Professional Documents
Culture Documents
Applicability of Burger Model in Predicting The Response of Viscoelastic Soil Beds PDF
Applicability of Burger Model in Predicting The Response of Viscoelastic Soil Beds PDF
ABSTRACT
This paper presents a critical review of various viscoelastic lumped parameter
models (Maxwell model, Kelvin-Voigt model, Poynting-Thompson model and
Burger model) commonly used to represent the stress-strain-time behavior of a
compressible viscoelastic medium. Efficacy of such models has been checked to
suggest a rational model so as to represent the time-dependent behavior of such
media. Though Burgers four element model can incorporate all the phenomena of
viscoelastic behavior of materials and all other models can be degenerated to other
lower order models, it has not yet found general acceptability amongst the soil
engineers to model such behavior of soils. Through this study, an effort has been
made to demonstrate that it is probably the most effective model to predict the
behavior of structures resting on such soils. Therefore, the efficacy of the model so
chosen has been demonstrated through a case-study available in literature. Based on
the studies, it has been inferred that the Burger model possesses an excellent
potential for proper representation of the time-dependent behavior of a saturated
viscoelastic medium when subjected to loading an unloading phenomena.
INTRODUCTION
Since the proposition of Winklers model to represent the behavior of soils by
a series of discrete springs, several such models have been proposed to simulate the
response of elastic foundations to the externally applied loads. However, such
models, consisting of either linear or nonlinear springs, are unable to simulate the
viscoelastic and time-dependent behavior exhibited by saturated clayey soils.
When subjected to any external loading, the elastic foundation beds primarily
undergo instantaneous settlement, i.e. the majority of the total settlement is recorded
at the instant of the application of load. Upon unloading, such beds exhibit an elastic
recovery which does not change with time. The force-deformation response of such
foundations can be modeled with the aid of Winkler springs and its variations.
Contrary to the former type of foundations, viscoelastic foundation beds exhibit
time-dependent response when subjected to the action of external loading, i.e., such
2611
foundation beds record an instantaneous settlement directly after the loading and an
additional settlement increasing linearly or non-linearly with time due to gradual
outflow of pore water. Upon unloading, an instantaneous elastic recovery is
noticeable, followed by a partial recovery that apparently attains a constant value
with time [Tan (1959)]. Modeling of the time-dependent response of the viscoelastic
beds is achieved by the use of mechanical elements such as Winkler springs (linear
or nonlinear), Viscous dashpots (Newtonian or non-Newtonian) and their various
combinations. Such modeling is a very interesting topic of research in the field of
geotechnical engineering since a few decades. Several researchers have used various
types of viscoelastic lumped parameter models to represent the behavior of a
viscoelastic medium [Kerr (1961), Schiffman et al. (1964)]. Other viscoelastic
models such as the Zener model, Bingham Elastosupple model, Sobotkas
Elastosupple model etc. are used at specific situations in order to represent certain
specific stress-strain-time behavior of the subgrade subjected to special loading
conditions. Complex models such as chains of Maxwell and Kelvin-Voigt are also in
use. The constitutive relationships for such models can be found in standard
reference books [Flugge (1967), Sobotka (1985)]. Murayama and Shibata (1961),
and Schiffman et al. (1964) have proposed higher order complex viscoelastic models
of five- and seven-parameters to represent the soil behavior. Due to the incorporation
of more-and-more parameters affecting the force-deformation response of a
viscoelastic medium, with the passage of time, these models turned out to be more
and more complex in their physical and mathematical representation. Such
progresses led to the improvement of the models so that the viscoelastic behavior of
soil under various conditions could be simulated properly.
The use of these various models has two prime aspects.
Suitability of a model to correctly predict the viscoelastic soil behavior.
Model parameter estimation.
This paper primarily focuses on presenting a critical discussion of the former
aspect in the following sections.
