Professional Documents
Culture Documents
1285
1. Introduction
Concepcin is placed in the middle of Chile, 500 km to the south of Santiago. The city is delimited by Valle
de la Mocha and lies on the foothill of the cordillera de la costa mountains, beside the Bo Bo river. The
position of the riverbed of Bo Bo river has changed throughout history. Therefore, parts of Concepcion lie
on fluvial deposits formed by ancient branches of the Bo Bo river. In particular, the prosperous and active
urban center of Concepcin is located on these areas. Despite the fact that these zones are highly prone to
flooding [1] residential districts and commercial zones have been built there.
The ground of Concepcin is composed of unconsolidated deposits of black sands and eolic silt [2],
mainly composed of volcanic sediments, i.e. basalt fragments and feldspars (as visible in Figure 1). Its origin
seems related to the Antuco volcano. The Bo Bo sand (Figure 1) is a clean uniform sand (!!" ! !!"! mm,
!"!" ! ! mm, !! ! !!!", !! ! !!!"), with angular particle form and generally without fine particles or
cementation and in some cases salty. Its colour is dark gray to black brown with fragments of black grey
volcanic rocks and very few granite boulders. In some zones thin layers of grey silt with less than 1 mm
thickness can be found. According to its stratigraphic sequence, in some areas clastic and granitic rock exist
[3].
1
Departament of Civil Engeenering, Universidad Catlica de la Santsima Concepcin, Alonso de
Ribera 2850, Concepcin. E-mail: mauro@ucsc.cl
1286
The present paper deals with the problem of high-cyclic loading and prediction of accumulated
settlements of foundations on Bo Bo sand. A high-cyclic loading means a loading with a large number of
cycles (N > 1000) of relatively small strain amplitudes (ampl < 10-3). Such cyclic loading is of practical
relevance for example in the case of traffic infrastructure (railways, highways), machine foundations or wind
power plant foundations (as studied in Section 3). Coastal structures are also subjected to a high-cyclic
loading due to waves. Such cyclic loading causes reorientations of grains leading to an accumulation of
deformation (compaction) in the sand and thus settlements of the foundations. An excessive accumulation of
inhomogeneous settlements, in particular occurring in loosely deposited sands as the Bo Bo sand deposits
below Concepcion, can lead to a loss of the serviceability of structures. Therefore, the prediction of these
permanent deformations during the design phase of foundations is desirable. However, such prediction is a
challenging issue since the cumulative deformations depend on numerous parameters like amplitude, soil
density, fabric of the grain skeleton, cyclic preloading history, average stress, etc.
Figure 1: Grain size distribution curve of the Bo Bo sand and photo of the grains.
The high-cycle accumulation (HCA) model of Niemunis et al. [4] can be applied for a prediction of
these permanent deformations by means of the finite element method. The HCA model is used in the
framework of a special calculation scheme. The first two cycles are calculated with a conventional
constitutive model (see Figure 2). The authors use hypoplasticity (Kolymbas [5], von Wolffersdorff [6],
Niemunis [7]) with the extension by the intergranular strain concept of Niemunis and Herle [8] for that
purpose. A severe problem of the original hypoplastic model is the prediction of excessive permanent strain
accumulation in case of small loading and unloading cycles. The intergranular strain concept solves this
problem. During the second cycle of the hypoplastic calculation, the strain path is recorded in each
integration point. The strain amplitude ampl is determined from this strain path, which is one of the most
important input parameters of the HCA model. During the following explicit calculation (see Figure 2), the
HCA model predicts the cumulative deformations due to the cyclic loading directly, without following the
stress or strain path during the individual cycles. The structure of the HCA model is similar to a viscoplastic
model, replacing time t by the number of cycles N. Therefore, the HCA model predicts the accumulation of
permanent strains in sand due to a high-cyclic loading in a similar manner as viscoplastic models predict
creep in cohesive soils under constant load.
The basic equation of the HCA model reads:
! ! !! !! ! ! !!"" ! ! !"! !
(1)
with stress rate !, elastic stiffness E, strain rate ! , rate of strain accumulation !!"" and plastic strain
rate !!"! . The plastic strain rate is necessary in order to keep the average stress possibly evolving during highcyclic loading within the yield surface.
