sand
specimens were tested under laboratory conditions and the excess pore water
pressure generation from these specimens was compared to those measured in situ
on the same soil as presented in Chapter 7. Aggregate sand is a clean, poorly
graded (SP) sand with fines content of less than 1%. This sand was obtained from
Capital Aggregates. The maximum and minimum void ratios were found to be 0.64
and 0.43, respectively. The median grain size, D50, is 0.7 mm, the uniformity
coefficient, Cu, is 3.75, and the coefficient of curvature, Cc, is 0.74. The grain size
distribution and index properties of the aggregate sand are shown in Figure 4-11
and Table 4-4, respectively.
100
90
80
Percent passing (%)
70
60
50
40
30
20
10
0
10 1 0.1 0.01
Grain size (mm)
82
Table 4-4 Index properties of aggregate sand
83
Table 4-5 CDSS Tests on Monterey #0/30 sand
Post consolidation
Vertical Cyclic
Overall Sand Relative effective shear
B-value at void skeleton density stress strain
the end of ratio void (Dr) (σv') (γ )
Test saturation (e) ratio (es) (%) (kPa) (%)
CS11 0.99 0.711 0.711 49.6 100.0 0.100
Note:
84
under a vertical effective stress of 100 kPa and then subjected to cycles of a single
shear strain level until full liquefaction (u* ≥ 0.9) was reached or 50 cycles of
loading were applied. The details of specimen preparation will be discussed in the
next section, and the results of these tests will be presented in Chapter 5.
In the second testing series, excess pore water pressure generation from
strain-controlled cyclic direct simple shear tests was compared to that from cyclic
triaxial tests. For this comparison, a series of cyclic triaxial tests was performed on
Monterey #0/30 sand specimens, which were reconstituted at a 50% relative
density. Thus, only medium dense sand was considered and the results from these
tests were compared to those obtained from cyclic direct simple shear tests at the
same relative density. Cyclic triaxial tests were performed at four different shear
strain levels ranging between 0.01% and 0.3%. The effective confining stress was
100 kPa for all of these tests. The testing conditions are presented in Table 4-6, and
the discussion of the results can be found in Chapter 5.
Post- Confining
consolidation Cyclic
B-value at effective Axial strain
relative shear strain
Test # the end of stress (ε)
saturation
density
(σ3')
γ=ε(1+ν)
(Dr)
85
The goal of the third testing series was to investigate the effect of non-
plastic fines on excess pore water pressure generation. Strain-controlled,
undrained, cyclic direct simple shear tests were conducted on silty sand specimens.
The specimens were prepared from mixtures of Monterey #0/30 sand and Sil-Co-
Sil 52 non-plastic silt. Three different controlling index parameters were taken into
account during the specimen preparation. The controlling index parameters are: (1)
relative density, (2) sand skeleton void ratio, and (3) overall void ratio. The
selection of these parameters is discussed in Chapter 6. The silty sand tests were
grouped according to the controlling index parameters. Tables 4-7, 4-8, and 4-9
present the testing conditions for the tests performed at each index parameter. For
the tests at constant relative density (Tables 4-7) and at constant sand skeleton void
ratio (Table 4-8), the fines content ranged from 5% to 20%. The tests at a constant
overall void ratio are shown in Table 4-9. The fines content under this category was
limited to 10% due to specimen preparation difficulties at higher fines content. It
should be noted that the series at 10% fines content for all of the three index
parameters are common and presented separately in Table 4-10. This is simply
because the index parameters of these specimens were such that they could be used
with each of the three categories. The results of all these tests are discussed in
detail in Chapter 6.
86
Table 4-7 CDSS Tests on sands containing fines at constant Dr=50%
Post consolidation
Fines B-value at Vertical Cyclic
Test content the end of Overall Sand Relative effective shear
(% ) saturation void skeleton density stress strain
ratio void (Dr) (σv') (γ )
(e) ratio (es) (%) (kPa) (%)
DFC51 5 0.98 0.634 0.720 50.5 100 0.100
87
Table 4-8 CDSS Tests on sands containing fines at constant es=0.76
Post consolidation
Fines B-value at Vertical Cyclic
Test content the end of Overall Sand Relative effective shear
(% ) saturation void skeleton density stress strain
ratio void (Dr) (σv') (γ )
(e) ratio (es) (%) (kPa) (%)
ESFC51 5 0.99 0.674 0.762 31.4 100 0.300
Note:
ESFC5: silty sand with 5% fines content at constant sand skeleton void ratio
ESFC15: silty sand with 15% fines content at constant sand skeleton void ratio
ESFC20: silty sand with 20% fines content at constant sand skeleton void ratio
Post consolidation
Fines B-value at Vertical Cyclic
Test content the end of Overall Sand Relative effective shear
(% ) saturation void skeleton density stress strain
ratio void (Dr) (σv') (γ )
(e) ratio (es) (%) (kPa) (%)
EFC51 5 0.96 0.590 0.674 71.4 100 0.300
Note: EFC5: silty sand with 5% fines content at constant overall void ratio
88
Table 4-10 CDSS Tests on sands containing fines
Post consolidation
Fines B-value at Vertical Cyclic
Test content the end of Overall Sand Relative effective shear
(% ) saturation void skeleton density stress strain
ratio void (Dr) (σv') (γ )
(e) ratio (es) (%) (kPa) (%)
FC101 10 1.00 0.590 0.767 50.0 100 0.300
The fourth and final testing series was performed on reconstituted aggregate
sand specimens. A total of 13 specimens were tested. A list of these tests, along
with their testing conditions is presented in Table 4-11. The results from these tests
were compared with those measured in situ (Rathje et al. 2005) in Chapter 7.
89
Table 4-11 CDSS Tests on aggregate sand
T8 1.00 41 25 0.0052
T9 0.98 39 25 0.0100
90
4.7 SPECIMEN PREPARATION
There are four specimen preparation procedures that are widely used in
liquefaction studies. These are:
1. Dry pluviation (pluvial spreading through air, pluvial spreading
through air with tapping, pluvial spreading through air with
vibratory compaction)
2. Water sedimentation
3. Moist undercompaction
4. Slurry deposition
All of these methods may be used to form clean sand specimens, and generally the
latter two are preferred to reconstitute sand specimens with fines.
91
field specimen preparation technique (Rathje et al. 2005), as well as to replicate in
situ soil deposition processes.
The procedure for preparing the aggregate sand specimens is as follows. A
101.6-mm (4-in) diameter wire reinforced membrane of moderate stiffness (C=1.0),
manufactured by the Norwegian Geotechnical Institute (NGI), was used to confine
the soil. This membrane was secured to the bottom platen by O-rings for sample
preparation. A mold was placed around the membrane and filled half with de-aired
water. Pre-weighed dry sand was poured into the mold with great caution. The
approximate desired soil density was achieved by adjusting the drop height of the
sand (Kammerer 2002). An approximate drop height of 5 mm was found to create a
loose fabric and result in initial relative densities between 32% and 44% for the
aggregate sand. The sample was then carefully placed in the cyclic direct simple
shear device, the top cap was lowered, and the membrane was secured to the top
cap with O-rings. The drain lines were connected to both the top cap and the base
platen of the specimen. A vertical pressure of 10 kPa was applied. To achieve the
desired relative density and to improve contact between the soil and end caps, the
base platen was gently tapped while the mold was in place. After the desired
specimen height was obtained, the mold was removed and the O-rings at the top
and bottom were tightened further around the membrane by application of hose
clamps. This practice was found to be useful in preventing potential bulging of the
non-reinforced ends of the membrane during development of high pore water
pressures. The dimensions of the sample were then measured and recorded.
92
layers when the upper layer is placed. This method has been widely used in
liquefaction studies. The main advantage of the undercompaction method is very
uniform specimens can be formed at a range of densities.
In this study, the moist undercompaction technique was used for
reconstituting Monterey #0/30 clean sand specimens as well as silty sand
specimens. Moist undercompaction eliminates the problems of particle segregation
associated with dry pluviation or water sedimentation.
The specimens prepared by moist undercompaction were constructed with
the following procedure: A predetermined amount of soil (clean sand or silty sand)
was mixed with de-aired water such that the degree of saturation in the soil was
approximately 50%. The moist soil was contained in a sealed bowl and allowed to
sit overnight to ensure uniform water content distribution. The wire reinforced
membrane was secured to the bottom platen by O-rings for sample preparation. A
mold was placed around the membrane and a vacuum was applied to pull the
membrane to the mold. Then, 3-in diameter filter paper was placed on the porous
stone at the bottom platen. Next, half of the moist soil was carefully spooned into
the mold. A manual tamper was then lowered on the soil and used to press the soil
down to its correct height. The tamper made several trips near the wall of the mold
so that a uniform distribution throughout the first lift was achieved. After the first
layer was compacted to its correct height, the tamper was removed and the surface
of the soil layer was roughened to a depth of about 3-mm. Next, the second half of
the soil was spooned into the mold and the compaction procedure was repeated.
The second lift was compacted to a height slightly greater than the target height to
allow for seating of the top cap. The sample was then carefully transferred to the
cyclic direct simple shear device, the top cap was lowered, and the membrane was
secured to the top cap with O-rings. The drain lines were connected to both the top
cap and the base platen of the specimen. A vertical pressure of 10 kPa was applied.
To achieve the desired relative density and to improve contact between the soil and
end caps, the base platen was gently tapped while the mold was in place. After the
93
desired specimen height was obtained, the mold was removed and the O-rings at
the top and bottom were tightened further around the membrane by application of
hose clamps. The dimensions of the soil specimen were then measured and
recorded.
In the case of triaxial tests, the same procedure was followed except that
five soil layers were used. Because the triaxial specimens (2.5-in diameter by 5-in
height) were much taller than the simple shear specimens (4-in diameter by 1-in
height), more soil layers were required to attain a homogenous fabric.
Undercompaction proved to work quite well on specimens having fines
content up to 20%. Table 4-12 presents detailed information regarding the
deformation during saturation and consolidation of the specimens formed by the
undercompaction procedure at 50% relative density. Figures 4-12 and 4-13 display
the volume change that occurred during the saturation and consolidation processes,
respectively. Clean sand specimens appear to undergo more volume change during
the saturation phase than the silty sand specimens. However, the silty sand
specimens tend to exhibit greater contraction during the consolidation phase.
94
Table 4-12 List of tests performed on specimens reconstituted by undercompaction
Saturation Consolidation
γ ∆V ∆V
Test #
(%) ∆V
3
εV (LVDT) (burette) εV
(cm ) (%) 3 3 (%)
(cm ) (cm )
Notes:
1. CS: clean sand FC10: sand with 10% fines FC20: sand with 20% fines
2. γ : shear strain ∆V : volume change εV : volumetric strain εa : axial strain
3. All specimens were prepared at 50% postconsolidation relative density
4. Volume change during consolidation was measured by using both the internal
vertical LVDTs and the volume change transducer connected to the burette.
Almost equal values of volume changes indicated 1-D consolidation.
95
1.8
1.6
1.4
Volume Change (cm 3)
1.2
1
0.8
0.6 Clean Sand
0.4 Sand w ith 10% FC
0.2 Sand w ith 20% FC
0
Specimen1
Specimen2
Specimen3
Specimen4
Specimen5
9
8
7
Volume Change (cm 3)
6
5
4
3 Clean Sand
2 Sand w ith 10% FC
0
Specimen1
Specimen2
Specimen3
Specimen4
Specimen5
97
Table 4-13. Results of the attempts made to form specimens with 35% and higher fines content
Change in
Change in Total
Fines Specimen height due
Specimen Target Dr height due change in Axial Strain
content Preparation Initial w (%) to H2O
# (%) to CO2 height (%)
(%) Method flushing
(mm) (contraction)
(mm)
1 Undercompaction
35 50 (moist 5 1.95 2.35 4.3 18.9
compaction)
98
Undercompaction
2 35 50 (moist 8 2.1 2.47 4.57 20.4
compaction)
Undercompaction
4 100 50 (moist 16.5 3.04 1.22 4.26 18.8
compaction)
Undercompaction
5 35 85 (moist 7.8 0.067 0.183 0.25 1.04
compaction)
98
Figure 4-14 Cyclic resistance versus relative density for Monterey sand (Polito
and Martin 2001)
100
of 50 kPa to 100 kPa, depending on the desired density, is applied on the porous
stone for approximately 2 to 3 days to consolidate the specimen. Following the
consolidation, the specimen along with the forming mold is placed in a freezer.
The frozen specimen is then taken out. The split mold is dismantled and the intact
specimen is mantled with a rubber membrane and with two porous stones at both
ends. The specimen is then placed in the cell and the drainage lines are connected.
A vacuum of 20 kPa is applied while thawing the specimen by placing a heater
close to the cell. Next, CO2 is percolated through the specimen for about two
hours under a very low gas pressure of approximately 10 kPa. After flushing with
CO2, deaired water is allowed to seep from the bottom of the specimen and then
conventional back-pressure saturation is performed.
Kuerbis and Vaid’s method (1988)
Kuerbis and Vaid (1988) introduced the so called “slurry deposition
method” for preparing well graded sand and silty sand specimens. The idea was to
obtain homogenous specimens and to simulate closely the in situ soil deposition
process. The procedure can be summarized as follows:
The sand and the silt portions are separately deaired prior to forming the
slurry. Deaired soils are deposited into a plexiglass cylindrical tube, which
contains deaired water and serves as a mixing tube. One end of this tube is sealed
with the porous stone underlain by a metal disk. The other end is closed with a
tapered rubber stopper. The slurry within the tube is then mixed continuously for
approximately 20 minutes by shaking and rotating the mixing tube. The slurry is
then allowed to settle holding the tube vertically. Next, the metal disk is removed
from the bottom of the tube and the tube is placed on the base platen of the testing
apparatus (triaxial in their case). This placement of the tube is done in a water
bath so as to maintain the full saturation of the specimen. The membrane is then
rolled over the mixing tube and sealed to the base platen with O-ring. The split
mold is assembled and a vacuum is applied so as to hold the membrane tight
101
against the mold. The rubber stopper is removed from the top of the mixing tube
and now the slurry can be deposited into the mold by carefully withdrawing the
mixing tube. Next, the top cap is placed with a great care and a slight vacuum is
applied to the specimen. The split mold can now be removed, and the specimen is
ready to be tested.