RATIONAL CHOICE OF A VISCOELASTIC LUMPED PARAMETER
MODEL
In general, for foundations resting on viscoelastic subgrade, the conventional
empirical approach by Buisman (1936) is followed to determine the primary and
secondary consolidation of soil separately. However, rational theories of secondary
consolidation treat both the consolidation processes to be occurring in unison. Biot
(1956) provided the three-dimensional theories of secondary consolidation under
effective viscoelastic stress-strain relationships. Tan (1957) used the two-parameter
Maxwell model to analyze the settlement characteristics of foundations on clayey
beds, which was supported by enough experimental evidences [Geuze and Tan
(1954)]. However, the experiments exhibited a considerable deviation from the
theoretical predictions [Schiffman et al. (1964)]. Taylor (1942) proposed the use of
Kelvin-Voigt (KV) model to determine the coupled elastic-viscoelastic behavior of
soft soil, which was improved further by adding an elastic element in series to the
Kelvin unit [Tan (1957)]. It is observed that the KV model cannot take into account
the immediate elastic settlement of a viscoelastic medium upon the application of
2612
2613
2614
viscous sub-elements. If the stress = 0 does not vary with time i.e. the element is
subjected to a constant stress, the element undergoes creep. The total strain at any
instant of time ME ( t ) is expressed as a summation of instantaneous elastic
( )
e
c
strain ME
and the creep strain ME
( t ) defined by the following expression:
0 0t
+
e
c
+ ME
ME ( t ) = ME
(t ) =
t ta
(1)
where, ta is the time period for which the constant normal stress ( 0 ) is applied, t is
the elapsed time, E is the modulus of elasticity of the Hookean elastic element, and
is the coefficient of normal viscosity of the Newtonian viscous element.
(a)
(b)
Figure 1. The rheological model of a Maxwell viscoelastic liquid element and its
stress-strain-time behavior
Figure 1(b) depicts the stress-time and the corresponding strain-time behavior
of the Maxwell viscoelastic liquid element. At the instant of unloading, elastic
recovery due to the retreat of the Hookean elastic element is observed and the initial
instantaneous strain is entirely recovered. The viscous strain at the instant of
unloading is maintained constant after unloading, which can be mathematically
expressed as follows [Equation (2)]:
ME =
0t a
t > ta
(2)
2615
(a)
(b)
Figure 2. The rheological model of a Kelvin-Voigt viscoelastic solid element and
its stress-strain-time behavior
For stresses remaining constant with time and in the absence of any initial
strain at time t = t0 , the total strain in the KV model at any particular instant of
k (t ) =
Et
1 exp = 0
E
E
t
1 exp
tr
t ta
(3)
2616
the KV element is loaded for an indefinite time-period, the strain-time curve tends
asymptotically to the elastic value ( = 0 E ) .
If the stress acting on the KV element is suddenly removed from the strained
body at time t = ta , the body will revert to the unstrained state following the equation
as follows [Equation (4)]:
E ( t ta )
Et
k ( t ) = 0 1 exp a exp
t ta
(4)
the relation for the total strain at any instant of time PT ( t ) for sustained loading
which is expressed as a summation of the instantaneous elastic strain and delayed
elastic strain as follows:
PT ( t ) =
E t
+ 0 1 exp 2 t ta
E1 E2
(5)
(a)
(b)
Figure 3. The rheological model of a Poynting-Thompson element and its stressstrain-time behavior
2617
Figure 3(b) depicts the stress-time and the corresponding strain-time behavior
of the Poynting-Thompson viscoelastic element. After removal of the stress, if the
Poynting-Thompson element is kept unloaded over an indefinite long period of time,
the strain-time curve follows an exponential decay i.e. the total recovery of the strain
occurs following an exponential curve, which is expressed as:
E2 ( t ta )
0
E2ta
1 + exp
exp
t ta
E2
PT ( k ) =
(6)
with a Kelvin-Voigt ( KV ) element. For stresses remaining constant with time, and
in the absence of any initial strain, the strain in the Burger element at any instant of
time is expressed as follows:
1
E2t
t
1
+ +
1 exp
t ta
2
E1 1 E2
BU ( t ) = 0
(7)
(a)
(b)
Figure 4. The rheological model of a four-parameter Burger element and its
stress-strain-time behavior
2618
model. If the Burger element is kept unloaded for a long period of time, creep
recovery of the element occur which is governed by the following equation
[Equation (8)]:
0
E2ta
1 exp
E2
2
BU ( t ) =
E2 ( t ta )
t t a + t
exp
2
(8)
the KV model is suitable for the representation of the long-term behavior of the soil.