1287
The rate of strain accumulation !!"" is calculated as the product of a scalar cumulative intensity ! !"! ,
and a tensorial direction of accumulation ! (flow rule):
!!"" ! ! !"" !! ! !!#$" !!!! !! !!! !!! !!!! !
(2)
The intensity of accumulation is obtained as the product of six functions, each considering a separate
influencing parameter, i.e. strain amplitude (!!"#$ ), cyclic preloading (!!!! ), void ratio (!! ), average mean
pressure pav (!! ), normalized average stress ratio ! !" (!! ) and polarization changes (!! ).
(3)
Critical friction angle!!!! : The critical state is characterized by large shear deformations without any
changes of shear stress and volume. The shear stress mobilized in the critical state is determined by the
critical friction angle c of the material. This constant can be determined from undrained monotonic triaxial
tests or from cone pluviation tests. In the present study the latter test method has been used (Figure 3). !! is
the inclination of the cone.
!!
! !"# !
!! !
!!
(4)
These parameters can be determined from the compression curves measured in oedometric tests
beginning from the loosest possible state [8]. The granular hardness !! and the exponent ! were determined
from the curves !!!! from fourteen oedometric compression tests performed on dry Bo Bo sand. For
1288
comparison purpose two different specimen geometries were used: a diameter of 100 mm with a height of 18
mm and a diameter of 150 mm with a height of 30 mm. Seven tests have been performed with the smaller
sample geometry while seven other ones were conducted with the larger diameter. The measured curves !!!!
for the loose sand are given on the left-hand side of Fig. 4 (dashed curves). The oedometric stiffness of the
smaller samples has been found somewhat lower than that of the larger ones. This is probably due to
boundary effects that are more significant in the case of the smaller samples. Eq. (4) was fitted to each curve
!!!! resulting in the constants !! and !. Afterwards the oedometric tests have been simulated using the
element test program IncrementalDriver of A. Niemunis (solid curves in Fig. 4). During an iterative
procedure, the parameters hs and n have been optimized until the best approximation of the test data has been
achieved. The optimum parameters are summarized in Table 1. Due to the larger stiffness measured for the
samples with diameter 150 mm, the corresponding granular hardness is larger and the exponent n is smaller
(Table 1).
Figure 4: Oedometric compression tests on loose (left-hand side) and dense (right-hand side) Bo Bo
sand using samples of 100 mm or 150 mm diameter. The dashed curves are the results from the experiments.
The solid curves have been obtained from the element test simulations.
Exponent !: This constant controls the influence of the material density on the peak friction angle. In order
to determine !, a test with triaxial compression may be performed on an initially dense specimen. Based on
the measured peak shear strength, ! can be determined from the following equation:
!!
!" !
!!! !
! !!! !
!!! ! ! ! ! !!!"!" !!
!! ! !!
!" !!
!!! !!"!# !!
The factors !! (peak stress ratio) and "!" !! (dilatancy angle) are defined as:
!! !
and
#!# !! ! !
!! !
!!
!!!
!!!
!!"!" ! !
!!"!# ! !
!! ! !!!!
!!
!!! ! !!!! ! !!
!! ! !!
(5)
(6)
(7)
(8)
!!! and !!! are the axial and radial stress components in the peak state. re = (e-ed)/(ec-ed) is the
pressure-dependent relative density. The factor a in Eq. (5) depends on the critical friction angle:
!!
! !!"!# ! !
! ! "!# ! !
(9)
Four drained monotonic triaxial tests with different initial densities (ID0 = (emax e)/(emax emin) = 0.17, 0.62,
0.77 and 0.90 have been performed. The samples measured 100 mm in diameter and 100 mm in height. They
1289
were prepared by air pluviation. The effective confining pressure was !! ! ! "!! kPa in all tests. The
measured curves of deviatoric stress and volumetric strain versus axial strain are given in Figure 5 (dashed
curves). Obviously, the shear strength and the dilatancy increases with increasing density (pyknotropy) [10].
Two more tests have been performed on medium dense samples with effective confining pressures of 200
and 400 kPa. The curves q(1) and v(1) for all three medium dense samples (ID0 = 0.61 0.65) are provided
as dashed curves in Figure 6, demonstrating the barotropy of the material.
Figure 5: Drained monotonic triaxial tests with variation of initial relative density. The dashed curves are the
results from the experiments. The solid curves have been obtained from the element test simulations.
