TECHNICAL TRANSACTIONS CZASOPISMO TECHNICZNE

CIVIL ENGINEERING BUDOWNICTWO

5-B/2014

MAGDALENA MIKOŁAJCZYK*

DYNAMIC COMPACTION OF COHESIVE SOILS –

THEORETICAL ASPECTS AND MODELING

DYNAMICZNE ZAGĘSZCZANIE GRUNTÓW SPOISTYCH –

ASPEKTY TEORETYCZNE I MODELOWANIE

Abstract
The principles of dynamic compaction of soils are presented in the paper, along with a short
description of its usefulness for cohesive soils. The main topic of the article is the description of
ways of modeling the phenomenon related to dynamic compaction. This is based on experimental
data and on recently developed computer models using the Finite Element Method. Existing
simplified ‘perfectly flexible plastic’ model is presented and used for computer modeling.
The model does not capture the highly plastic and nonlinear behavior of cohesive soils under
dynamic compaction. Additionally, a modified Cam-Clay constitutive model will be briefly
described, which can address the above mentioned issues. Computer modeling method of the
phenomenon will be discussed, together with a short description of the dynamic characteristics
of the process.
Keywords: dynamic compaction, cohesive soils, Final Elements Method

Streszczenie
W artykule przedstawiono podstawy zagęszczania dynamicznego wraz z krótkim opisem jego
stosowalności dla gruntów spoistych. Podstawowym tematem artykułu jest jednakże opis
modelowania tego zjawiska. Jest on oparty na danych doświadczalnych oraz, ostatnio rozwi-
niętych, metodach opartych na Metodzie Elementów Skończonych. Przedstawiono podstawy
modelu doskonale plastycznego (stosowanego w modelowaniu komputerowym). Model ten
nie odzwierciedla jednak wysoce plastycznego i nieliniowego zachowania gruntów spoistych
pod wpływem dynamicznego zagęszczania. Zmieniony model konstytutywny Cam-Clay został
krótko opisany, może on do pewnego stopnia rozwiązać powyższe problemy. Zostaną przedsta-
wione sposoby praktycznego modelowania zjawiska, a także podstawy komputerowego mode-
lowania zjawiska, wraz z dyskusją dynamiki procesu.
Słowa kluczowe: zagęszczanie dynamiczne, grunty spoiste, metoda elementów skończonych

M.Sc. Eng. Magdalena Mikołajczyk, Ph.D. student, Institute of Geotechnics, Faculty of Environmental
*

Engineering, Cracow University of Technology.

178
Notation

c – cohesion [kPa]
Cc – wave velocity [m/s]
d – crater depth [m]
D – tamper diameter [m]
E – elastic modulus [kPa]
f – wave propagation frequency [Hz]
F – force [MN]
G – shear modulus [kPa]
H – tampering height [m]
M – constrained modulus [kPa]
t – time [s]
W – tamper weight [Mg]
H – tampering height [m]
λ – wave length [m]
ρ – density [Mg/m3]
υ – Poisson’s ratio [–]
φ – friction angle [deg]
ψ – dilation angle [deg]

1. Introduction

Dynamic compaction (DC) is an engineering process used to increase structural bearing