However, as observed from Figure 2(b), when kept unloaded for an indefinitely long
period of time, the Kelvin-Voigt model undergoes total recovery of the strain which
is not a usual phenomenon. Thus, this model also fails in representing properly the
stress-strain-time behavior of viscoelastic medium.
It is observed from Figure 3(b) that the Poynting-Thompson element is able
to represent the instantaneous elastic deformation upon sudden loading, delayed
elastic deformation or creep upon maintaining the sustained loading, instantaneous
recovery upon unloading, and exponential strain recovery due to sustained unloading
of the element. However, this model also lacks the ability to capture the permanent
deformation obtained as a result of loading of a viscoelastic subgrade.
Thus, amongst all the models discussed, Burger model can be stated to be the
most generalized model which can be degenerated to all other lower models with
suitable omission of the Hookean springs and/or Newtonian dashpots. It is observed
from Figure 4(b) that the Burger model is capable of expressing correctly the
complex phenomena associated with the long-term reversible and irreversible
deformation, along with the modeling of instantaneous strain and recovery due to
sudden loading and unloading of the element.
(a)
(b)
Figure 5: Loading and unloading sequence; and, comparative study of the
predicted and observed Settlement of pile-top and pile-bottom with elapsed time
[Data from static load test on a pile at M.I.T Centre for Space Research (Lambe
and Whitman, 1979)]
2619
model can suitably model the pile-soil settlement behavior with time. The model has
been further checked through several case studies where the tests were conducted for
short and long durations. Comparison of the data presented above with all the four
models is a very interesting idea; however, for the sake of space and brevity these are
not presented here, and are to be included in an upcoming full paper.
CONCLUSIONS
In the light of the above discussions, it can be concluded that the Burger
model possesses an excellent potential for proper representation of the timedependent behavior of a viscoelastic medium (saturated clays/ clayey soils) subjected
to loading and unloading. The correctness of the prediction of time-settlement
behavior of pile tip using this model has been established and explored through an
in-situ pile load test reported in literature. The predictions made are found to be in
very close agreement with that of field measurement. Thus, it is established that the
Burger model can be used with confidence in determining the time-dependent
behavior of a viscoelastic medium.
REFERENCES
Biot, M. A. (1956) Theory and deformation of a porous viscoelastic
anisotropic solid Journal of Applied Physics, Vol. 27, No. 5, pp. 459 467.
Buisman, A. S. K. (1936) Results of long duration settlement tests
Proceedings of the 1st International Conference of Soil Mechanics and Foundation
Engineering, Harvard, USA, pp. 103 106.
Flugge, W. (1967) Viscoelasticity, Blaisdell Publishers, London.
Geuze, E. C. W. A. and Tan, T. K. (1954) The mechanical behavior of
clays Proceedings of the 2nd International Conference on Rheology, New York, pp.
247 259.
Kerr, A. D. (1961) Viscoelastic Winkler foundation with shear interactions
Journal of the Engineering Mechanics division: Proceedings of the ASCE, Vol. 87,
No. EM3, pp. 13 30.
Murayama, S. and Shibata, T. (1961) Rheological properties of clays 5th
International Conference of Soil Mechanics and Foundation Engineering, Paris,
France, Vol. 1, pp. 269 273.
Schiffman, R. L., Ladd, C. C. and Chen, A. T. F. (1964) The secondary
consolidation of clay Rheology and Soil Mechanics: Proceedings of the
International Union of Theoretical and Applied Mechanics Symposium, Grenoble,
Berlin, pp. 273 303.
Sobotka, Z. (1985) Rheology of Materials and engineering structures:
Rheology Series 2, Wiley Interscience Publication, Netherlands.
Tan, T. K. (1957) Three-dimensional theory on the consolidation and flow
of the clay-layers Scientica Sinica, Vol. 6, No. 1, pp. 203 215.
Tan, T. K. (1959) Structure mechanics of clays Scientica Sinica Col. 8,
No. 1, pp. 83 97.
Taylor, D. W. (1942) Research report on consolidation of clays
Massachusetts Institute of Technology, Department of Civil and Sanitary
Engineering, Publication No. 82.
2620