Figure 6: Drained monotonic triaxial tests with variation of effective confining pressure. The dashed curves
are the results from the experiments. The solid curves have been obtained from the element test simulations.
The parameter has been calibrated from Eq. (5) based on the test performed on the sample with the
largest initial density (ID0 = 0.90). Afterwards, all triaxial compression tests have been simulated by means of
IncrementalDriver. The parameter was slightly adjusted to deliver a perfect agreement between the
measured peak strength and that predicted by hypoplasticity (Figure 4). The parameter is also slightly
affected by the parameters hs and n which differ for the two tested geometries in the oedometric tests (see
Table 1). The results from the simulations with the optimum -value are given as solid curves in Figures 4
and 5. Obviously, some aspects of the experimental data are reproduced well by the constitutive model (e.g.
the stress strain and dilatancy curves in the test with 3 = 200 kPa) while in some other cases there are larger
deviations between the experiments and the prediction (e.g. some of the v(1) curves)
Exponent !: The constant ! effects an increase of the stress rate ! with increasing density at D =
constant. It can be obtained from oedometric tests on specimens with different initial densities. For that
purpose additional oedometric compression tests on dense sand have been performed. The resulting curves
e(p) are shown on the right-hand side of Figure 4. From the tests on the loose and the dense samples, the
oedometric stiffness !!! (loose) and!!!! (dense) have been determined for the same effective mean stress !!.
The parameter ! can then be obtained from:
!!
with ! !
!!!!!! !! !! !
!!!!!!
!"!
and
!!
(10)
!!!!!!! !! !!!!!! !
!!!!!!!! !
(11)
!!! and !!! (Eq. 10) are the pyknotropy factors for loose and dense sand respectively, with !! and !!
evaluated at the p value under consideration. After has been determined from Eq. (10) it has been
optimized by recalculations of the dense oedometer tests (solid curves in Figure 4). The optimum parameters
for the two different sample geometries are summarized in Table 1. Note, that a negative value was
necessary to reproduce the stiffness measured in the oedometric tests on dense samples in the case of the
diameter 100 mm (see Table 1). This negative value is necessary because the density-dependence expressed
by the hypoplastic equations without fe is too large, i.e. the negative value of must counteract.
1290
The extension of hypoplasticity by intergranular strain needs five more constants ! ! , !! , !"!# , !! , !.
! ! and !! are constants that represent the increases of stiffness due to changes of the strain path direction
by 90 or 180, respectively. !"!# is the strain range for which the stiffness is constant and at its maximum
value. The parameters!!! and ! describe the degradation of the stiffness with continued monotonic loading
after the change of the strain path direction. These five parameter can be obtained from resonant column tests
and triaxial tests with changes of the direction of loading. In the case of Bo Bo sand the parameters have
been determined from a drained cyclic triaxial test performed on a medium dense sample (ID0 = 0,65) with an
average stress of pav = 200 kPa, av = qav/pav = 0.75 and a deviatoric stress amplitude of qampl = 60 kPa. The
following assumptions have been made: R = 10-4, r = 0.1, = 6 and mT = mR/2 1. The chosen values of r
and render the material response during the cycles nearly perfectly elastic, i.e. the undesired ratcheting
effect is prevented. This is appropriate in the present case because the accumulation of deformations is
predicted during the subsequent calculation using the HCA model. In order to calibrate the constant mR, five
cycles of the cyclic triaxial test were simulated with IncrementalDriver. The constant mR has been varied
until the strain amplitude ampl measured in the test could be reproduced in the simulations. The optimum
constants are summarized in Table 1. Note that the obtained mR-values are relatively small, owing to the
relatively large values of the granular hardness hs. It has to be kept in mind that mR increases the basic
stiffness predicted by the hypoplastic model, i.e. mR is coupled with the hs and n values previously
determined. Therefore, the mR and mT values also depend on the sample geometry used in the oedometric
tests.
2.3 HCA model
For the calibration of the HCA model usually at least 11 drained cyclic triaxial tests with different stress
amplitudes, initial densities, average mean pressures and average stress ratios are necessary. A simplified
procedure has been proposed by Wichtmann et al. [11]. Following this procedure, the HCA model parameters
can be fully or partly estimated from correlations with granulometry (mean grain size d50, uniformity
coefficient Cu) or index quantities (minimum void ratio emin). These correlations have been developed based
on approximately 350 cyclic triaxial tests on clean sands with !!!!!! ! !"" ! !!!!!! and !!! ! !! ! !