Slurry in Batch Consolidometer (Brandon et al., 1991)
In their study, Brandon et al. (1991) developed a procedure for
reconstituting uniform silty sand specimens. A slurry of silty sand is prepared by
mixing the dry soil with water. The slurry is then consolidated to Ko conditions in
a batch consolidometer. Next, the consolidated soil is trimmed to form the
desired triaxial specimens. This method results in a dense fabric and therefore is
not applicable for reconstituting loose fabricated specimens.
Romero and Boulanger’s Approach (1998)
In a master study conducted by Romero and Boulanger at University of
California at Davis in 1998, a new approach for preparing silt specimens was
developed. Their approach involved one-dimensional consolidation in a specially
designed loading frame with a three-way split mold. This method seems to be
similar to the batch consolidometer procedure discussed above with the exception
that it does not require trimming of the specimen (the batch consolidometer does
involve trimming) after consolidation, thus minimizes the specimen disturbance.
As in the case of batch consolidometer, this approach also yields dense
specimens.
Applicability of the Slurry Method to the GCTS CDSS Specimens
The GCTS CDSS system accommodates a 4-inch diameter specimen. The
height of the specimen is desired to be 1 inch or less, and the soil specimen is
reconstituted in a wire-reinforced membrane. The GCTS CDSS configuration
does not allow in-place specimen preparation. Therefore, the specimen has to be
prepared outside and then transferred into the cell.
102
As outlined above, there is no a standard slurry deposition method for
reconstituting laboratory soil specimens. To be able to apply the slurry method
with the GCTS CDSS system, the following procedure was adopted:
1. The split mold is assembled around the membrane.
2. The desired soil weight is determined from the target void ratio.
3. The portions of sand and silt are first dry mixed thoroughly in a
bowl until visible homogeneity has been achieved.
4. Deaired water is added slowly while stirring the soil mixture. A
water content of about 25% was found to yield a slurry mixture for
sand containing 35% fines. Lower water contents resulted in a
sticky and non-uniform mixture while higher water contents
caused segregation of the sand particles and thus a non-
homogenous slurry.
5. The slurry is then deposited slowly, with the help of a spoon, into
the split mold.
6. Next, the mold is carefully transferred into the cell and the top cap
is applied.
7. Drainage lines are connected.
8. A vacuum of 10 kPa is applied before the split mold is removed.
Effectiveness of the method and observations from initial trials
To test the effectiveness of the slurry method and its applicability to the
GCTS CDSS set-up, five trials were attempted to form specimens with 35% fines
content. The results are summarized in Table 4-14, and the following observations
were recorded:
103
Table 4-14 Trials of the slurry method with the GCTS CDSS set-up
Dr after
Water Dr after specimen
Fines Dr after top
Specimen Target Dr content of slurry placed transferred
content cap lowered
# (%) the slurry in the mold to cell
(%) (%)
(%) (%) (%)
1 35 50 21 71 80 85
2 35 50 24 66 77 81
3 35 50 30 70 81 83
4 35 50 25 69 79 83
5 35 50 20 73 79 82
104
prepared to almost zero relative density so as to account for the
volume change due to handling. After the specimen was
transferred and the top cap was applied, further densification was
achieved by tapping on the split mold when necessary.
105
Table 4-15 Stress-controlled CDSS tests on Monterey #0/30 sand
Post- Vertical
consolidation effective
B-value at relative CSR
Test # the end of stress
density
saturation (Dr) (σv')
(%) (kPa)
VT1 1.00 56 100 0.11
106
double amplitude shear strain was observed or until a predetermined number of
loading cycles were applied.
A typical stress-controlled test result is shown in Figure 4-15 for a sample
prepared to 56% relative density and subjected to a uniform cyclic shear stress
ratio, τ/σv′, of 0.11. The pore water pressure generation followed a smooth
progressive pattern until reaching a pore water pressure ratio, u*, of about 1.0,
corresponding to zero vertical effective stress. As the pore water pressure ratio
neared a value of 1.0, the previously small shear strains, γ, began to rapidly
increase, reaching values in excess of 10% within a few additional loading cycles.
In this test, liquefaction occurred (as defined by u* = 1.0) in approximately 30
cycles.
The response of a denser specimen (reconstituted to a relative density of
78%) is presented in Figure 4-16. The shear strain in this case, developed
gradually and liquefaction of the soil (as the defined by 5% double amplitude
strain) was observed after approximately 105 cycles of loading. The maximum
pore water pressure ratio during this test was around 0.9.
107
0.4
0.3
0.2
0.1
τ/σ 'v
0
-0.1
-0.2
-0.3
-0.4
0 5 10 15 20 25 30 35 40
Num ber of Cycles, N
0.8
ru (∆ u/σ 'v)
0.6
u*
0.4
0.2
0
0 5 10 15 20 25 30 35 40
Num ber of Cycles, N
20
15
10
5
γ, %
0
-5
-10
-15
-20
0 5 10 15 20 25 30 35 40
Num ber of Cycles, N
Figure 4-15 Undrained cyclic direct simple shear loading response of a specimen
at 56% relative density and CSR of 0.11
108
0.4
0.3
0.2
0.1
τ/σ 'v
0
-0.1
-0.2
-0.3
-0.4
0 10 20 30 40 50 60 70 80 90 100 110 120
Num ber of Cycles, N
0.8
ru (∆ u/σ 'v )
0.6
u*
0.4
0.2
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Num ber of Cycles, N
12
4
γ, %
-4
-8
-12
0 10 20 30 40 50 60 70 80 90 100 110 120
Num ber of Cycles, N
Figure 4-16 Undrained cyclic direct simple shear loading response of a specimen
at 78% relative density and CSR of 0.12
109
The stress-strain responses from both loose and dense specimens
discussed above are presented in Figure 4-17. The response of the loose specimen
indicates initial loops with relatively small strains and large shear stresses. As the
specimen softens the shear strains increase dramatically and result in loops such
as those Figure 4-17a. The response of the dense specimen shows that the
specimen does not soften as dramatically as in the loose case.
The measured responses from the stress-controlled cyclic direct simple
shear tests of seven different specimens are shown in Figure 4-18. The test results
are presented in terms of the applied CSR and the number of cycles to reach 6%
double amplitude shear strain. Four of the seven tests were conducted on loose
specimens reconstituted at relative densities between 42% to 61% while the other
three tests were performed on dense specimens that prepared to relative densities
of 70% to 78%. The variability in the specimen relative densities is due to the
water sedimentation sample preparation technique, which does not provide
precise control of specimen density. Nonetheless, the results from this study agree
well with those from De Alba et al. (1976).
110
30
25
20
15
10
5
(kPa)
0
-20 -15 -10 -5 -5 0 5 10 15 20
τ
-10
-15
-20
-25
-30
γ (%)
30
25
20
15
10
5
(kPa)
0
-8 -6 -4 -2 -5 0 2 4 6 8
τ
-10
-15
-20
-25
-30
γ (%)
111
0.40
Hazirbaba-Dr=42%
Hazirbaba-Dr=50-60%
Hazirbaba-Dr=70-78%
0.30 De Alba et al. (1976)
Dr=82%
Dr=68%
CSR (τ /σv')
Dr=54%
0.20
0.10
0.00
1 10 100 1000
4.9 SUMMARY
The details of the testing program for the assessment of pore water
pressure generation characteristics of sands and silty sands are presented in this
chapter. The components of the details are: (1) description of the testing system
used, (2) properties of the soils tested, (3) description of the strain-controlled
testing program, (4) testing procedures followed, (5) specimen preparation issues,
and (6) verification of the testing system.
The testing system used for this research is an NGI-type cyclic direct
simple shear system manufactured by Geotechnical Consulting and Testing
Systems (GCTS). The system is unique in that it allows for backpressure
112
saturation. The original design of the system was modified to meet the particular
testing needs for this research (e.g., internal instrumentation).
The soils used in this research program were: (1) Monterey #0/30 sand, (2)
aggregate sand, and (3) Sil-Co-Sil 52 non-plastic silt. Monterey #0/30 sand and
the aggregate sand were used as clean sands. Mixtures of Monterey #0/30 sand
and Sil-Co-Sil 52 non-plastic silt were used to form the silty sand specimens.
Index properties of these soils were presented in Section 4.5.
The testing program consisted of multiple series of strain-controlled cyclic
direct simple shear and cyclic triaxial tests. The testing series can be grouped as
follows:
1. CDSS tests on Monterey #0/30 sand. Three different relative
densities (Dr = 32%, 50%, 93%) were evaluated and a series of
tests with shear strain levels ranging from 0.005% to 0.3% were
performed at each relative density.
2. CTX tests on Monterey #0/30 sand. A series of tests were
performed at Dr = 50% with shear strains ranging from 0.01% to
0.3.
3. CDSS tests on silty sands at a constant relative density of 50%.
Tests were performed on specimens containing 5%, 10%, 15%,
and 20% fines. The shear strain levels for each of these test series
ranged from 0.03% to 0.3%.
4. CDSS tests on silty sands at a constant sand skeleton void ratio of
0.76. In this group, similar to the previous group, tests were carried
out at different fines contents (FC = 5%, 10%, 15%, 20%). The
shear strain levels were 0.03%, 0.1% and 0.3%.
5. CDSS tests on silty sands at constant void ratio of 0.59. This group
was limited up to 10% fines content. Therefore, two series of tests
113
(FC = 5%, 10%) were performed at shear strains of 0.03%, 0.1%,
and 0.3%.
6. CDSS tests on the aggregate sand at relative densities of 32-44%.
The tests were performed at two different initial vertical effective
stresses. The aim of this group was to provide pore water pressure
generation data on a sand that had been tested in situ.
Specimen preparation, particularly in the case of silty sand, was one of the
many challenges faced during this research. Monterey #0/30 sand specimens and
silty sand specimens were prepared by the moist undercompaction technique, and
the aggregate sand specimens were prepared by water sedimentation. It was found
that at fines content larger than 20%, the silty sand specimens tended to yield a
very contractive and metastable structure when prepared in a medium to loose
state.
Finally, the testing system was verified by a series of stress-controlled
tests on Monterey #0/30 sand specimens, which indicated good agreement with
De Alba et al. (1976).
114
Chapter 5: LABORATORY PORE WATER PRESSURE
GENERATION IN CLEAN SANDS
5.1 INTRODUCTION
This chapter presents measurements of excess pore water pressure from
clean sand specimens. Loose, medium, and dense sand specimens were subjected
to various uniform strain cycles under undrained conditions, and the excess pore
water pressure response for these specimens was recorded. For each considered
strain level, a freshly reconstituted specimen was used. Staged tests were avoided
to eliminate the staging effect on pore water pressure generation as discussed
previously in Chapter 4.
The effect of relative density on excess pore water pressure generation is
evaluated. Additionally, cyclic direct simple shear and cyclic triaxial testing
techniques are compared in terms of excess pore water pressure generation for
clean sands.
115
direct simple shear conditions. The details regarding the specimen preparation and
testing procedures can be found in the previous chapter. In terms of relative
densities, the loosely fabricated specimens were reconstituted at 32%, the medium
specimens were at 50%, and the dense specimens were at 93%. The initial vertical
effective stress, (σv'), was kept constant at 100 kPa for all of the tests.
Figures 5-1a and 5-1b show the results of the tests performed on loose
specimens. The excess pore water pressure ratio (u* = ratio of the measured excess
pore water pressure to the initial vertical effective stress, ∆u/σv') is presented
versus the number of loading cycles, (N), for shear strain levels of γ=0.03%,
γ=0.1%, and γ=0.3%. As can be seen from Figure 5-1a, the excess pore water
pressure ratio increased significantly during the first 10 loading cycles for all
shear strain levels, and continued its development at a slower rate with further
application of loading cycles. To better evaluate the history of excess pore water
pressure generation, and clearly portray the quick rise within the first 10 loading
cycles, the number of loading cycles is presented on a logarithmic scale in Figure
5-1b. This figure shows that at the smallest applied shear strain of 0.03%, the
excess pore water pressure ratio progressively increased from 2.7% at the end of
the first cycle of loading to 34% after 50 loading cycles. Similarly, the specimen
tested at 0.1% shear strain showed a progressive increase with increasing number
of loading cycles. The excess pore water pressure ratio at this strain level was
8.6% after the first cycle, and went up to approximately 62% at the end of 50
loading cycles. Initially the same trend was observed for the specimen subjected
to the largest shear strain of 0.3%. The first loading cycle in this case generated an
excess pore water pressure ratio of 20.6%. However, after 20 cycles of loading the
excess pore water pressure ratio curve flattens out and remains constant at about
95%. At this point the specimen has reached full liquefaction. It is clear that as the
shear strain level increases, more excess pore water pressure is generated for the
same number of loading cycles.
116
a) u* versus N in linear-linear space
Figure 5-2 Excess pore water pressure generation as a function of cyclic shear
strain for Monterey #0/30 sand at Dr=32%
118
To evaluate the rate at which the excess pore water pressure develops, the
increment between consecutive loading cycles can be taken into account. Figure
5-3 illustrates this approach conceptually. The initial excess pore water pressure
after the first cycle is defined as u*o. The increment in the excess pore water
pressure during the ith cycle, ∆u*i, is obtained from the difference between the
excess pore water pressure ratio at the end of the ith cycle (u*i) and at the end of
the i-1th cycle (u*i-1). Because pore water pressure generation appears to increase
logarithmically with number of loading cycle (Figure 5-1b), the rate of pore water
pressure generation will be defined as the increment between consecutive loading
cycles (∆u*i) divided by the logarithmic difference of the number of consecutive
loading cycles log(Ni/Ni-1).