capacity of soils – for building purposes (in particular – foundation soils straightening).There
are a few distinct types of that process, but only DC using heavy tampering issues is shortly
discussed in this paper. It was invented a long time ago, but in modern practice it has been
developed by Louis Mennard and his commercial company MENARD Corp [1–3].
The process can be described as follows: the tamper (usually a steel cylinder of 2–3 meters
of diameter, several tons of weight) is raised up to a pre-calculated height using a special
crane to be then released. It gains speed quickly (and thus kinetic energy) and then it hits the
ground, compacting it. A crater forms, sometimes of a depth of 1m or more. The structural
parameters of soils are then greatly improved (albeit only in the vicinity of actual impact
site). The procedure is repeated for neighbouring places following a pre-planned regular
pattern. Sometimes more than one hit of the tamper is required. The procedure is referred to
as the “multiple passes” approach. The entire area is leveled afterwards, using bulldozers,
and sometimes other materials (gravels) are introduced [3, 4].
The following parameters are of main interest: the tamper weight W, the tampering height
H, the tamper diameter D of the bottom area, the crater depth d, the type of soil to be treated,
including its actual layered structure.
This technology is used for improvement of foundations for large-scale projects, where
no other known technology can be used in an economical way. It has been successfully
utilized for granular soils, where its functioning is quite well understood. It can be used for
fine soils as well, but the physics behind this is far more complicated [4].
 179
From an engineering point of view, there are three main differences between Dynamic
Compaction of granular and fine soils. Firstly, there are water-pore pressure issues. In
coarse, granular soils after tampering, water is filtered away through small pores between
the grains. The water pressure slowly decreases after that. This process can take a few
months to complete. In fine soils it is, however, hardly possible. Water will not filter
through soils that easily. When the pore water pressure is high enough, it creates cracks
in the body of soil around the point of tampering, so that water may escape. Sometimes
this flow is so intensive that shallow marshes or puddles are visible on the surface after
tampering [2, 4, 5].
Secondly, the situation with compressed gas bubbles is different for fine and coarse soils.
In fine soils a large part of air trapped in the soil pores does not escape during dynamic
compaction, and while this will not crack the soil body, it is being compressed, thus the
overall volume of pores can be decreased, resulting in some soil compaction [2].
Thirdly, there is a difference between fine and coarse soils regarding remoulding issues.
With great stresses caused by DC the internal structure of the soil may be destroyed, so at that
point the soil may behave like a liquid. This happens when the stresses overcome the forces
holding the grains together. This phenomenon is called a “soil liquefaction”. After some time,
however, the soil “remoulds” as it returns to the solid state, albeit with different parameters.
This may lead to a general improvement of the soil condition (as it gets compacted that
way), but if fine soils are surrounded by significant inlets of coarse material (that is fine soils
are “supported” by a coarse “skeleton”) liquefaction of the fine part will be of no use, as it
shall “remould” into the same shape as before the treatment. Because of that, DC must be
performed in such a way that liquefaction of coarse material happens before, or nearly at the
same time, as of the fine one [2].
Dynamic Compaction can thus be utilized for both granular, and fine soils, or a combination
of both, but there is a need for good modeling tools and guidelines. Louis Mennard has
proposed some rough estimations [2, 3], based on actual engineering practice, but a more
refined approach is needed, because not all soils can be effectively treated that way. The
other problem is that DC is an efficient, though costly technology. Excess tampering may be
unnecessary representing wasted time and money, or even counterproductive, since a poorly
designed DC process may even weaken the soil. A lot of energy is put into the ground in
a dynamic way. This may cause serious problems for neighbouring buildings which poses
a risk of cracking or even failing. An estimation of the allowable distance from the nearest
structure is therefore very important.

2. Common modeling approaches

The described phenomenon has been studied extensively. Some of the main findings and
theories used are summarized below. In most cases the calculations are done using the Finite
Element (FE) method.
As a first step for modeling of dynamic consolidation of soils, some models based on
the linear equivalent viscous-elastic theory are introduced and implemented. However,
they are not realistic for a medium like soils characterized by complicated behavior under
dynamically changing pressures, among other issues. Some parameters however, may be
extracted from that models for further use. Models based on the elastic principle are much
180
better as they can capture the nonlinear behavior under changing and variable stress paths.
There is still a serious problem with capturing and modeling of dynamic processes, including
Dynamic Compaction [6].
The linear viscoelastic model, yet only partially useful, has one big advantage: only one
stiffness matrix is needed for each time step. The equivalent for elasto-plastic models are
much more complicated. Moreover, their implementation is more difficult [6].
To be able to solve these problems simultaneously, a new approach was referred to as the
hypo-plasticity model. It was developed for stress-strain under simple loading. Then it was
extended to more complicated loadings.
The model is based on dividing an effective stress tensor so that the spherical and the
deviator stress which determine the behavior are seen. The theory was implemented into
the FE code, and some reasonable results were obtained [6], but only for saturated sand
foundations (although the authors of the method have stated that the model had also been
verified using triaxial tests on silty clays).
Another widely used model is a Cam-Clay plasticity model. It was developed only for
deformation and strain analysis of partially or fully saturated granular soils, using three-
phase continuum theory [7].
Firstly, the energy expressions for the constitutive models are given. They are then
related to effective stresses and deformations of the soil matrix, the pressures and the volume
changes. Also, perhaps most importantly, they are linked to the seepage forces and the
corresponding pressure gradients. The dissipation inequality is then calculated as well as
condition of convexity of the yield function [7]. Next, conditions describing deformation
bands for drained and undrained states are given. Finally, specific constitutive models for
partially saturated soils have been developed, taking into account the calculated plasticity
of the media. An interesting point is that the matrix suction is treated as another strain-like
variable. Again, the model can only be partially used for fine soils, or for a mixture of different
types of soils, under dynamic loads.