[11]. In the present study on Bo Bo sand, the four constants Campl, Ce, Cp and CY have been estimated from
the correlations provided in [11], using d50 = 0.73 m and Cu = 2.15, while the parameters CN1, CN2 and CN3
have been determined from the strain accumulation curve acc(N) (see Figure 7) measured in the drained
cyclic triaxial test mentioned in Section 2.2. The constants CN1, CN2 and CN3 determine the shape of the
accumulation curves predicted by the HCA model. Table 1 summarizes the HCA model parameters obtained
for Bo Bo sand.
Table 1: Constants of hypoplasticity with intergranular strain and of the HCA model for Bo Bo sand. The
different stiffness measured in the oedometric tests with different diameters (100 or 150 mm) leads to
different sets of material constants.
!"#$%!"#$%
!! ! "!# !
!! ! "!" !
!!
!!" !
!!" !
!!" !
!!
!!
!! !
!! !
!#!# !
&''%((%
&)'%((%
"!#$$!
'()%#$%#!
"!'+"!
"!%)%!
""!#"!
%!*%!
'!('!
'%)&''(")!
"!''"!
"!#"(!
"!('*!
"!$*+!
"!%'$!
"!)"!
%!
%!$!
"!"""%!
*+"$#!"#$%
!! !
!!
!!! !
!!! !
!!! !
!! !
!!#$# !
!! !
!! !
!!"# !
!#!" !
!#$#
!#!" !
&''%((%
&)'%((%
"!%!
&!
"!""%$*!
"!"())!
"!""""#!
"!*"%&!
%!+!
'!&'$+!
"!*('&!
%""!
"!$*(!
! "!!! !
1,0
0,8
q [kPa]
60
150
0,6
200
p [kPa]
0,4
0,2
0,0
10
10
10
10
10
Zyklenanzahl
Number
of Cycles N [-]
10
Dehnungsamplitude
Strain amplitude ampl [10-4]
Bleibende
Dehnung acc [%]
Strain
accumulation
1291
5
4
3
2
1
0
100
101
102
103
104
Zyklenanzahl
Number
of Cycles N [-]
105
Figure 7. Strain accumulation and strain amplitude curves measured in the drained cyclic triaxial test
1292
-!
".*)(!
".*))!
5. References
[1] Mardones M. & Vidal C, La zonificacin y evaluacin de los riesgos naturales de tipo geomorfolgico: un instrumento
para la planificacin urbana en la ciudad de Concepcin. EURE (Santiago) v.27 n.81 Santiago set.2001.
[2] Quezada J, Geologa urbana y ambiental de la ciudad de Concepcin. Memoria para optar al ttulo de Gelogo,
Universidad de Concepcin, 1996.
[3] Puga P, Estudio experimental de coeficientes de permeabilidad en arenas. Memoria para optar al ttulo de ingeniero
civil, Universidad Catlica de la Santsima Concepcin, 2012.
[4] Niemunis A, Wichtmann T, Triantafyllidis T. A high-cycle accumulation model for sand. Comput Geotech, 32(4): 24563, 2005.
[5] Kolymbas D, A rate-dependent constitutive equation for soils, Mechanics Research communications, 1(4):367-372,
1997.
[6] Wolffersdorff P. A hypoplastic relation for granular materials with a predefined limit state surface. Mechanics of
Cohesive Frictional Materials 1996, 1:251 -271.
[7] Niemunis A, Extended hipoplastic models for soils, Dissertation submitted for habilitation. Bochum, January 2003
[8] Niemunis A., Herle I. Hypoplastic model for cohesionless soils with elastic strain range. Mechanics of CohesiveFrictional Materials 2:1997, 279 - 299 pp.
[9] Herle I. & Gudehus G. Determination of parameters of a hypoplastic constitutive model from properties of grain
assemblies. Mech. Choes. Frict., 4,461-486
[10] Villalobos F.A. Mecnica de suelos. Editorial Universidad Catlica de la Santsima Concepcin, Chile. 2014.
[11] Wichtmann T, Niemunis A, Triantafyllidis T. Improved simplified calibration procedure for a high-cycle
accumulation model. Soil Dynamics and Earthquake Engineering 70 (2015):118-132.