Excess pore water pressure ratio, u*
* * *
∆u i = u i - u i-1
Ni-1 Ni
Number of Cycles, N
Figure 5-3 Conceptual representation of excess pore water pressure generation per
logarithmic cycle
119
Thus, the rate of excess pore water pressure generation is defined per
logarithmic cycle of N. Based on this definition, the pore water pressure
generation rates were calculated for three loading cycles (N=2, N=5, and N=10),
and plotted versus induced shear strains in Figure 5-4 for the loose sand. The rate
of pore water pressure generation increases both with increasing shear strains, and
with increasing number of cycles until full liquefaction is reached. After full
liquefaction is reached, the pore water pressure generation rate goes to zero
(Figure 5-1b).
The results of the tests from the medium dense specimens are shown in
Figures 5-5 and 5-6. In this series, tests were performed at four shear strain levels,
γ=0.005%, γ=0.03%, γ=0.1%, and γ=0.3%. The addition of a shear strain level of
0.005% was used to study the threshold strain. As can be seen from Figure 5-5a,
only minor excess pore water pressure (u* ≅ 1.5%) was generated even after 50
Figure 5-4 Excess pore water pressure generation rate per cycle for Monterey
#0/30 sand at Dr=32%
120
cycles of loading at the smallest shear strain of 0.005%. Thus, this strain level can
be regarded as almost non-destructive. At a larger strain of 0.03%, the excess pore
water pressure builds up in a progressive manner from about 1.6% after the first
loading cycle to 19% after 50 cycles of loading. The progress of the excess pore
water pressure for all shear strain levels is shown in Figure 5-5b. The excess pore
water pressure ratio for the case of 0.1% shear strain went from 6% at the end of
the first loading cycle to a value of 52% after 50 cycles of loading. As expected,
the largest pore pressure ratios were again obtained in the case of the largest
121
b) u* versus N for γ=0.005% to γ=0.3%
applied shear strain of 0.3%. The first loading cycle in this case generated an
excess pore water pressure ratio of approximately 15%. The progressive increase
with increasing number of loading cycles continued until N=30, where a
maximum of 90% pore water pressure ratio was reached. Further loading of the
specimen did not increase the pore water pressure ratio, thus the excess pore water
pressure history curve in this case flattens out and remains constant at 90%
between N=30 and N=50. The pore water pressure generation curves for the
medium dense sand are presented in Figure 5-6 and they suggest that the
threshold shear strain is close to 0.005%.
Figures 5-7 and 5-8 summarize the results of the tests performed on dense
sand specimens. The smallest shear strain level considered was γ=0.01%, and at
122
Figure 5-6 Excess pore water pressure generation as a function of cyclic shear
strain for Monterey #0/30 sand at Dr=50%
123
Figure 5-8 Excess pore water pressure generation as a function of cyclic shear
strain for Monterey #0/30 sand at Dr=93%
this strain value there was a small amount of excess pore water pressure generated
(u*< 4%). The excess pore water pressure ratio remained below 10% (u*= 8%) at
0.03% shear strain, and only reached a value of 30% at 0.1% strain. However, the
specimen subjected to the shear strain of 0.3% generated an excess pore water
pressure ratio of 20% after the first loading cycle, and the pore water pressure rise
at this strain level went up to 87% after 50 cycles of loading. Unlike the results
from the loose and medium dense sands, the excess pore water pressure history
curve of the dense sand specimen at this large strain level showed a continuous
increase during the application of 50 loading cycles. No flattening of the pore
water pressure history curve was observed. However, further cycling would have
most likely increased the pore water pressure ratio to 90-95% (full liquefaction),
where the curve would have flattened out. Figure 5-8 presents the pore water
124
pressure generation curves for the dense sand specimens. The curves for various
numbers of loading cycles follow a similar pattern to that observed in loose and
medium dense specimens. At the smallest shear strain level of 0.01% only a small
amount of pore water pressure was induced, therefore γ=0.01% can be considered
close to the threshold shear strain for the dense sand specimens.
125
Figure 5-9 Comparison of excess pore water pressure measurements from
Dr=32%, 50%, and 93% specimens at γ=0.03%
126
response from the medium specimen is very similar to that of the loose specimen,
although the actual values of excess pore water pressure ratio are larger in the
loose specimen for all loading cycles. Also, the ratio of pore water pressure
generation in the dense specimen seems to be more linear (in semi-log space) than
in both the medium and loose specimens.
Figure 5-11 displays the comparison at the largest shear strain level of
0.3%. The shape of the pore water pressure responses of the medium and loose
specimens are again similar. The main difference is that the excess pore water
pressure in the medium specimen is about 5% less than in the loose specimen.
This difference appears to be maintained approximately constant at all loading
cycles. Interestingly, the pore water pressure generation in the dense specimen
initially is the same as the loose specimen. However, because the rate of excess
pore water pressure generation is slower in the dense specimen, it quickly falls
below the looser specimen after four cycles of loading.
127
Dobry et al. (1982) reported similar trends of pore water pressure
generation at three different relative densities. Figure 5-12 presents the
measurements of pore water pressure generation for Monterey #0 sand at a shear
strain level of 0.1% from Dobry et al. (1982). Although the relative density
indicated a significant effect on the generation of pore water pressure, Dobry et al.
(1982) concluded that at numbers of cycles less than 10, the relative density effect
is only moderate and thus they presented a generalized pore water pressure
generation curve (Figure 2-9) that was applicable to various relative densities and
various specimen preparation techniques.
128
The relative density effect can further be examined from the pore water
pressure generation curves for 10 and 30 cycles of loading, as shown in Figures 5-
13a and 5-13b, respectively. These data indicate that at shear strains between
0.03% and 0.3%, specimens prepared to an initial relative density of 32%
generate noticeably larger excess pore water pressures than those reconstituted at
50% and 93% post consolidation relative densities. The threshold shear strain
does not appear to be significantly affected by the density, based on the results
from the Dr=50% and Dr=93% specimens. However, because a small strain test
was not performed on the loose specimen, this observation is not certain.
a) Comparison of pore water pressure generation curves for Dr=32%, 50%, and
93% specimens at N=10
Figure 5-13 Effect of relative density on pore water pressure generation curves
(continued)
129
b) Comparison of pore water pressure generation curves for Dr=32%, 50%, and
93% specimens at N=30
Figure 5-13 Effect of relative density on pore water pressure generation curves
130
Figure 5-14 Excess pore water pressure generation as a function of relative
density
Finally, the effect of the relative density on the initial excess pore water
pressure ratio and the rate of excess pore water pressure generation were
evaluated. This was done for three different number of loading cycles, N=1, N=2,
and N=10. As discussed earlier (see discussion of Figure 5-3), the initial excess
pore water pressure after the first loading cycle is defined as u*o. Based on this
approximation, Figure 5-15 presents a comparison of u*o versus shear strain for all
relative densities. At small strains (γ=0.03%), the values of u*o were almost
identical. At γ=0.1%, the values of u*o were only slightly affected by relative
density, with Dr=32% producing the largest value of u*o. Interestingly, at γ=0.3%,
u*o was the same for Dr=32% and 92%, with Dr=50% being slightly lower.
Figures 5-16 and 5-17 illustrate the variation of excess pore water pressure
generation rate per log cycle (% per log cycle) versus shear strain for N=2 and
131
Figure 5-15 Effect of relative density on initial excess pore water pressure after 1
loading cycle for Monterey #0/30 sand
Figure 5-16 Effect of relative density on excess pore water pressure generation
rate during the second cycle for Monterey #0/30 sand
132
Figure 5-17 Effect of relative density on excess pore water pressure generation
rate during the 10th cycle for Monterey #0/30 sand
N=10, respectively. Evaluation of these two plots shows that the excess pore
water pressure rate increases with increasing shear strain, and decreases with
increasing density. The difference is more pronounced at larger loading cycles
(N=10).
133
5.4 EFFECT OF TESTING SYSTEM ON PORE WATER PRESSURE
GENERATION
Almost all of the strain-controlled test results that have been reported in
the literature by various researchers are from cyclic triaxial tests. This may be due
to the fact that triaxial equipment is widely available and the test is relatively easy
to perform. However, as discussed earlier in Chapter 3, triaxial stress conditions
do not provide an ideal duplication of the in situ stress conditions during an
earthquake. A more realistic representation of in situ stresses is possible with the
use of a simple shear system. This section investigates the effect of test type on
the excess pore water pressure generation in clean sand, and examines the
relationship between cyclic direct simple shear and cyclic triaxial systems in
terms of excess pore water pressure generation.
134
cycles. Dramatic excess pore water pressure developments were observed at shear
strains of 0.1% and 0.3%. At 0.1% shear strain, the excess pore water pressure
ratio ranged from 18% at the end of the first cycle to 92% after 50 cycles of
loading. Finally, for the specimen tested at 0.3% shear, it took only 10 cycles to
reach an excess pore water pressure ratio of 95%. Further cyclic loading did not
induce additional pore water pressure, thus the pore water pressure ratio curve
flattened out and remained constant at a value of 95% until the test was
terminated after 50 cycles of loading.
Figure 5-18 Excess pore water pressure measurements from cyclic triaxial tests on
Monterey #0/30 sand specimens
135
Figure 5-19 shows the pore water pressure generation curves in terms of
excess pore water pressure ratio versus shear strain for a range of number of
cycles. These curves suggest that the threshold shear strain is on the order of
0.01%. For practical purposes the value of 0.01% can essentially be taken as the
threshold strain because no significant excess pore water pressure ratio was
measured at this strain level (u* ≤ 5%) for all values of N.
Figure 5-19 Pore water pressure generation curves obtained from cyclic triaxial
tests on Monterey #0/30 sand specimens
136
The pore water pressure generation curve at 10 cycles of loading is
compared to previously published data on clean sands by various researchers in
Figure 5-20. All of the pore water pressure generation curves shown in Figure 5-
20 are from triaxial tests. The results obtained from this study are in excellent
agreement with Dobry et al. (1982). This agreement is especially important
because Dobry et al. (1982) tested Monterey #0 sand, which is essentially the
same as the sand tested in this study. Also, the specimen preparation methods of
Dobry et al. (1982) and this study are identical, and the relative densities are
similar (i.e. 45% vs. 50%). Silver and Park (1976) reported results from Crystal
Silica #20 sand, and these results were found to be reasonably close to Dobry et
al. (1982) and this study. Of the published data, Erten and Maher (1995b) reported
Figure 5-20 Pore water pressure generation from cyclic triaxial tests and
previously published studies
137
relatively small excess pore water pressure ratios in loose Ottawa C-109 sand
specimens. The grain size distribution and index properties of Ottawa C-109 sand
are compared to those of Monterey #0/30 in Figure 5-21. The two sands display
very similar gradation curves and index properties and thus should display similar
cyclic test results.
5.4.2 Comparison of excess pore water pressure generation from cyclic direct
simple shear and cyclic triaxial tests
For stress-controlled cyclic liquefaction tests, it has already been
established that different types of tests yield different results (Finn et al. 1971,
Seed and Peacock 1971, Castro 1975). Relationships have been proposed between
the results of stress-controlled cyclic triaxial and cyclic direct simple shear tests.
The following equation is usually used to relate the results from the two types of
tests:
100
90 Monterey #0/30
Percent passing (%)
80 Ottaw a C-109
70 Monterey #0/30:
60 Cu=1.6
Cc=1.1
50 Class=SP
40
Ottawa C-109:
30 Cu=2
20 Cc=1.12
10 Class=SP
0
10 1 0.1 0.01
Grain size (mm)
Figure 5-21 Grain size distribution and index properties of Monterey #0/30 and
Ottawa C-109 sands
138
(CSR) CDSS = α (CSR) CTX …………………………………………….…(5-1)
where (CSR)CDSS = the cyclic stress ratio (τ/σv') to cause liquefaction in a cyclic
direct simple shear test, (CSR)CTX = the cyclic stress ratio (σd/2σ3′) to cause
liquefaction in a cyclic triaxial test, and α is a conversion factor. The value of α is
typically 0.7 for normally consolidated sand (Finn et al. 1971, Seed and Peacock
1971, Castro 1975).
The difference between the results from cyclic triaxial test and cyclic
direct simple shear test often is ascribed to the different stress states in the tests,
although the stress paths are also different. The state of stress in both systems is
depicted in Figure 5-22. As discussed in detail in Chapter 3, the stress and strain
on the horizontal plane of a simple shear specimen can be measured directly for a
given vertical effective stress. Yet, the entire state of stress within the specimen is
not completely defined, because Ko is not known and therefore the horizontal
effective stress, (σh'), is not known. Also, the lack of complementary shear
stresses on the vertical sides of the simple shear specimen complicates the stress
state. In a triaxial specimen, the vertical cyclic stress and strain are measured
during the test. The corresponding shear stress is computed on the 45o plane and
the shear strain is computed using elastic theory and Poisson’s ratio. The main
difference in the states of stress is that the cyclic direct simple shear test has an
initial, anisotropic state of stress (approximate Ko conditions) and the cyclic
triaxial test has an initial, isotropic state of stress.
139
σv' σv' σv'
τxy τxy
γxy
σ h' σ h'
εv
εv
σh'=σv' σh'=σv'
Figure 5-22 State of stress for specimens under cyclic direct simple shear and
cyclic triaxial test conditions (modified from Park and Silver, 1975)
140
To evaluate the effect of test type and initial state of stress on the
generation of excess pore water pressure, the results of cyclic direct simple shear
tests for the case of 50% relative density are compared to those obtained from
cyclic triaxial tests at the same relative density. Figure 5-23 displays the
comparison of the pore water pressure generation histories at shear strains of
0.03%, 0.1%, and 0.3%. The cyclic direct simple shear test results display less
pore water pressure generation than the cyclic triaxial test results, particularly at
large strains. To better evaluate the responses from the two testing systems, pore
water pressure generation curves at N=10 were produced and they are presented
in Figure 5-24. At cyclic shear strains between 0.005% and 0.03%, there is only
Figure 5-23 Comparison of cyclic direct simple shear and cyclic triaxial
measurements
141
a minor difference in the generation of the excess pore water pressure measured,
and thus the pore water pressure curves are close to each other. At relatively
larger strains (>0.03%), the cyclic direct simple shear results display smaller pore
water pressure values than the triaxial results.