3. Force – load function

Another approach was proposed in [8]. Here, to overcome the general problem of
dynamic loads, a different modeling tool has been developed. A specially constructed force-
time function was superimposed on a fairly classic elastic model. This has led to a hybrid
model, which was then introduced into the widely used Finite Element Method (FE)
software. This approach is especially interesting for the author, although another software
is to be used, named Z_SOIL v13.09 2d. This software is widely used for various tasks
involving foundations analysis in different conditions. It has become an industrial standard
for FE analysis. Being able to model DC that way, shall greatly increase the usefulness of the
software referred to [8]. This is, however, a model for coarse soils.
The proposed function [8] consists of two parts: The left (ascending) part and the
right (descending) part. Both are chosen to provide the composite characteristics of the
phenomenon. The shapes of the two parts of the plot were based on both experimental and
analytical data.
 181

Fig. 1. Force-time function, along with the interpolation plot

To simplify the analysis an interpolation function is proposed (Fig. 1), using the following
polynom:

F (t ) = 10000000t 4 − 747786t 3 − 8715.6t 2 + 873.15t + 0.0042 (1)

where:
t – denotes the time [s].
The function was based on regeresion analysis of the available data.
The figure shows that main differences between the lines (original data and interpolation)
occurs in the very last part (about 0.045s). This can be easily rectified by using a more
complicated polynominal function, or another type of interpolation, but it seems to have no
big impact on the results.

4. FE model

To build a working model, boundary conditions must be specified. The Figure 2 shows
the original plot [8] together with the author’s implementation of this plot using Z_SOIL.
(Fig. 2). The model is axi-symmetric.

Fig. 2. Boundary conditions for modeling [8] (left), as used in

Z_SOIL software suite (right)
182
This is a provisional model, based on [8], for coarse soils. The soil parameters used are:
E = 5000 kPa, φ = 25 deg, ψ = 5 deg, c = 5 kPa, ν = 0,35, ρ = 1.8 mg/m3. Some other data
were calculated: M = 8025 kPa, G = 1852 kPa, Cc = 66.8 m/s, wave propagation frequency
for f = 10 Hz, λ = 6.7 m. They were used for determination of the size of FE mesh and mesh
density identification.

5. Conclusions

Dynamic compaction technology has been known for some time now. It has been
successfully used for different conditions, however an advanced tool for a correct design of
the process is needed, especially for fine soils.
A proposed polynominal force-time function in its current form can be a meaningful step
for formulating better modeling tools for dynamic loads assessment.
The results obtained by the Z_SOIL model differ from the expected deformations and
settlements. This may mean that the model will have to be modified taking into account the
highly nonlinear behavior of soils. The next step will be the improvement of the Z_SOIL
model and its implementation for cohesive soils. It is emphasized, however, that much more
work is required to attain a useful model for fine soils under dynamic loads, including the
multiple-passes case.

References

[1] Gaszyński J., Pabian Z., Modeling research of dynamic compaction of cohesive soils.
Badania modelowe udarowego zagęszczania gruntu spoistego, Czasopismo Techniczne
3-Ś/2011, Kraków 2011, 28-50.
[2] Menard L., Broise Y., Theoretical and practical aspects of dynamic compaction, 1975,
Geotechnique, 25, 3-18.
[3] Menard Soltraitementi Freyssinet Polska, Means of foundations enchencing. Metody
wzmacniania podłoża gruntowego, Materiały reklamowe, 2011.
[4] Mostafa K., Numerical modeling of dynamic compaction in cohesive soils, disstertation
for the phd title, University of Akron, Akron 2010.
[5] Pabian Z., Experimental research of cohesive soils deformation process under dynamic
loads. Proces deformacji gruntów spoistych wywołany obciążeniem udarowym w świetle
badań eksperymentalnych, praca doktorska, Politechnika Krakowska, 2005.
[6] `Zhou Ch., Shen Z.J., Yin J.H., Biot dynamic consolidation finite element analysis using
a hypo-plasticity model, 13th World Conference on Earthquake Engineering, Vancouver,
Canada, 2004, Paper No.351.
[7] Borja R.I., Cam-Clay plasticity. Part V: a mathematical framework for three-phase de-
formation and strain localization analyses of partially saturated porous media, Comput.
Methods. Appl. Mech. Engineering 193, 2004, 5301-5338.
[8] Pan J.L., Selby A.R., Simulation of dynamic compaction of loose granular soils, Advanc-
es in Engineering Software, 33, 2002, 631-640.