As discussed earlier, the stress conditions in the simple shear and triaxial
systems are different (Figure 5-22). Finn et al. (1971) pointed out that the vertical
stress applied in the simple shear test is not a good index of the confinement of
the soil specimen. Therefore, when comparing the results of cyclic direct simple
shear and cyclic triaxial tests, the different initial confining stress conditions
should be taken into account.
Figure 5-24 Pore water pressure generation curves from cyclic direct simple shear
and cyclic triaxial tests at N=10
142
A more complete comparison may be achieved if the excess pore water
pressures are normalized by the mean effective confining stress, instead of the
vertical effective stress. The mean effective confining stress is defined as:
1 + 2K o
σm '= σ v ' ………………………………………………...……(5-2)
3
where Ko = the lateral earth pressure coefficient; and σv' = the vertical effective
stress. In a triaxial test, an isotropic stress state exists and Ko = 1, thus the mean
effective confining stress is equal to the initial effective cell pressure (σm' = σ3').
For the simple shear test, where the stress state is anisotropic, Ko may be assumed
to be about 0.5. Thus, the initial mean effective stress in the simple shear test
becomes:
2
σm '= σ v ' …………………………………………………………….(5-3)
3
The cyclic direct simple shear test pore water pressure data were replotted as a
function of the mean confining stress, σm', for the case of Ko = 0.5 along with the
triaxial data in Figure 5-25. The values of excess pore water pressure ratios
obtained from the two testing systems are now closer, particularly during the first
10 loading cycles. However, these new data suggest that the largest shear strain of
0.3% generated an excess pore water pressure approximately 1.3 times the mean
confining stress (u* ≅130%) in the simple shear specimen at the end of 50 loading
cycle. A pore ratio of 130% is not reasonable. However, this final pore water
pressure ratio may suggest that Ko increases as liquefaction is approached and
thus normalizing the pore water pressure by the initial mean effective stress is not
correct and cannot explain the difference between cyclic triaxial and cyclic direct
simple shear results. The difference may then be a result of the different stress
paths induced in the tests.
143
Figure 5-25 Comparison of cyclic direct simple shear and cyclic triaxial
measurements in terms of mean confining stress
5.5 SUMMARY
The excess pore water pressure generation in clean sand specimens was
evaluated using strain-controlled, undrained, cyclic direct simple shear and cyclic
triaxial tests. The cyclic direct simple shear tests were performed on three
different groups of sand specimens. These are: (1) specimens prepared at 32%
relative density (loose), (2) specimens prepared at 50% relative density (medium),
and (3) specimens prepared at 93% relative density (dense). The cyclic triaxial
tests were conducted on a set of specimens that were reconstituted at 50% relative
density.
144
The measurements of excess pore water pressure from the cyclic direct
simple shear tests were presented separately for each group of specimens. Then,
the effect of density on the generation of excess pore water pressure was
evaluated by comparing the responses of loose, medium, and dense sand
specimens. This comparison showed that, in general the excess pore water
pressure generation in the specimens prepared at 32% relative density was
consistently the largest for shear strain levels between 0.03% and 0.3%. However,
it was observed that the pore water pressure generated after the first cycle of
loading was not significantly affected by relative density, but that the relative
density has a larger effect on the rate of pore water pressure generation. Similar
trends of pore water pressure generation at various densities were observed by
Dobry et al. (1982).
The results of strain-controlled cyclic triaxial tests were assessed and
verified with previously published studies. A comparison between the
measurements of excess pore water pressures from cyclic triaxial tests and cyclic
direct simple shear tests was made to investigate the effect of the testing system
on the generation of pore water pressures. The excess pore water pressure
generation obtained from the cyclic direct simple shear tests was found to be
somewhat lower than that from the cyclic triaxial tests for a given shear strain
level. This result could not be ascribed to the initial stress state, but may be
caused by the different stress paths induced in the two test types.
145
Chapter 6: EFFECT OF NON-PLASTIC FINES ON PORE
WATER PRESSURE GENERATION
6.1 INTRODUCTION
The role of fines content on the generation of excess pore water pressure
in liquefiable soils is the focus of this chapter. Numerous laboratory studies have
been performed by various researchers to clarify the effect of fines content on the
cyclic resistance of soils. However, a review of these studies shows that no clear
conclusion can be drawn as to how altering the fines content influences the cyclic
resistance, and in particular the pore water pressure generation characteristics, of
soil. This study aims to evaluate the pore water pressure generation in sands with
various non-plastic fines contents. Strain-controlled, undrained, cyclic direct
simple shear tests were performed on clean sands and sands containing up to 20%
fines. Three different controlling index parameters were considered during the
preparation of clean sand and silty sand specimens. These are: (1) relative density,
(2) sand skeleton void ratio, and (3) overall void ratio. The test results from each
of these categories are presented, and evaluated in terms of pore water pressure
generation characteristics.
146
forces (Mitchell 1993). A silty sand can be considered as a delicate matrix
composed of two submatrices: (1) a coarse-grain matrix, consisting of sand
particles, and (2) a fine-grain matrix, consisting of silt particles (Thevanayagam
1998). The interaction between these two matrices and the arrangement of the
soil particles within each matrix are critical to understanding the behavior of silty
sands.
The effect of fines on the minimum void ratio, which is defined as the
void ratio at the densest possible packing of soil particles without crushing, has
been theoretically and experimentally studied (McGeary 1961, Lade at al. 1998).
The theoretical variation of minimum void ratio with fines content for spherical
particles is illustrated schematically in Figure 6-1. Points A and C represent the
minimum void ratios of soils with no fines and 100% fines, respectively. Fine
particles are filling the voids that are formed by the large particles between points
A and B. The overall volume remains constant between these two points while the
weight of the soil increases as the fines fill in the void space. Point B is reached
when the voids formed by the large particles are completely filled with fines.
Additional fines beyond point B push the large particles apart. Due to the
tendency of the smaller grains to pack with higher void ratios (Lade et al. 1998),
the void ratio continuously increases with increasing fines content between points
B and C.
In an effort to search for the effect of particle size ratio (ratio of the particle
diameter of fines to that of large particles) on the minimum void ratio, McGeary
(1961) performed laboratory studies on steel shot with various diameters ranging
between 0.16 mm and 3.15 mm. The large particle diameter was kept constant at
3.15 mm, and six different smaller particle diameters ranging between 0.91 mm
and 0.16 mm were used to obtain different particle size ratios. The large particles
were first deposited into a container at their densest configuration. The smaller
particles were then added while vibrating the container. This process was repeated
147
for the six different particle size ratios and it was found that at all fines contents
smaller fine grain diameters result in smaller void ratios (Figure 6-2).
Figure 6-1 Theoretical variation of the minimum void ratio with fines content
(Yamamuro and Covert 2001)
148
Figure 6-2 Effect of fines content and fine particle diameters on minimum void
ratio (McGeary 1961)
149
The maximum void ratios obtained in mixtures of large and small spheres
follow a similar pattern to that obtained for the minimum void ratios (Lade et al.
1998). This is because the small particles initially fill the voids between large
particles and thus cause a decrease in the maximum void ratio, while further
increases in fines spreads the large particles apart and thus increases the
maximum void ratio.
The variations in the maximum and minimum void ratios with fines
content for the soil mixtures used in this research are presented in Figure 6-3.
Monterey #0/30 sand was used as the sand, and mixed with Sil-Co-Sil 52 non-
plastic silt at various percentages to form silty sand specimens. The grain size
distributions and properties of these soils were presented previously in Chapter 4.
The minimum void ratios were determined by using the ASTM D4253
procedure (vibratory table) and also from Modified Proctor (ASTM D1557)
compaction tests. In the ASTM D4253 procedure, the minimum void ratios were
achieved by densifying dry soil in a standard mold with a volume of 2830 cm3
using a vertically vibrating table. The two different procedures yielded identical
results between 0% and 20% fines content. Smaller minimum void ratios were
obtained from the Modified Proctor compaction tests at fines contents larger than
20%. It is important to note that the ASTM D4253 recommends a maximum value
of 15% fines content. Therefore, it is not surprising that the minimum void ratios
from the vibratory table test are larger than the Modified Proctor values at high
fines contents.
For the maximum void ratios, ASTM D4254 was followed. Although this
standard is not explicitly recommended for soils with significant fines, two of the
three methods suggested (namely Method B and Method C) were found to be
applicable to soils containing fines. In Method B, a thin-wall cylindrical tube with
150
Figure 6-3 Maximum and minimum void ratios versus fines content for Monterey
#0/30 sand and Sil-Co-Sil 52 non-plastic silt mixtures
151
open ends is placed in a standard mold and filled cautiously with dry soil. Then,
the tube is slowly lifted up, thus the soil flows out and fills the mold. The soil is
then leveled with the top of the mold. The minimum density (maximum void
ratio) is determined from the weight of the soil and the volume of the mold. In
Method C, a 2000-ml graduated glass cylinder is filled with soil of a known
weight. Then, the open end of the cylinder is closed with a rubber stopper, and the
cylinder is inverted so as to allow for the loosest particle arrangements. The
minimum density (maximum void ratio) is determined from the weight and
volume of the soil. The values of the maximum void ratios from Methods B and C
were similar up to 35% fines content. At higher fines contents, Method C
produced larger maximum void ratios, as can be seen in Figure 6-3.
Although the index densities (emax and emin) were determined for a range of
fines content between 0 and 100%, this investigation was limited to 20% fines
content due to the extreme contractive tendency of the specimens at higher fines
contents, as discussed in detail in Chapter 4. For the range that was considered
during this study (FC=0-20%), the minimum and maximum void ratios obtained
from the various methods were found to be very similar.
152
e max − e
Dr = ⋅ 100 …………………………………………………...(6-1)
e max − e min
The concept of relative density for silty sands may be subjected to some
criticism due to the difficulties in obtaining maximum and minimum void ratios,
particularly for sands with more than 15% fines content. However, reproducible
values of maximum and minimum void ratios, such as those presented in Figure
6-3, can be obtained if the ASTM procedures (D4253 and D4254) are followed
with extra caution. Thus, repeatable relative density values are possible even for
sands with high fines contents.
To evaluate the effect of fines on the generation of pore water pressure at a
constant relative density, a value of 50% was selected to represent a medium
dense state. Hence, all of the specimens tested under this category were prepared
to a 50% post-consolidation relative density.
The sand skeleton void ratio has been used by various researchers (e.g.,
Kuerbis et al. 1988; Polito and Martin, 2001) as a controlling index parameter in
investigations of the effect of fines on liquefaction resistance. The concept of sand
skeleton void ratio was discussed earlier in Chapter 2. If the sand skeleton void
ratio of a silty sand exceeds the maximum void ratio of the clean sand, the sand
matrix exists with a ratio greater than it could attain in the absence of fines, which
means that the sand particles are, on average, no longer in contact, and thus the
behavior is no longer governed by the sand skeleton (Salgado et al. 2000).
Another index parameter usually used in the evaluation of fines effect on
the cyclic behavior of soils is the overall void ratio. This concept considers the
voids formed by both sand particles and fines. Different investigations (i.e. Chang
et al. 1982, Koester 1994, Polito and Martin 2001) have been performed at
constant overall void ratios and varying fines content to evaluate the cyclic
resistance of liquefiable soils, yet none of these studies focused directly on the
generation of pore water pressure in strain-controlled tests.
153
The selection of the sand skeleton void ratio and the overall void ratio was
found to be a challenging task. For the selection of the sand skeleton void ratio,
the relationship between sand skeleton void ratio (es) and overall void ratio (e)
(Equation 2-2) for different values of fines content (FC) was plotted (Figure 6-4).
Equation 2-2 indicates that for a constant fines content, the sand skeleton void
ratio increases linearly with the overall void ratio. The maximum and minimum
void ratios for the clean sand (Monterey #0/30) are represented by the horizontal
dashed lines. In choosing a constant sand skeleton void ratio for testing over a
range of fines content, achievable overall void ratios must be considered (e.g. emin,
emax for each FC, as shown in Figure 6-3). Thus, the relationships for each FC in
Figure 6-4 are plotted only between emin and emax for that FC. A sand skeleton
void ratio of 0.76 provided achievable overall void ratios for FC = 5 to 20%. If a
horizontal line representing this sand skeleton void ratio is drawn, the intersection
of this line and each fines content line corresponds to the overall void ratio at that
fines content. Similarly, a vertical line representing a constant overall void ratio
intersects each fines content line at the sand skeleton void ratio of that fines
content. The selection of the overall void ratio required some initial trials. It was
found that overall void ratios corresponding to sand skeleton void ratios larger
than the maximum index void ratio of the clean sand lead to highly contractive
and metastable specimens. As a result, a value of 0.59 was selected as the
constant overall void ratio to be used for testing.
154
es=0.76
e=0.59
Figure 6-4 Sand skeleton void ratio versus overall void ratio for various fines
contents
155
6.4 TESTS AT CONSTANT RELATIVE DENSITY
Clean sand specimens and sand specimens with 5%, 10%, 15%, and 20%
fines content were prepared by the moist undercompaction technique at a constant
post-consolidation relative density of 50%. These specimens were tested under
strain-controlled, undrained, cyclic direct simple shear conditions. The details
regarding specimen preparation, consolidation, and testing procedure can be
found in Chapter 4.
Three different shear strain levels, γ=0.03%, γ=0.1%, and γ=0.3%, were
considered. Figure 6-5 presents the excess pore water pressure histories at the
smallest induced shear strain of 0.03%. The clean sand specimen and the
specimen prepared with 5% fines content experienced larger pore water pressures
than those with higher fines contents. During the first 10 loading cycles, the
excess pore water pressure was slightly higher in the specimen with 5% fines
Figure 6-5 Pore water pressure histories at constant Dr=50% and γ=0.03% for
various fines contents
156
content than that in the clean sand specimen. The excess pore water pressure ratio
for these two specimens progressively increased from approximately 2% at the
end of the first cycle to about 8% at the end of 10 loading cycles. Further loading
resulted in more pore water pressure in the clean sand specimen. The pore water
pressure ratio reached 19% after 50 loading cycles for the clean sand specimen
while it reached only 11% for the specimen with 5% fines. The specimens with
10%, 15%, and 20% fines content did not undergo significant pore water pressure
during the first 10 loading cycles (u* < 2%). Interestingly, a noticeable increase in
pore water pressure occurred in the specimen with 20% fines content between
N=10 and N=50. The pore water pressure ratio increased from approximately 2%
at N=10 to 11% at N=50. The specimens with 10% and 15% fines content
experienced the smallest pore water pressure ratios, less than or equal to 4% after
50 loading cycles.
Figure 6-6 shows the results of the tests performed at a larger shear strain
level of 0.1%. The pore water pressure ratio at the end of the first loading cycle
remains below 10% for all of the specimens at this strain level. As expected, pore
water pressures increase with increasing number of cycles. Larger pore water
pressure ratios were obtained for the clean sand specimen than any of the other
specimens during the first 10 loading cycles. After N=10, the specimen with 20%
fines content continuously developed the largest pore water pressure and reached
a pore water pressure ratio of 61% at N=50. The pore water pressure ratios for the
specimens with 10% and 15% fines content are the smallest for all loading cycles.
The specimen with 5% fines content followed a trend that is between the clean
sand specimen and the specimens with 10% and 15% fines content.
The pore water pressure histories at the largest shear strain of 0.3% are
displayed in Figure 6-7. After the first loading cycle at this strain level, the
specimens with 10% and 15% fines content generated a pore water pressure ratio
157
Figure 6-6 Pore water pressure histories at constant Dr=50% and γ=0.1% for
various fines contents
Figure 6-7 Pore water pressure histories at constant Dr=50% and γ=0.3% for
various fines contents
158
of about 9%, and the clean sand specimen and the specimens with 5% and 20%
fines content experienced a ratio of approximately 18%. As loading continued, the
pore water pressures progressed in each specimen. The smallest pore water
pressures were recorded in the specimen with 10% fines content, whereas the
largest pore water pressures were generated in the specimen with 20% fines
content.
The pore water pressure generation curves (u* versus γ) at 50% constant
relative density for specimens with various fines content up to 20% are presented
in Figures 6-8 and 6-9. It appears that the fines content has a noticeable influence
on the threshold shear strain. The specimens with fines content larger than 5%
showed almost zero pore water pressure generation (N=1 and 5 in Figure 6-8,
N=10 and 30 in Figure 6-9) at shear strain of 0.03%. As previously discussed in
Chapter 5, the threshold shear strain for the 50% relative density clean sand
specimens was found to be between 0.005% and 0.01%, and the clean sand
specimens generated appreciable pore water pressure at a shear strain level of
0.03% (Figures 6-8 and 6-9). The specimen with 5% fines content also showed an
appreciable amount of pore water pressure generation at a shear strain of 0.03%,
and appears to behave more similar to the clean sand specimen in terms of pore
water pressure generation at this strain level. Thus, for specimens prepared at a
constant relative density the immediate effect of fines content on the generation of
pore water pressure was observed at fines content larger than 5% where the pore
water pressures at relatively small strains (i.e. γ=0.03%) were significantly
reduced. This may be a result of an increase in the threshold shear strain.
In general, the pore water pressure generation curves in Figures 6-8 and 6-
9 show that the pore water pressure in clean sand will be larger than that in silty
sand (at the same relative density) containing up to 15% fines. It was also
observed that, depending on the number of loading cycles considered, the pore
159
Figure 6-8 Pore water pressure generation curves at constant Dr=50% and N=1
and 5 for various fines contents
160
Figure 6-9 Pore water pressure generation curves at constant Dr=50% and N=10
and 30 for various fines contents
161
water pressure in clean sand may be equal to, or less than, that in sand containing
20% fines (i.e. Figure 6-9).
Figures 6-10 to 6-12 show the effect of fines, at a constant relative density,
on the generation of excess pore water pressure. Each figure presents a different
strain level and illustrates the effect of fines at three different numbers of loading
cycles; N=5, N=10, and N=30. Figure 6-10 demonstrates the variation in excess
pore water pressure ratios with fines content at the shear strain of 0.3%. It is
evident that an increase in fines content up to 10% results in a decrease in excess
pore water pressure, and that a further increase in fines content between 10% and
20% causes an increase in excess pore water pressure ratio. Hence, the 10% fines
content appears to be the level at which minimum excess pore water pressure
develops.
A similar trend was also observed at a smaller shear strain of 0.1%, as
displayed in Figure 6-11. As the fines content increases up to 10%, a significant
decrease occurs in the excess pore water pressure. An additional increase in fines
content leads to an increase in excess pore water pressure, and at 20% fines
content and large numbers of cycles the pore water pressure may even be larger
than for clean sands. In Figure 6-11, the data point for N=30 at 20% fines content
shows such a condition.
At the smallest shear strain of 0.03%, there appears to be no significant
difference between the excess pore water pressures from clean sand and sand with
5% fines content, for N=5 and N=10, as shown in Figure 6-8. The excess pore
water pressure ratio decreased with further increases in fines content, and
remained nearly constant between 10% and 20% fines content for N=5 and N=10.
The trend for N=30 is analogous to those presented earlier at larger strain levels.
162
Figure 6-10 Excess pore water pressure ratio versus fines content at Dr=50% and
γ=0.3%
163
Figure 6-11 Excess pore water pressure ratio versus fines content at Dr=50% and
γ=0.1%
164
Figure 6-12 Excess pore water pressure ratio versus fines content at Dr=50% and
γ=0.03%
165
The sand skeleton void ratios and overall void ratios are included with
each data point in Figures 6-10 to 6-12. The sand skeleton void ratio increased
with increasing fines content and ranged from 0.71 for clean sand specimens to
0.90 for sand specimens containing 20% fines. At FC = 10% and 20%, the sand
skeleton void ratio is close to or greater than the maximum void ratio of the clean
sand (emax,clean sand=0.85). Thus, the increased pore water pressure in these
specimens may indicate the transition from behavior dominated by the sand
skeleton to behavior dominated by the fines skeleton. The overall void ratio
decreased with increasing fines content and varied from 0.71 for clean sand
specimens to 0.52 for sand specimens containing 20% fines.
166
Figure 6-13 Pore water pressure histories at constant es=0.76 and γ=0.03% for
various fines contents
generated significantly less pore water pressure with increasing number of cycles.
The excess pore water pressure ratio was less than 4% after 10 loading cycles, and
around 7% or less after 50 loading cycles in all of the silty sand specimens. The
smallest pore water pressure generation was observed in the specimen containing
20% fines. The specimens with 10% and 15% fines content appear to have nearly
identical pore water pressure ratios at all loading cycles, which are slightly larger
than the specimen containing 20% fines yet smaller than the specimen containing
5% fines.
The pore water pressure histories at a constant sand skeleton void ratio of
0.76 and shear strain of 0.1% are presented in Figure 6-14. The clean sand
specimen generated the largest pore water pressure at this strain level at all
loading cycles, as it did for the previous strain level. The smallest pore water
pressure occurred in the specimen containing 10% fines. After the first loading
167
Figure 6-14 Pore water pressure histories at constant es=0.76 and γ=0.1% for
various fines contents
cycle, the excess pore water pressure ratio was approximately 9% in the clean
sand specimen, and about 4% or less in the silty sand specimens. The specimens
containing 5% and 15% fines produced almost equal pore water pressure ratios at
all loading cycles. At the end of 50 loading cycles, the clean sand specimen
generated an excess pore water pressure ratio of 62% while the silty sand
specimens experienced a ratio of 25% or less.
Figure 6-15 illustrates the pore water pressure histories at the largest
considered shear strain level of 0.3%. The general trends at this strain level are
similar to those presented in Figure 6-14 for γ=0.1%. The largest pore water
pressure generation was observed in the clean sand, and the silty sand specimens
developed less pore water pressure at all loading cycles. The values of pore water
pressure ratio obtained from the specimens containing 5%, 15%, and 20% were
168
Figure 6-15 Pore water pressure histories at constant es=0.76 and γ=0.3% for
various fines contents
close to each other while the specimen containing 10% fines yielded the lowest
values of pore water pressure ratio at all loading cycles.
The pore water pressure generation curves at a constant sand skeleton void
ratio for clean sand and sand containing fines up to 20% are shown in Figure 6-16
for N=1 and 5, and in Figure 6-17 for N=10 and 30. Both of these figures imply
that at a constant sand skeleton void ratio the fines content increases the threshold
shear strain. At N=1 in Figure 6-16, the smallest shear strain of 0.03% resulted in
about a 3% excess pore water pressure ratio for the clean sand whereas nearly
zero pore water pressure was measured in the silty sand specimens. Similarly, at
N=5 the clean sand specimen generated a pore water pressure ratio of about 10%
while the pore water pressure ratios from the silty sand specimens were less than
2%. This trend was maintained at larger numbers of cycles (N=10, N=30) as
presented in Figure 6-17. Therefore, it can be concluded that at a constant sand
169
Figure 6-16 Pore water pressure generation curves at constant es=0.76 and N=1
and 5 for various fines contents
170
Figure 6-17 Pore water pressure generation curves at constant es=0.76 and N=10
and 30 for various fines contents
171
skeleton void ratio, the presence of fines decreases the pore water pressure
generation and leads to an increase in the value of the threshold shear strain when
compared to clean sand. Evaluation of the pore water pressure generation curves
at larger shear strains shows a similar trend to that observed at γ=0.03%. The
responses from the clean sand specimens represent the upper bound of the pore
water pressure generation curves presented in Figures 6-16 and 6-17. The lower
bound was obtained from the specimens containing 10% fines. The specimens
containing 5%, 15%, and 20% fines yielded values of pore water pressure close to
one another, which fall between the clean sand values and the values obtained
from the specimens containing 10% fines.
To better portray the effect of fines content on the generation of pore
water pressure at a constant sand skeleton void ratio, the excess pore water
pressure ratio was plotted versus fines content and presented for each strain level
in Figures 6-18 to 6-20. For each fines content, the values of relative density and
overall void ratio are shown. Note that the relative density tends to increase with
fines content and the overall void ratio decreases. It is important to recall that all
of these specimens have the same sand skeleton void ratio, which is less than the
maximum void ratio of the sand, and thus the behavior is dominated by the soil
skeleton.
At the largest shear strain of 0.3% (Figure 6-18) there is a continuous
decrease in the excess pore water pressure ratio with increasing fines content up
to 10%. The pore water pressure ratio shows an increase between 10% and 15%
fines content and then a slight decrease between 15% and 20% fines content.
Figure 6-19 presents a similar trend at a smaller shear strain (γ=0.1%). This
decrease in pore water pressure generation most likely can be attributed to the
increase in relative density and decrease in void ratio. At the smallest shear strain
level of 0.03%, as displayed in Figure 6-20, the pore water pressure ratio
decreases with increasing fines content up to 10%, and appears to remain
172
approximately constant thereafter. At this small strain level, pore water pressure
generation is controlled mostly by the proximity to the threshold strain, and thus
the large fines content specimens do not show an increase in pore water pressure
generation.
Figure 6-18 Excess pore water pressure ratio versus fines content at es=0.76 and
γ=0.3%
173
Figure 6-19 Excess pore water pressure ratio versus fines content at es=0.76 and
γ=0.1%
174
Figure 6-20 Excess pore water pressure ratio versus fines content at es=0.76 and
γ=0.03%
175
6.6 TESTS AT CONSTANT OVERALL VOID RATIO
In this testing series, the overall void ratio was held constant and the pore
water pressure generation in clean sand and in sand containing up to 10% fines
was evaluated. Because attempts at higher fines content resulted in very
contractive and metastable specimens, this testing series was limited to 10% fines
content. The issue of contractive specimens at relatively high fines content was
discussed in detail in Chapter 4. The selection of applicable overall void ratio was
presented earlier in Section 6.3.
The pore water pressure histories from clean sand specimens and sand
specimens containing 5% and 10% fines at a constant overall void ratio of 0.59
are shown in Figures 6-21 to 6-23.
Figure 6-21 illustrates the pore water pressure histories at the smallest
considered shear strain of 0.03%. The largest pore water pressure ratios were
obtained from the clean sand specimen with about 1% at the end of first loading
cycle, and 8% after 50 loading cycles. The specimen containing 5% fines
generated smaller pore water pressure ratios; less than 1% at the end of the first
loading cycle and 5% at the end of 50 loading cycles. The pore water pressure
ratios obtained from the specimen containing 10% fines progressed from 0.5%
after the first loading cycle to only 3% after 50 loading cycles.
At a larger shear strain level of 0.1%, as seen in Figure 6-22, larger pore
water pressures were recorded, yet the same trends as in the previous shear strain
level (γ=0.03%) were observed; the largest pore water pressure generation was
recorded in the clean sand specimen, and the smallest pore water pressure was
from the specimen containing 10% fines.
The pore water pressure histories at the shear strain of 0.3% are shown in
Figure 6-23. The pore water pressure generation in the clean sand specimen
yielded a pore water pressure ratio of 20% at N=1, and a ratio of about 90% at
176
Figure 6-21 Pore water pressure histories at constant e=0.59 and γ=0.03% for
various fines contents
Figure 6-22 Pore water pressure histories at constant e=0.59 and γ=0.1% for
various fines contents
177
Figure 6-23 Pore water pressure histories at constant e=0.59 and γ=0.3% for
various fines contents
N=50. The values of pore water pressure ratio obtained from the silty sand
specimens were again lower than those recorded for the clean sand specimen, and
both of the silty sand specimens at this strain level exhibited almost equal pore
water pressures during the first 10 loading cycles (u*=10% at N=1; u*=30% at
N=10). Further cycling resulted in larger pore water pressure ratios in the
specimen containing 10% fines than in the specimen containing 5% fines. It
should be noted that this was not the case at the smaller strains (γ=0.03%,
γ=0.1%) presented earlier.
The pore water pressure generation curves (u* versus γ) at a constant
overall void ratio of 0.59 are presented for N=1 and 5 in Figure 6-24, and for
N=10 and 30 in Figure 6-25. At the smallest shear strain of 0.03%, for N=1 and
N=5, clean sand and silty sand (5% FC, and 10% FC) generated only a small
178
Figure 6-24 Pore water pressure generation curves at constant e=0.59 and N=1
and 5 for various fines contents
179
Figure 6-25 Pore water pressure generation curves at constant e=0.59 and N=10
and 30 for various fines contents
180
amount of pore water pressure (u* < 4%), as can be seen in Figure 6-24. As the
shear strain increases, the pore water pressure generation in the clean sand
specimen becomes larger than that in both of the silty sand specimens. Although
the two silty sand specimens were prepared with different fines content (5% FC,
10% FC), their pore water pressure generation curves were found to be close to
each other, and they can almost be represented by a single pore water pressure
generation curve. Similar trends were observed at larger number of cycles, with
larger pore water pressure generation values (Figure 6-25).
To evaluate the effect of fines content on the generation of pore water
pressure at constant overall void ratio, the pore water pressure ratio was plotted
against the fines content in Figures 6-26 to 6-28 for all three shear strain levels
that were considered.
The effect of fines content at the largest shear strain level of 0.3% is
shown in Figure 6-26. The pore water pressure ratio at all loading cycles
(N=5,10,30) decreases first between the clean sand and the sand containing 5%
fines, and then remains approximately constant up to 10% fines content. Thus, at
a constant void ratio of 0.59, the fines cause less pore water pressure generation in
silty sand specimens, with comparison to the clean sand specimen, when
subjected to a shear strain of 0.3%. At smaller shear strain values of 0.1% and
0.03%, the pore water pressure ratio decreases with increasing fines content, yet
the decrease between clean sand and sand containing 5% fines is more significant
than what occurs between 5% and 10% fines content (Figures 6-27 and 6-28). The
results in Figure 6-26 through 6-28 are consistent with the other test series in that
pore water pressure generation decreases with increasing fines content up to 10%.
However, the results in these figures are also somewhat surprising because they
show decreasing pore water pressure generation as the relative density decreases.
181
Figure 6-26 Excess pore water pressure ratio versus fines content at e=0.59 and
γ=0.3%
182
Figure 6-27 Excess pore water pressure ratio versus fines content at e=0.59 and
γ=0.1%
183
Figure 6-28 Excess pore water pressure ratio versus fines content at e=0.59 and
γ=0.03%
184
6.7 PORE WATER PRESSURE GENERATION RATE OF SANDS AND SILTY
SANDS
In this section, the excess pore water pressure generation rate per log cycle
is evaluated for sands and silty sands. The discussion about the concept of the rate
of excess pore water pressure generation can be found in Chapter 5.
The rate per log cycle during N=2 and N=10 at a constant relative density
of 50% is presented in Figure 6-29. At all strain levels, the smallest rates were
obtained from the silty sand specimens with 10% fines during both N=2 and
N=10. The largest rate (~50% per log cycle) during N=2 was obtained from the
specimen with 20% fines content at 0.3% shear strain. Interestingly, the specimen
with 20% fines content had approximately the same pore water pressure rates at
0.1% and 0.3% shear strain levels during N=10. These results are similar to the
observations of absolute pore water pressure, the FC = 10% specimens displayed
the smallest pore water pressure generation.
Figure 6-30 shows the rate of pore water pressure generation in sand and
silty sand specimens prepared at a constant sand skeleton void ratio of 0.76. The
rates calculated for N=2 and N=10 show that the silty sand specimens generate
pore water pressure at significantly smaller rates than the clean sand specimens.
Finally, the rate of excess pore water pressure generation from sand and
silty sand specimens reconstituted at a constant overall void ratio of 0.59 is shown
in Figure 6-31. Similar to the case of constant sand skeleton void ratio, the rate of
pore water pressure generation for clean sand specimens is larger than for silty
sand specimens.
185
Figure 6-29 Rate of pore water pressure generation in sand and silty sand
specimens prepared at Dr=50%
186
Figure 6-30 Rate of pore water pressure generation in sand and silty sand
specimens prepared at es=0.76
187
Figure 6-31 Rate of pore water pressure generation in sand and silty sand
specimens prepared at e=0.59
188
6.8 DISCUSSION
This section presents a discussion of the results of this study. The trends
of this study are summarized and compared with the results from other studies
presented in the literature. These previous studies were discussed in Chapter 2.
At a constant relative density of 50%, the excess pore water pressure
generation indicated an initial decreasing trend between 0% and 10% fines
content, and then an increasing trend between 10% and 20% fines contents. In
terms of cyclic resistance, as discussed previously in Chapter 2 (Figure 2-22),
Polito and Martin (2001) reported a constant cyclic resistance below 36% fines
content (the limiting fines content for the soils they tested). However, if one
considers the data points presented by Polito and Martin (2001), rather than their
proposed constant trend, a different conclusion can be drawn. Figure 6-32 shows
the actual cyclic resistances reported by Polito and Martin (2001) for 0% to 20%
fines content and Dr=32%. These data were presented in their original form in
Figure 2-22. The data show that the cyclic resistance initially increases, which
refers to a decrease in excess pore water pressure, with increasing fines content up
to about 7%, and then decreases (an increase in excess pore water pressure) with
further increases in fines. Thus, the data points presented by Polito and Martin
(2001) appear to be consistent with the trends observed during this study at
constant relative density.
The observed initial decrease (up to 10% FC) and then increase (10%FC
to 20% FC) in pore water pressure generation for the constant relative density
tests may be explained by the transition from behavior dominated by the sand
skeleton to behavior dominated by the fines skeleton.
189
0.3
0.25
0.2
Cyclic resistance
0.15
0.1
0
0 5 10 15 20
Silt content (%)
Figure 6-32 Cyclic resistance values at Dr=30% (after Polito and Martin 2001)
The results of the constant relative density tests also indicated that the
threshold shear strain is influenced by the fines content. Based on the small pore
water pressure generation at 0.03% strain, it is inferred that the threshold shear
strain for specimens with more than 10% fines is larger than that of clean sand.
The investigation at a constant sand skeleton void ratio of 0.76 showed
that the presence of non-plastic fines results in less pore water pressure
generation. The pore water pressure generation curves (u* versus γ) for constant
sand skeleton void ratio display significantly larger pore water pressure
generation in clean sad specimens when compared to silty sand specimens. The
trend in pore water pressure generation with increasing fines content generally
shows a continuous decrease up to 10% fines, and then slight increase followed
by a slight decrease again. This decreasing trend in pore water pressure generation
appears to be in agreement with the observed increase in cyclic resistance with
190
fines content by Kuerbis et al. (1988) at a constant sand skeleton void ratio.
However, as discussed in Chapter 2, Polito and Martin (2001) reported a constant
cyclic resistance for Monterey sand and increasing cyclic resistance for Yatesville
sand up to the limiting fines content at a constant sand skeleton void ratio. Thus,
the findings of Polito and Martin (2001) are not in agreement with the results
from this research.
It is also important to note that at a constant sand skeleton void ratio, the
change in pore water pressure with increasing fines content appears to be strain
dependent. At relatively small shear strains (γ=0.03%), the large fines content
specimens do not show an increase in pore water pressure generation. Thus, it can
be hypothesized that at small strain the pore water pressure generation is
controlled by the proximity of the strain to the threshold shear strain. Thus, if the
addition of fines increases the threshold strain, pore water pressure is reduced at
small strains when fines are present.
The investigation at a constant overall void ratio was limited to 10% fines
content due to specimen preparation difficulties at larger fines contents. A
constant void ratio of 0.59 indicated a decrease in excess pore water pressure with
increasing fines up to 10% fines content. This decreasing trend appears to be in
conflict with previously published cyclic resistance studies. For the range of the
fines content considered, previous studies (e.g., Chang 1982, Koester 1994, Erten
and Maher 1995a, Polito and Martin 2001) consistently reported a decrease in
cyclic strength, which indicates an increase in pore water pressure.
Based on the results presented in this chapter, the following hypotheses
with regard to the effect of fines on the generation of excess pore water pressure
can be developed:
1. The addition of fines results in an increase in the threshold shear
strain for pore water pressure generation. This observation is
consistent with studies of nonlinear soil properties, which show
191
that fine-grained materials have a larger threshold strain.
However, it should be noted that this study used non-plastic fines
and most nonlinear property studies used plastic soils.
2. At small strains (γ < 0.05-0.1%), the excess pore water pressure is
influenced most by proximity to the threshold shear strain.
3. At large strains, if the sand skeleton void ratio remains smaller
than the maximum void ratio of the sand, the excess pore water
pressure decreases with fines content. This case was observed in
the constant Dr tests. Dr also plays a role in the generation of pore
water pressure as seen during the constant es tests where increasing
Dr indicated decreasing excess pore water pressure.
Finally, the results from all three testing series (Dr, es, and e) are
summarized in Figure 6-33 along with the data for clean sand from Dobry (1985).
As discussed earlier in Chapter 2, Dobry (1985) presented a band for the excess
pore water pressure generation from various sands, specimen preparation
techniques, relative densities and initial effective confining stresses. It should be
noted that Dobry (1985) curves were from CTX, and that based on the results
presented in Chapter 5 CTX-measured pore water pressure values are larger than
CDSS-measured values. An important trend in Figure 6-34 is that the presence of
fines, regardless of the index parameters used, indicates a decrease in the excess
pore water pressure generation.
192
Figure 6-33 General evaluation of the effect of fines content on pore water
pressure generation
6.9 SUMMARY
The effect of fines content on excess pore water pressure generation was
evaluated from strain-controlled, undrained, cyclic direct simple shear tests.
Monterey #0/30 sand was used to form the clean sand specimens, and mixtures of
Monterey #0/30 sand and Sil-Co-Sil 52 non-plastic silt were used to reconstitute
193
the silty sand specimens with up to 20% fines. The evaluation was made at three
different controlling index parameters: (1) a constant relative density of 50%, (2)
a constant sand skeleton void ratio of 0.76, and (3) a constant overall void ratio of
0.59. The selection of these values incorporated an optimization in terms of the
number of tests required as well as the relationship between sand skeleton void
ratio and overall void ratio. The tests were performed at shear strain levels
ranging between 0.03% and 0.3%.
The data in this study show that at a constant relative density, pore water
pressure generation decreases with increasing fines content up to about 10% fines
for a given shear strain. At large values of fines content, where the sand skeleton
void ratio is close to or greater than the maximum void ratio of the clean sand, the
pore water pressure generation starts to increase. At a constant sand skeleton void
ratio (less than the maximum void ratio for the clean sand), pore water pressure
generation tends to decrease with increasing fines content for the values of fines
content considered. Finally, at a constant overall void ratio, pore water pressure
generation decreases with increasing fines up to 10%.
Several hypotheses were developed regarding the effect of fines on pore
water pressure generation. These hypotheses relate the effect of fines content on
threshold strain and on pore water pressure generation at strains greater than the
threshold strain. Additionally, pore water pressure generation appears to increase
when enough fines are present to create a sand skeleton void ratio greater than the
maximum void ratio of the clean sand. At this point, the behavior of the soil is
dominated by the fines skeleton rather than the sand skeleton.
194
Chapter 7: COMPARISON OF IN SITU AND LABORATORY
MEASUREMENTS OF PORE WATER PRESSURE
GENERATION
7.1 INTRODUCTION
This chapter presents a comparison between in situ and laboratory
measurements of pore water pressure generation in liquefiable sand. The in situ
pore water pressure measurements were conducted in the field by Chang (2002)
with a newly developed in situ liquefaction testing technique that dynamically
loads a soil deposit with a truck-mounted hydraulic vibrator (called a vibroseis).
The soil response was recorded with embedded pore water pressure transducers
and geophones, and these data were used to compute the pore water pressure ratio
and shear strains induced in the soil. The laboratory pore water pressure
measurements were conducted using strain-controlled cyclic direct simple shear
testing. A total of 13 reconstituted sand specimens were prepared by water
sedimentation at relative densities between 32% and 44%, and tested at different
strain levels. Strain levels ranging from 0.005% to 0.96% were applied at initial
vertical effective stresses of 25 kPa and 100 kPa.
195
mounted hydraulic vibrator (called a vibroseis) to generate seismic waves that
propagate through the test area and induce a controlled number of cycles of shear
strain and shear stress. The soil response is recorded with embedded geophones
and miniature pore water pressure transducers, integrated in a single unit called a
liquefaction sensor. The collected data are used to compute the excess pore water
pressure ratio, u*, and shear strains, γ, induced in the soil. In situ liquefaction
testing was performed by Chang (2002) on a large (1.2 m by 1.2 m by 1.2 m)
reconstituted test specimen of clean, uniform aggregate sand. The index properties
and grain size distribution of this aggregate sand were presented in Chapter 4. A
schematic cross-section of the in situ test specimen is shown in Figure 7-1. The
specimen was constructed by water sedimentation at a relative density of 35%.
Because of the shallow location of the test specimen, the vertical effective stress
was relatively small (~13 kPa). Four liquefaction sensors were installed in such a
manner that they were at the corners of a 0.6 m by 0.6 m square array. An
additional sensor was located at the center of this array. The settlement was
monitored at the surface and within the specimen at levels where the liquefaction
sensors were located. A 0.8 m layer of air-dried sand was placed on top of the
specimen to impede vertical drainage and to create an overburden; thus increasing
the stresses acting on the test specimen.
Staged loading was performed with the vibroseis, with loading levels
progressing from those that induced no pore water pressure generation to a final
stage that induced significant pore water pressure generation. The induced shear
strains at specific locations were evaluated from relative displacements between
adjacent sensor locations. These displacements were computed from the particle
velocity data measured by the embedded geophones. The pore water pressure
generation was directly measured with the embedded pore water pressure
transducers. Thus, the coupled behavior between the dynamic response of the soil
skeleton, represented by shear strain, and the generated pore water pressure was
196
Vibroseis
truck ~0.8 m
Waterproof
liner
Backfill sand
0.3 m
Foundation 1 2
3.3 m 0.3 m
Legend 1.2 m
5
Liquefaction sensor 0.3 m
Accelerometer 3 4
0.3 m
Settlement plate 0.3 m 0.3 m
1.2 m
Figure 7-1 Schematic cross-section of the in situ testing set up (Rathje et al. 2005)
197
captured. Figure 7-2 shows the in situ measurements of pore water pressure
generation in terms of mean shear strain amplitudes and number of loading
cycles. During the first four stages of loading, the induced mean shear strains did
not exceed 0.003% and themeasured excess pore water pressure ratios were less
than about 1%. The mean shear strain and pore water pressure ratio increased to
about 0.007% and 8%, respectively, with application of increased loads in the
next three stages. In the last testing stage, the largest load was applied and caused
the sample to experience a mean shear strain of 0.014%. The excess pore water
pressure ratio after 20 cycles of loading in the final stage was 68%. The data in
Figure 7-2 are the basis of the comparison with the laboratory investigation,
described in the next section. Further details regarding the field experiments can
be found in Chang (2002) and Rathje et al. (2005).
Figure 7-2 In situ pore water pressure generation curves (Rathje et al. 2005)
198
7.3 LABORATORY EXPERIMENTS
To make a valid comparison between in situ and laboratory pore water
pressure responses, a series of strain-controlled, undrained, cyclic direct simple
shear tests were performed on the same material that was used in the field
experiments (Hazirbaba and Rathje 2004). A total of 13 reconstituted specimens
were prepared by the water sedimentation technique, which is the same specimen
preparation technique used for the field specimens. Each specimen was subjected
to cycles of a single shear strain amplitude until full liquefaction (u* ≥ 0.9) was
reached or 30 cycles of loading were applied. A list of the tests performed, along
with their testing conditions, and the specimen preparation, consolidation, and
testing procedures was presented previously in Chapter 4.
199
a) Pore water pressure histories for shear strains between 0.0055% to 0.96%
b) Pore water pressure histories for shear strains between 0.0055% to 0.0175%
Figure 7-4 Laboratory measurements from this study and Dobry et al. (1982)
201
pore water pressure generation at, or slightly less than, 0.01% and more pressure
is generated at larger strain levels with more cycles of loading.
The pore water pressure generation from the aggregrate sand was also
compared to that presented previously in Chapter 5 for Monterey #0/30 (Figure 7-
5). In this comparison, both data are from strain-controlled, undrained, cyclic
direct simple shear tests, yet the specimen preparation techniques are different.
The aggregate sand data, as pointed out earlier, were obtained from specimens
prepared by the water sedimentation technique at varying relative densities
(Dr=32-44%), whereas the Monterey #0/30 sand data were from specimens
prepared by the moist undercompaction technique. Two sets of data are presented
for Monterey #0/30 sand, (Dr=33% and Dr=50%) to match the range of relative
densities of the aggregate sand specimens. The comparison in Figure 7-5 shows
Figure 7-5 Pore water pressure generation curves for Monterey #0/30 and
aggregate sands
202
larger values of pore water pressure ratio at all shear strains for the aggregate sand
prepared by water sedimentation. Therefore, it appears that sample preparation
does affect pore water pressure generation in strain-controlled tests. This is
inconsistent with the observations of Ladd et al. (1989).
Figure 7-6 shows a comparison of the laboratory measurements of pore
water pressure generation obtained from this study and from Dobry et al. (1982),
with the in situ pore water pressure measurements from Rathje et al. (2005). The
in situ results indicate a threshold shear strain of about 0.005%, which is slightly
smaller than that measured in the laboratory. At a relatively larger strain of
Figure 7-6 Laboratory measurements at σv'=100 kPa along with in situ pore water
pressure measurements
203
0.015%, the in situ measured excess pore water pressure ratio increases to nearly
40%, while it increases to only about 10% in the laboratory specimens. The same
trend was observed at N=5 and N=20 loading cycles. The measurement of a lower
threshold shear strain and larger pore water pressure generation in situ may be
attributed to the fact that the vertical effective stress acting on the in situ specimen
was very low, approximately 13 kPa. Threshold strain is affected by overburden
pressure (Dobry and Swiger 1979), and this can explain the smaller threshold
strain measured in situ. The same trend can be extended to larger strain levels,
where lower effective stresses may result in larger induced excess pore water
pressures. To investigate this issue, a series of tests was performed at a vertical
effective stress of 25 kPa.
204
Figure 7-7 Laboratory-measured pore water pressure generation curves after 10
loading cycles at σv'=25 kPa and σv'=100 kPa
a) Pore water pressure generation under σv'=25 kPa and σv'=100 kPa at γ=0.034%
Figure 7-8 Effect of confining pressure on pore water pressure generation (continued)
205
b) Pore water pressure generation under σv'=25 kPa and σv'=100 kPa at γ=0.1%
c) Pore water pressure generation under σv'=25 kPa and σv'=100 kPa at γ=0.294%
206
water pressure generation for shear strain levels of 0.034%, 0.1%, and 0.294%. At
each of these shear strain levels, the pore water pressure generates more quickly at
25 kPa than at 100 kPa. This observation is particularly evident during the first
cycle of loading.
207
b) In situ and laboratory pore water pressure measurements after N=10
208
than the laboratory values. Unfortunately, 25 kPa is the smallest vertical effective
stress at which the GCTS CDSS device can operate. Still, the laboratory tests at
100 kPa and 25 kPa indicate that the low effective stresses in the field can explain
the large pore water pressure ratios measured in situ at relatively small strains.
7.4 SUMMARY
A series of strain-controlled, undrained, cyclic direct simple shear tests
was performed on a uniform, clean aggregate sand. The sand specimens were
prepared by water sedimentation at relative densities between 32% and 44%. Each
specimen was subjected to cycles of a constant shear strain level. Strain levels
ranging from 0.005% to 0.96% were applied at vertical effective stresses of 25
kPa and 100 kPa. Measurements of pore water pressure generation from this
laboratory study were compared to those previously measured in situ by Chang
(2002) and Rathje et al. (2005).
Laboratory-measured values of pore water pressure generation were
generally lower than measured in the field. The low initial effective stress state for
the in situ tests most likely caused the discrepancy. Both the laboratory and in situ
testing results reflected a threshold strain below which no pore water pressure
generation occurs. The threshold strain ranged from about 0.005% to 0.01%, with
the lower value indicated in situ at low effective stresses. The excess pore water
pressure generation was shown to be strongly influenced by the vertical effective
stress. Larger vertical effective stress results in less pore water pressure
generation. The laboratory and in situ measurements of pore water pressure
generation both show a sudden increase beyond the threshold strain. However, the
pore water pressure rises more quickly for the in situ measurements presented
here, most likely because of the low overburden stress in the in situ test specimen.
209
Chapter 8: SUMMARY AND CONCLUSIONS
8.1 SUMMARY
This dissertation presents results from an experimental research program
carried out at The University of Texas at Austin on excess pore water pressure
generation in clean sands and silty sands under cyclic loading conditions. The
aim of this research was primarily to find a unified picture regarding the effect of
fines on excess pore water pressure generation, which is the main responsible
mechanism for the occurrence of liquefaction. The main components of the
research program are: (1) the testing program, (2) an assessment of excess pore
water pressure generation in clean sands, (3) an assessment of excess pore water
pressure generation in silty sands, and (4) a comparison between in situ and
laboratory measured excess pore water pressures.
210
undrained loading conditions. Some modifications were needed on the original
design of the system to meet the particular testing needs of this research (e.g.,
internal instrumentation).
Three different soils were used in this research program: Monterey #0/30
sand, aggregate sand, and Sil-Co-Sil 52 non-plastic silt. Monterey #0/30 sand and
aggregate sand were used as clean sands. Monterey #0/30 sand was mixed with
different percentages of Sil-Co-Sil 52 non-plastic silt to reconstitute the silty sand
specimens.
The testing program consisted of multiple series of strain-controlled cyclic
direct simple shear and cyclic triaxial tests. The tests performed included: (1) a
series of cyclic direct simple shear tests on Monterey #0/30 sand specimens at
three different relative densities (Dr = 32%, 50%, 93%), (2) a series of cyclic
triaxial tests on Monterey #0/30 sand specimens at Dr = 50%, (3) a series of
cyclic direct simple shear tests on silty sands at a constant relative density of 50%,
(4) a series of cyclic direct simple shear tests on silty sands at a constant sand
skeleton void ratio of 0.76, (5) a series of cyclic direct simple shear tests on silty
sands at a constant overall void ratio of 0.59, and (6) a series of cyclic direct
simple shear tests on the aggregate sand.
Monterey #0/30 sand specimens and silty sand specimens were prepared
by the moist undercompaction technique and aggregate sand specimens were
prepared by water sedimentation. Because the moist undercompaction technique
for medium to low density specimens resulted in highly contractive and
metastable specimens at fines contents larger than 20%, this study was limited to
specimens with 20% fines or less.
The development of the testing program also included a series of stress-
controlled tests on Monterey #0/30 sand specimens for verification of the testing
system.
211
8.1.2 Assessment of excess pore water pressure generation in clean sand
The assessment of excess pore water pressure generation in clean sand
involved strain-controlled, undrained, cyclic direct simple shear and cyclic triaxial
tests. The cyclic direct simple shear tests were performed on three different
density groups of sand specimens (Dr=32%, 50%, 93%). The cyclic triaxial tests
were conducted on a set of specimens that were reconstituted at 50% relative
density. The shear strains used in this part of the research ranged from 0.005% to
0.3% for the cyclic direct simple shear tests, and from 0.01% to 0.3% for the
cyclic triaxial tests.
The measurements of excess pore water pressure from the cyclic direct
simple shear tests were presented for each density group. Results from various
densities allowed for an investigation on the effect of relative density on excess
pore water pressure generation.
The results of strain-controlled cyclic triaxial tests were evaluated and
verified with previously published studies. A comparison between the
measurements of excess pore water pressures from cyclic triaxial tests and cyclic
direct simple shear tests was made to investigate the effect of the testing system
on the generation of pore water pressure.
212
skeleton void ratio of 0.76, and (3) a constant overall void ratio of 0.59. The
selection of these values incorporated an optimization in terms of the number of
tests required as well as the relationship between the sand skeleton void ratio and
the overall void ratio. The tests were performed at shear strain levels ranging
between 0.03% and 0.3%. The test results from each of these categories are
presented, and evaluated in terms of pore water pressure generation
characteristics.
Several hypotheses were proposed to explain the effect of fines on pore
water pressure generation. These hypotheses relate the effect of fines content on
threshold strain and on pore water pressure generation at strains greater than the
threshold strain.
213
8.2 CONCLUSIONS
The conclusions of this research program can be viewed under three
different categories: (1) conclusions reached during the development of the testing
program, (2) conclusions drawn from the tests on clean sands, and (3) conclusions
drawn from the tests on silty sands. Each of these categories is outlined in the
following sections.
214
8.2.2 Conclusions from the tests on clean sands
215
vertical effective stress; larger vertical stress resulted in less pore
water pressure generation at each strain level.
• In situ and laboratory measured pore water pressures were
compared. Both the laboratory and in situ testing results reflected a
threshold strain, below which no pore water pressure generation
occurs. The threshold shear strain for the aggregate sand was
between 0.005% to 0.01%, with the lower value indicated in situ at
low effective stresses. At strains above the threshold values, in situ
measurements indicated larger pore water pressure generation than
laboratory measurements. These larger pore water pressure values
can be ascribed to the low effective stress present in the in situ
tests.
216
• At relatively large shear strains (~0.3%), the excess pore water
pressure decreases with increasing fines content provided the sand
skeleton void ratio remains smaller than the maximum void ratio of
the sand.
• The relative density was found to play a role in the generation of
pore water pressure for silty sands. This was observed in the
constant sand skeleton void ratio tests where increasing relative
density showed decreasing excess pore water pressure.
• The results from all three testing series (Dr, es, and e) were
combined and compared to previously reported clean sand data
from Dobry (1985). The presence of fines, regardless of the index
parameters used, indicated a decrease in the excess pore water
pressure generation.
217
studied with a strain-controlled approach. The plasticity of the fines will play a
role in pore water pressure generation because the nature of the fines will change
the interaction between the coarse and fine grains.
As discussed in Chapter 3, various types of laboratory tests have been
developed to investigate the liquefaction phenomenon (e.g., cyclic direct simple
shear test, cyclic torsional shear test, cyclic triaxial test). The objective of a
laboratory test is to replicate the loading conditions as closely as possible.
Although cyclic direct simple shear and cyclic torsional shear tests provide
perhaps the best representation of field conditions, they are still subject to
limitations that stem from the specimen size, specimen preparation procedures,
and compliance effects of the system. To overcome these limitations and obtain
more reliable data on liquefaction of sands, De Alba et al. (1976) used a large
scale, 1-D shaking table. They measured the cyclic resistance of sands for various
densities, and the data De Alba et al. (1976) obtained still serve as a very reliable
reference for laboratory liquefaction studies on sands. The quality and reliability
of the data come from the ability of the test type to almost perfectly duplicate in
situ loading conditions. Similar large scale shaking table tests on silty sands
would provide more information regarding the pore water pressure generation
characteristics of these soils. The shaking table system described in De Alba et al.
(1976) can be used for this purpose. Specimens can be prepared either as layered
silty sand with water sedimentation (this would represent a layered soil deposit
consisting of sand and silt layers) or as a slurry. In both cases, with the help of the
large size of the specimen, attainment of relatively loose fabric should be
possible. Cyclic direct simple shear stresses and strains will be induced in the test
specimen by the shaking table. In such a test, the instrumentation appears to be
very challenging. However, the instrumentation discussed by Rathje et al (2005)
for their in situ liquefaction test may be adopted, although at a smaller scale.
Rathje et al. (2005) used embedded liquefaction sensors (an integrated unit of a
218
pore water pressure transducer and a geophone) to monitor the response of the
soil. Thus, excess pore water pressure generation curves for the silty sand may be
obtained and evaluated.
An investigation on the post-liquefaction behavior of silty sands will
provide insight regarding post-earthquake settlements of natural soil deposits and
volume change characteristics. The generation of excess pore water pressure due
to seismic loading in natural soil deposits is typically followed by the dissipation
phase. Dissipation of pore water pressure leads to volume change in the soil
deposit, and this volume change will mostly be observed in the form of settlement
of the ground surface. Extensive damage may occur due to these settlements.
Therefore, it is important to characterize and be able to predict the post-cyclic
behavior of granular soils. There have been investigations on post-earthquake
volume change characteristics of clean sands (e.g., Lee and Albaisa 1974,
Tokimatsu and Seed 1987, Ishihara and Yoshimine 1992). Ishihara and
Yoshimine (1992) proposed a procedure for prediction of the settlements
occurring in clean sand deposits due to pore water pressure dissipation. Their
procedure was based on findings from laboratory tests on clean sands. They
correlated the volumetric strain caused by the dissipation of pore water pressure
with the density of sand and the factor of safety against liquefaction. Similarly,
the post-earthquake settlements of silty sands can be investigated. Because
previous studies on clean sands indicated that the post-liquefaction volumetric
strain is strongly influenced by the shear strains induced in the soil deposit
(Nagase and Ishihara 1988) strain-controlled test rather than stress-controlled tests
are appropriate for the research in this area. Silty sand specimens may be
subjected to strain-controlled, undrained, cyclic direct simple shear tests at
various shear strain levels to generate excess pore water pressure. The generated
pore water pressure can be dissipated by allowing drainage after a predetermined
number of loading cycles or a certain value of excess pore water pressure, and
219
volume change and settlement characteristics can be evaluated. Preliminary
observations from the tests performed during this study indicated a trend in the
favor of silty sands with less post-cyclic volume change when compared to clean
sands. An extensive testing program is needed to investigate various factors (e.g.
fines content, density, shear strain level, number of loading cycles) that may play
role on the post-liquefaction volume change and settlement behavior of silty
sands.
To evaluate the excess pore water pressure generation at larger fines
contents (i.e. FC > 20%), development of new specimen preparation techniques
that will allow for tests on liquefiable specimens with large fines content is
necessary. Research in this area would significantly assist systematic laboratory
liquefaction studies on silty sands. For example, it would have been possible to
portray a more complete picture of the pore water pressure generation
characteristics at all fines contents up to 100%, had silty sand specimens with
large fines content been attained successfully in this study. Additionally,
investigations on testing undisturbed specimens with large fines contents may be
considered.
Additional recommendations based on the results of this research include
investigations for the use of current findings in deformation-based liquefaction
analyses. The cyclic strain-approach presented by Dobry et al. (1982) was
discussed in Chapter 2. The liquefaction potential in the cyclic strain approach is
evaluated from the pore water pressure generation curves. Based on the results of
this study, the fines (for the range considered here, FC = 0% to 20%) generally
cause a decrease in pore water pressure generation (Figure 6-33). Development of
a family of pore water pressure generation curves with more data for each value
of fines content is recommended. Such a family of pore water pressure generation
curves will allow for the use of the cyclic-strain procedure in the evaluation of
liquefaction potential of silty sands. Each family of curves can be developed from
220
a large number of strain-controlled tests that takes into account various index
parameters and types of fines (non-plastic, plastic). Thus, readily available pore
water pressure generation curves could be used for liquefaction evaluation by the
cyclic-strain approach.
221
References
Amer, M. I., Kovacs, W. D., and Aggour, M. (1987), “Cyclic Simple Shear Size
Effects,” Journal of the Geotechnical Engineering Division, ASCE, Vol.
113, No. 7, pp. 693-707.
ASTM D4253 (2000), “Standard Test Methods for Maximum Index Density and
Unit Weight of Soils Using a Vibratory Table,” pp. 1-14.
ASTM D4254 (2000), “Standard Test Methods for Minimum Index Density and
Unit Weight of and Calculation of Relative Density,” pp. 1-9.
ASTM D854 (2000), “Standard Test Methods for Specific Gravity of Soil Solids
by Water Pycnometer,” pp. 1-7.
Boulanger, R. W., Clarence, K. C., Seed, H. B., Seed, R. B., and Sousa, J. B.
(1993), “A Low-Compliance Bi-Directional Cyclic Simple Shear
Apparatus,” Geotechnical Testing Journal, GTJODJ, Vol. 16, No. 1, pp.
36-45.
222
Boulanger, R. W., Seed, R. B., Chan, C. K., Seed, H. B., and Sousa, J. (1991),
“Liquefaction Behavior of Saturated Sands Under Uni-Directional and Bi-
Directional Monotonic and Cyclic Simple Shear Loading,” Rep. No.
UCB/GT/91-08, University of California, Berkeley.
Budhu, M., and Britto, A. (1987), “Numerical Analysis of Soils in Simple Shear
Devices,” Soils and Foundations, Vol. 27, No. 2, pp. 31-41.
De Alba, P., Seed, H. B., and Chan, C. K. (1976) “Sand Liquefaction in Large
Scale Simple Shear Tests,” Journal of the Geotechnical Engineering
Division, ASCE, Vol. 102, No. GT9, pp. 909-927.
223
Dobry, R. (1985) Liquefaction of Soils During Earthquakes, National Research
Council (NRC), Committee on Earthquake Engineering, Report No.
CETS-EE-001, Washington DC.
Dobry, R., Ladd, R. S., Yokel, F. Y., Chung, R. M., and Powell, D. (1982),
“Prediction of Pore Water Pressure Buildup and Liquefaction of Sands
During Earthquakes by the Cyclic Strain Method,” National Bureau of
Standards Building Science Series 138, 150 pp.
Dyvic, R., Berre, T., Lacasse, S., and Raadim, B. (1987), “Comparison of truly
Undrained and Constant Volume Direct Simple Shear Tests,”
Geotechnique, Vol. 37, No. 1, pp. 3-10.
Finn, W. D. L., Lee, K. W., and Martin, G. R. (1977) “An Effective Stress Model
for Liquefaction,” Journal of the Geotechnical Engineering Division,
ASCE, Vol. 103, No. GT6, pp. 517-533.
224
Finn, W. D. L., Vaid, Y. P., and Bhatia, S. K. (1978), “Constant Volume Cyclic
Simple Shear Testing,” Proceedings, 6th World Conference on Earthquake
Engineering, New Delhi, India, pp. 2157-2162.
Franke, E., Kiekbusch, M., and Schuppener, B. (1979), “A New Direct Simple
Shear Device,” Geotechnical Testing Journal, Vol.2, No. 4, pp. 190-199.
Hight, D. W., Gens, A., and Symes, M. J. (1983), “The Development of A New
Hollow Cylinder Apparatus for Investigating the Effects of Principal
Stress Rotation in Soils,” Geotechnique, Vol. 33, No. 4, pp. 355-383.
Ishihara, K., Troncoso, J., Yasuhiro, K., and Yoshiki, T. (1980), “Cyclic Strength
of Tailing Materials,” Soils and Foundation, Vol. 20, No. 4, pp. 127-142.
225
Kayen, R. E. and Mitchell, J. K. (1997), “Assessments of Liquefaction Potential
by Arias Intensity,” Journal of Geotechnical and Geoenvironmental
Engineering, ASCE, Vol. 123, No. 12, pp. 1162-1174.
Kuerbis, R. H., Negussey, D., and Vaid, Y. P. (1988) “Effect of Gradation and
Fines Content on the Undrained Response of Sand,” Geotechnical Special
Publication No. 21, ASCE, pp. 330-345.
Ladd, R. S., Dobry, R., Dutko, F. Y., Yokel, F. Y., and Chung, R. M. (1989),
“Pore-Water Pressure Buildup in Clean Sands Because of Cyclic
Straining,” Geotechnical Testing Journal, GTJODJ, Vol. 12, No. 1, pp.
77-86.
Lade, P. V., Liggio, C. D., Jr., and Yamamuro, J. A. (1998), “Effects of Non-
Plastic Fines on Minimum and Maximum Void Ratios of Sand,”
Geotechnical Testing Journal, GTJODJ, Vol. 21, No. 4, pp. 336-347.
Law, T. W., Cao, Y. L., and He, G. N. (1990), “An Energy Approach for
Assessing Seismic Liquefaction Potential,” Canadian Geotechnical
Journal, Vol. 27, No. 3, pp. 320-329.
226
Lee, K. L., and Albaisa, A. (1974), “Earthquake Induced Settlements in Saturated
Sands,” Journal of the Geotechnical Engineering Division, ASCE, Vol.
100, No. GT4, pp. 387-406.
Lucks, A. S., Christian, J. T., Brandow, G. E., and Hoeg, K. (1972), “Stress
Conditions in NGI Simple Shear Test,” Journal of the Soil Mechanics and
Foundations Division, ASCE, Vol. 98, No. SM1, pp. 155-160.
Mitchell, J. K. (1993), Fundamentals of Soil Behavior, 2nd ed., John Wiley and
Sons, Inc., pp. 437.
Mulilis, J. P., Chan, C. K., and Seed, H. B. (1975), “The Effects of Method of
Sample Preparation on the Cyclic Stress-Strain Behavior of Sands,”
Report No. EERC 75-18, Earthquake Engineering Research Center,
University of California, Berkeley.
Mulilis, J. P., Seed, H. B., Chan, C. K., Mitchell, J. K., and Arulanadan, K.
(1977), “Effects of Sample Preparation on Sand Liquefaction,” Journal of
the Geotechnical Engineering Division, ASCE, Vol. 103, No. GT2, pp.
91-108.
227
Peacock, W. H. and Seed, H. B. (1968), “Sand Liquefaction Under Cyclic
Loading Simple Shear Conditions,” Journal of Soil Mechanics and
Foundations Division, ASCE, Vol. 94, No. SM3, pp. 689-708.
Roscoe, K. H. (1953), “An Apparatus for the Apllication of Simple Shear to Soil
Samples,” Proceedings, 3rd International Conference on Soil Mechanics
and Foundation Engineering, Vol. 1, pp. 186-191.
Saada, A. S., Fries, G., and Ker, C. C. (1983), “An Evaluation of Laboratory
Testing Techniques in Soil Mechanics,” Soils and Foundations, Vol. 23,
No. 2, pp. 98-112.
228
Salgado, R., Bandini, P., and Karim, A. (2000), “Shear Strength and Stiffness of
Silty Sands,” Journal of Geotechnical and Geoenvironmental
Engineering, ASCE, Vol. 126, No. 5, pp. 451-462.
Sancio, R. B., Bray, J. D., Stewart, J. P., Youd, T. L., Durgunoglu, H. T., Onalp,
A., Seed, R. B., Christensen, C., Baturay, M. B., and Karadayilar, T.
(2002), “Correlation Between Ground Failure and Soil Conditions In
Adapazari, Turkey,” Soil Dynamics and Earthquake Engineering, Vol. 22
(9-12), pp. 1093 - 1102.
Seed, H. B. (1979), “Soil Liquefaction and Cyclic Mobility Evaluation for Level
Ground During Earthquakes,” Journal of the Geotechnical Engineering
Division, ASCE, Vol. 105, No. GT2, pp. 201-255.
Seed, H. B., and Idriss, I. M. (1971), “Simplified Procedure for Evaluating Soil
Liquefaction Potential,” Journal of the Soil Mechanics and Foundations
Division, ASCE, Vol. 97, No. SM9, pp. 1249-1273.
Seed, H. B., and Idriss, I. M. (1982), Ground Motions and Soil Liquefaction
during Earthquakes, Monograph Series, Earthquake Engineering Research
Institute, Berkeley, California.
Seed, H. B., Martin, P. P., and Lysmer, J. (1976), “Pore-Water Pressure Changes
during Soil Liquefaction,” Journal of the Geotechnical Engineering
Division, ASCE, Vol. 102, No. GT4, pp. 323-346.
229
Seed, H. B., Mori, K., and Chan, C. K. (1975b), “Influence of Seismic History on
the Liquefaction Characteristics of Sands,” Report No. EERC 75-25,
Earthquake Engineering Research Center, University of California,
Berkeley.
Seed, H. B., and Peacock, W. H. (1971), “Test Procedure for Measuring Soil
Liquefaction Characteristics,” Journal of the Soil Mechanics and
Foundations Division, ASCE, Vol. 97, No. SM8, pp. 1099-1119.
Seed, H. B., Tokimatsu, K., Harder, L. F., and Chung, R. M. (1985), “Influence of
SPT Procedures in Soil Liquefaction Resistance Evaluations,” Journal of
Geotechnical Engineering, ASCE, Vol. 111, No. 12, pp. 1425-1445.
Shen, C. K., Sadigh, K., and Hermann, L. R. (1978), “An Analysis of NGI Simple
Shear Apparatus for Cyclic Soil Testing,” Rep. No. 654, American Society
for Testing and Materials.
Tatsuoka, F., Toki, S., Miura, S., Kato, H., Okamoto, M., Yamada, S., Yasuda, S.,
and Tanizawa, F. (1986), “Some Factors Affecting Cyclic Undrained
Triaxial Strength of Sand,” Soils and Foundations, Vol. 26, No. 3, pp. 99-
116.
Thevanayagam, S., Fiorillo, M., and Liang, J. (2000), “Effect of Non-Plastic Fines
on Undrained Cyclic Strength of Silty Sands,” Geotechnical Special
Publication No. 107, ASCE, pp. 77-91.
230
Tokimatsu, K. and Seed, H. B. (1987): “Evaluation of Settlements in Sands due to
Earthquake Shaking,” Journal of Geotechnical Engineering, ASCE, Vol.
113, No. GT8, pp. 861-878.
Zienkiewicz, O. C., Chan, A. H. C., Pastor M., Paul, D. K., and Shiomi, T.
(1990), “Static and Dynamic Behavior of Geomaterials – A Rational
Approach to Quantitative Solutions, Part-1 Fully Saturated Problems,”
Proc. Royal Society of London, A429, pp. 285-309.
231
Vita
Kenan Hazırbaba was born in Mardin, Turkey, on April 9th, 1973, the first
child of Mehmet and Emel Hazırbaba. He graduated from Mardin Technical High
School before attending Istanbul Technical University (ITU). He earned his
Bachelor of Science in Civil Engineering from ITU in 1996. He then attended
Bogazici University (BU) in Istanbul and received his Masters of Science in Civil
Engineering from BU in 1999. During his studies at BU, he worked as a research
and teaching assistant. He started his Ph.D. study at The University of Texas at
Austin in August 2000.
232